Press Releases on Tachyum

About Tachyum

Tachyum is transforming the economics of AI, HPC, public and private cloud with the world’s first Universal Processor Prodigy unifying the functionality of a CPU, an HPC GPGPU, and AI accelerators to deliver industry-leading performance, cost and power efficiency. Tachyum has offices in the United States, Slovakia, Taiwan and the Czech Republic. For more information, visit https://www.tachyum.com/.

Tachyum Open Sources 281GB/s TDIMM™ for the Future of AI and Computing

Tue, 25 Nov 2025 00:00:00 +0000

LAS VEGAS, November 25, 2025 –Tachyum® today announced details about how its TDIMM™ is bringing the future of AI and computing, enabling AI models with parameters many orders of magnitude greater than those of any existing solution at a fraction of the cost. TDIMM is key to reducing the estimated cost of OpenAI data center $3 trillion and 250,000 megawatts of power to $27 billion and 540 megawatts.

Prodigy Ultimate provides up to 21x higher AI rack performance than NVIDIA Rubin Ultra NVL576. Tachyum DDR5 DIMM (TDIMM) bandwidth increase is 5.5 times, from 51 GB/s to 281 GB/s. Prodigy Ultimate’s 24 channels provide 6.7 TB/s, 11x increase over 12-channel CPUs. The TDIMM supports 256 GB in standard, 512 GB in tall, and 1 TB in extra tall height, and TSV increases capacity up to 8 times to 3 PB per Prodigy node.

Existing DDR5 RDIMM provides 64-bit data, 16-bit ECC, 40 DQS, 14 C/A, and 16 control signals for a total of 146 signals in 288-pin connector. Tachyum TDIMM provides 128-bit data, 16-bit ECC, 36 DQS, 14 C/A, and 16 control signals for a total of 206 signals in 484-pin connector, only 38% signals increase for doubling the memory bandwidth. From a cost perspective, TDIMM needs 10% fewer DRAMs and is expected to be 10% lower cost.

The physical dimensions of connectors and DIMMs are the same as for DDR5 RDIMM, and for tall DDR5 MRDIMM. The extra tall TDIMM further doubles capacity when the server enclosure is 3U instead of 2U, which is the typical server height today. The contact pins have 0.5 mm pitch, the same as DDR5 SODIMM, so no new contacts need to be developed. Only the plastic mold needs to be modified, which is a low-intensity engineering task.

The DDR5 x4 RDIMM activates 20 DRAM chips and TDIMM only 18 DRAM chips per access, saving power. The DDR5 RDIMM switches 124 pins (64+16+40+2+2) at a high rate and TDIMM switches 184 pins (128+16+36+2+2). The TDIMM power consumption is expected to be 30% higher for 2x bandwidth. Using newer DRAM chips will put TDIMM power consumption at about the same level as older DDR5 RDIMM.

The DDR5 TDIMM doubles DDR5 RDIMM or MRDIMM bandwidth and capacity without waiting years for expensive DDR6 memory. The TDIMM quadruples bandwidth in comparison to DDR5 RDIMM. The existing DDR5 controller and PHY IP changes are simple by increasing the data path from 80 bits to 144 bits. The command/address and data buffers chips do not need to change, making adoption less than a year from now.

Minor changes to DDR6 controller, PHY, and MRDIMM chips will double bandwidth from 6.7 TB/s to 13.5 TB/s in 2027, exceeding Nvidia Rubin’s 13 TB/s. The TAI reduces bandwidth up to 4x, making TAI inference like with 54 TB/s of bandwidth. Evolutionary changes in 2028 would double TDIMM based AI chips bandwidth to 27 TB/s.

Making TDIMM open source and royalty-free to companies and the whole world, including JEDEC, will drive mainstream volume adoption and low cost. China was late with DDR5, but they can leapfrog to performance and density as DDR6 next year, making low-cost volume manufactured TDIMM available before Prodigy launch. Interested parties should contact Tachyum now to get access to TDIMM technology and agreements.

“The TDIMM is key in reducing the cost of AI systems trained on all the knowledge from $8 trillion and 276 gigawatts to $78 billion and 1 gigawatt in 2028,” said Dr. Radoslav Danilak, founder and CEO of Tachyum. “The TDIMM ushers in the era of affordable AI trained on all written knowledge produced by humanity, accessible to many companies and nations.”

Tachyum believes that its technology should be democratized and accessible to serve all mankind. Opening of TDIMM technology will be followed this year by making the Instruction Set Architecture (ISA) open and releasing software to open source. In 2023, Tachyum announced that TAI TPU™ technology is available for licensing to ensure wide proliferation of Tachyum technology outside of data centers to edge and IoT devices.

The Prodigy Universal Processor delivers orders of magnitude higher AI performance, 3x the performance of the best x86 processors, and 6x HPC performance of the fastest GPGPU. Eliminating the need for expensive dedicated AI hardware and dramatically increasing server utilization, Prodigy reduces data center CAPEX and OPEX significantly while delivering unprecedented performance, power and economics.

Those interested in reading the technical details for TDIMM architecture can download the Tachyum TDIMM Module Product Brief.

Follow Tachyum

About Tachyum

Tachyum Prodigy Can Reduce OpenAI $3T 250 GW to $27B 540 MW

Mon, 17 Nov 2025 00:00:00 +0000

LAS VEGAS, November 17, 2025 – Tachyum today announced details regarding how performance improvements are achieved for its 2nm Prodigy® Universal Processor. It will enable AI models with parameters many orders of magnitude more than those of any existing solution at a fraction of the cost.

Prodigy Ultimate provides up to 21.3x higher AI rack performance than NVIDIA Rubin Ultra NVL576. Tachyum DDR5 DIMM (TDIMM) increases bandwidth 5.5 times from 51 GB/s to 281 GB/s. Prodigy Ultimate’s 24 channels provide 8 GB/s, an 8.2x increase over the previous design, and 11x increase over 12-channel CPUs. The TDIMM supports 256 GB in standard, 512 GB in tall, and 1 TB in extra tall height. TSV increases capacity up to 8 times to 2 x 96 TB per Prodigy Ultimate, or up to 3 PB per node.

The TAI data types reduce bandwidth up to 4x, making TAI inference performance like with 27 TB/s of bandwidth. Tachyum 16-socket node provides 16 GB cache to reduce memory bandwidth, which is 130x more than 126MB of NVIDIA B300, and older NVIDIA V100 has HBM memory size comparable to Prodigy’s node caches.

U.S. AI companies have spent billions on hardware. China focuses on efficiency. The Chinese DeepSeek inference is activating 20% parameters, enabling 5x larger models for the same bandwidth and compute. The Chinese Moonshot AI Kimi K2 activates 32 billion parameters out of 1 trillion, enabling 32x larger models for the same processing.

Human brains contain 150 trillion synapses. AI ChatGPT 4 has 1.8 trillion parameters, and BaGuaLu has 174 trillion parameters. The ultimate AI, trained on all the knowledge of humanity, is exceeding 100,000,000 trillion parameters. The Nvidia solution would cost $8 trillion and require 276 gigawatts of power. Tachyum’s solution will cost $78 billion and require 1 gigawatt of power, making it accessible to multiple companies and nations.

OpenAI CEO Sam Altman in September stated plans to build 250 gigawatts of AI data centers by 2033. India consumes 250 GW of electricity emitting twice as much carbon dioxide as ExxonMobil. 60 million GB300 with 288 GB memory for FP4 weights will need 17.3 exabytes of memory to support 34,560,000 trillion parameters. With $50,000 per B300 on the solution level, it would cost $3 trillion.

Using $200 per 48GB DDR5, the 17.3 exabytes (EB) would cost $72 billion. The Tachyum TAI data type reduces that to 8.6 EB for $36 billion and retraining to TAI reduces that to 4.3 EB for $18 billion. The $50,000 24-channel Prodigy Ultimate attaches 6TB of low-cost 256 GB DIMMs. The 4.3 EB of memory requires 720,000 Prodigy Ultimate costing $36 billion, or 360,000 with tall TDIMMs costing $18 billion, or 180,000 with extra tall TDIMMs costing $9 billion. That is 100x less than $3 trillion needed for NVIDIA B300.

At 3 kW per Prodigy Ultimate with DRAM, a 180,000 Prodigy Ultimate would consume 540 MW, or 460x less than the NVIDIA B300 solution. Inference performance of 60,000,000 B300 with 18 petaflops is 1,080 zettaflops. The 180,000 Prodigy Ultimate, each at 400 FP4 petaflops delivers 72 zettaflops, and models retrained for TAI inference deliver 216 zettaflops, which appears to be optimal performance for inference. If needed, the 720,000 Prodigy Ultimates deliver 864 TAI zettaflops for $54 billion and 2.2 GW of energy.

“With Prodigy Ultimate, the unsustainable $3 trillion datacenter consuming 250 gigawatts of power can be shrunk to a $27 billion consuming 540 megawatts in 2028 instead of 2033,” said Dr. Radoslav Danilak, founder and CEO of Tachyum. “With Prodigy Ultimate, humanity can move to an era of AI trained on all written knowledge produced by mankind.”

Those interested in reading the technical details for Tachyum’s latest Prodigy Universal Processor architecture can download the Tachyum Prodigy Feature Upgrade document.

Follow Tachyum

About Tachyum

Tachyum Unveils 2nm Prodigy with 21x Higher AI Rack Performance than the Nvidia Rubin Ultra

Wed, 12 Nov 2025 00:00:00 +0000

LAS VEGAS, November 12, 2025 –Tachyum® today announced details and specifications for its 2nm Prodigy® Universal Processor, which will enable AI models with parameters many orders of magnitude larger than those of any existing solution at a fraction of the cost.

Prodigy Ultimate provides up to 21.3x higher AI rack performance than Nvidia Rubin Ultra NVL576. Prodigy Premium provides up to 25.8x higher AI rack performance than Vera Rubin 144. Technical details of the 2nm Prodigy, the first ever chip to exceed 1,000 PFLOPs on inference, will be published within a week. Nvidia Rubin delivers 50 PFLOPs.

The global competition in AI continues to accelerate, with China and the United States leading the race. Current AI models demonstrate massive computational scales — for instance, ChatGPT 4 features approximately 1.8 trillion parameters, while human brains contain an estimated 150 trillion synapses. Emerging systems such as BaGauLu reach 174 trillion parameters, but the ultimate breakthrough is expected to come from models trained on the collective knowledge of humanity, exceeding 100 000 000 trillion (10²⁰) parameters. Traditional large-scale AI solutions could cost over $8 trillion and require more than 276 gigawatts of power. In contrast, the Tachyum solution is projected to achieve comparable capabilities at an estimated cost of $78 billion and a power requirement of just 1 gigawatt — making it accessible to multiple companies and nations.

In addition to open-sourcing all software, Tachyum is making its memory technology available, using standard components, allowing 10x increase of DIMM-based memory bandwidth available for licensing by memory or processor companies, including JEDEC adoption to achieve high adoption and low cost. In 2023, Tachyum announced licensable Tachyum AI (TAI) data types, and its Tachyum Processing Unit (TPU) core is available for licensing. Tachyum is in the process of making the Instruction Set Architecture (ISA) open.

Tachyum has continually upgraded its Prodigy design to address ever-changing requirements in server, AI and HPC markets with up to 5x integer performance, up to 16x higher AI performance, 8x DRAM bandwidth, 4x chip-to-chip and I/O bandwidth, 4x scalability by supporting 16 sockets, and 2x power efficiency, with lower cost per core.

The Prodigy chip was upgraded to 2nm to significantly reduce power consumption. Reducing chiplet die size improves cost despite expensive 2nm wafers. Each chiplet in the Prodigy package integrates 256 high-performance custom 64-bit cores. The power consumption reduction is critical, as multiple chiplets occupy a single package. Backed by a recent $220 million investment, the 2nm Prodigy is being readied for tape-out.

Multiple Prodigy SKUs cover a wide range of performance and applications, including big AI, exascale supercomputing, HPC, digital currency, cloud/hyperscale, big data analytics, and databases. Prodigy Ultimate integrates 1,024 high-performance cores, 24 DDR5 17.6GT/s memory controllers and 128 PCIE 7.0 lanes. The Prodigy Premium comes with 16 DRAM channels, and 512 to 128 cores scalable to 16 socket systems. Entry-level Prodigy comes with 8 or 4 DRAM controllers and 128 to 32 cores.

Prodigy features, scalability, and price segmentation ensure rapid market penetration. Tachyum provides out-of-the-box native system software, operating systems, compilers, libraries, many applications, and AI infrastructure frameworks. It also allows running unmodified Intel/AMD x86 binaries and mixing them with native applications. This ensures that Tachyum systems can be operational by customers from day one.

“With tape-out funding now secured after a long wait, the world’s first Universal Processor can proceed to production, designed to overcome the inherent limitations of today’s data centers,” said Dr. Radoslav Danilak, founder and CEO of Tachyum. “The distinct markets addressed by Prodigy are the AI, server, and HPC markets, requiring fast and efficient chips. Tachyum’s Prodigy Premium and Ultimate will supercharge workloads with superior performance at a lower cost than any other solution on the market.”

Those interested in reading the full specifications for Tachyum’s latest Prodigy Universal Processor architecture can download the solutions brief.

Follow Tachyum

About Tachyum

Tachyum Opens Offices in Taiwan to Expand Reach of Prodigy Universal Processor

Tue, 21 Oct 2025 00:00:00 +0000

LAS VEGAS, October 21, 2025 – Tachyum® today announced the opening of new offices in Taiwan, the world’s most advanced technology manufacturing and technological hub, in advance of production of the Prodigy® Universal Processor.

As Tachyum continues towards tape-out for Prodigy, the company has chosen to establish manufacturing, testing, assembly, ODM, logistics and support facilities in Taiwan to ensure worldwide shipments of the completed product without the need for intermediaries. This decision was based on Tachyum’s desire to simplify and speed up the supply chain while reducing costs and environmental impact.

The opening of its new facilities was made possible by efforts within the Taiwan government to strengthen economic ties with enterprises in the European Union. This mutually beneficial relationship between Taiwan and Tachyum Slovakia and Tachyum Czech Republic accelerates business and economic ties between the two regions.

“As we prepare for the final stage in Prodigy’s development, it is important to us to have facilities around the world to ensure the smooth production and distribution of the world’s first Universal Processor,” said Dr. Radoslav Danilak, founder and CEO of Tachyum. “With an established presence in the United States, Slovakia and the Czech Republic, turning to a technology hub like Taiwan provides tremendous benefits in terms of logistics by creating a strong foothold in Asia. With this latest expansion, we are well-positioned to service customers looking to leverage the benefits of the smallest, fastest and greenest general-purpose chip regardless of their location.”

The Prodigy Universal Processor offers industry-leading performance for AI/ML, HPC, and cloud workloads, with a single homogeneous architecture. Each chiplet in the Prodigy package integrates 256 high-performance custom 64-bit cores to deliver orders of magnitude higher performance for AI, and 3x the performance of the highest-performing x86 processors, and 6x the performance of the highest-performing GPGPU for HPC. By eliminating the need for expensive dedicated AI hardware and dramatically increasing server utilization, Prodigy reduces data center CAPEX and OPEX significantly while delivering unprecedented performance, power, and economics.

Follow Tachyum

About Tachyum

Tachyum Signs $220 Million Funding and $500 Million Purchase Order

Tue, 14 Oct 2025 00:00:00 +0000

LAS VEGAS, October 14, 2025 – Tachyum® today announced that it has executed a binding agreement for its Series C funding with a $220 million investment to be wired into Tachyum’s account within a month, by a European investor that positions the company to become one of the top enablers within the AI data center market. The Series C investor has also signed a $500 million purchase order for Prodigy® chips, which will increase AI performance by an order of magnitude and reduce the cost of ultra large LLM/AI models by approximately two orders of magnitude.

The lead investor from the US took a semiconductor company public last year in the US with a valuation of nearly a billion USD, and the same investor took another US company public through a SPAC for several billion USD.

With a total of more than $300 million invested over three funding rounds, Tachyum is poised to revolutionize AI data centers by solving the most challenging barriers to AI and expects that Series C will propel the company to an IPO, potentially as early as 2027.

There is tremendous growth that is attributed to increased demand for AI and cloud workloads, requiring hyperscale AI data centers to ensure applications and workloads can be achieved. The Stargate Project is investing $500 billion in US AI infrastructure to win the AI race. The Stargate UAE, brokered by the US, is building the world’s largest gigawatt-scale AI data centers outside the US, located in a 5-gigawatt campus in Abu Dhabi. Humain, Saudi Arabia’s AI company, is constructing 100 MW data centers in Riyadh and Dammam. In addition to companies looking to leverage the benefits of AI, there has been a marked interest from governments looking to win AI supremacy.

With the explosive costs of AI, there is a strong market pull for Tachyum’s Prodigy to minimize costs and democratize AI for organizations across the globe, which will enable them to benefit from the AI revolution. Countries in Asia are building semiconductor fabs while Europe is approving €30 billion on AI data centers for the next 2 years. Tachyum was selected by the EU for the Important Project of Common European Interests (IPCEI), which enables the EU’s AI to match AI in the US at a fraction of cost.

The Kimi K2 AI model has approximately 1 trillion parameters, ChatGPT 4.5 has an estimated 10 trillion parameters, and BaGauLu has 174 trillion parameters, which exceeds the 100-150 trillion synapses estimated in Human Brain. To accommodate the needs of next generation AI, Tachyum is using its latest investment funds to produce its Prodigy chips to further reduce cost and energy for inference and training ultra large AI models without sacrificing accuracy or task performance. These advancements are accelerating innovation by reducing capital and operational costs associated with inference and training ultra large state-of-the-art AI models.

Tachyum has continuously upgraded its Prodigy design while waiting for the Series C investment necessary for tape-out. With Series C closing as outlined above, Tachyum will finalize the tape-out process and schedule. Tachyum will disclose Prodigy’s upgraded specifications and performance in the near future.

“With the capital raised from this latest funding round, we are able to get complete Prodigy tape-out with the latest innovations and designs to meet ever-changing market demands,” said Dr. Radoslav Danilak, founder and CEO of Tachyum. “We are seeing how the battle for AI supremacy is currently being waged and we’re excited to bring to market a disruptive chip that will enable AI models with parameters many orders of magnitude larger than the synapses of the human brain at an affordable price at a fraction of cost of existing solutions.”

The Prodigy® Universal Processor offers industry-leading performance for AI/ML, HPC, and cloud workloads, with a single homogeneous architecture. Each chiplet in the Prodigy package integrates 256 high-performance custom 64-bit cores to deliver orders of magnitude higher performance for AI, and 3x the performance of the highest-performing x86 processors, and 6x the performance of the highest-performing GPGPU for HPC. By eliminating the need for expensive dedicated AI hardware and dramatically increasing server utilization, Prodigy reduces data center CAPEX and OPEX significantly while delivering unprecedented performance, power, and economics.

Follow Tachyum

About Tachyum

Tachyum Supports Next Stage of AI Revolution Behind FP4 Data Type

Wed, 08 Oct 2025 00:00:00 +0000

LAS VEGAS, October 8, 2025 –Tachyum® today announced that its AI team has successfully demonstrated an algorithm to perform LLM training in the quantized 4-bit FP4 format, dramatically reducing memory and compute requirements while delivering faster, more cost-effective and energy-efficient training without sacrificing model accuracy or downstream task performance. This advancement, detailed in the company’s latest white paper, “Tachyum demonstrates supercharged LLM training in only 4 bits”, offers transformative potential for LLM development, accelerating innovation by reducing capital and operational costs associated with training state-of-the-art AI.

Tachyum’s AI team has demonstrated that a foundation model fine-tuned on a task-specific dataset in FP4 data type, which represents value using 4-bit floating-point format rather than standard FP32 or BF16, achieves parity with traditional FP32 training baselines.

Quantization reduces numerical precision of values by reducing the number of bits in each numerical representation. Quantized models thereby achieve greater compression, requiring fewer compute, storage and memory demands. FP4 promises up to 4x better memory efficiency than 16-bit formats and up to 8x better efficiency than 32-bit.

LLMs increase in size by as much as 10x generation to generation, creating excessive training times that introduce significant delays and impact overall costs. The ability to train in FP4 addresses these pain points by improving performance, shortening training times and further reducing TCO.

“With AI model sizes doubling every 3 to 6 months, there is an ever-growing need for greater efficiency to narrow the ratio between model sizes and processing times,” said Dr. Radoslav Danilak, founder and CEO of Tachyum. “With models moving from 1T to 10T parameters – with the goal to achieve human brain processing of about 100T synapses – traditional FP32 or BF16 formats simply are a barrier to the AI revolution. By quantizing to the 4-bit FP4 format, we continue to transform the economics of AI by delivering industry-leading performance, cost and power efficiency required to properly train LLMs.”

As a Universal Processor offering industry-leading performance for all workloads, Tachyum Prodigy®-powered data center servers can seamlessly and dynamically switch between computational domains (such as AI/ML, HPC, and cloud) with a single homogeneous architecture. By eliminating the need for expensive dedicated AI hardware and dramatically increasing server utilization, Prodigy reduces CAPEX and OPEX significantly while delivering unprecedented data center performance, power, and economics. Prodigy integrates 256 high-performance custom-designed 64-bit compute cores to deliver up to 18x the highest performing GPU for AI applications, 3x the performance of the highest-performing x86 processors for cloud workloads, and up to 8x that of the highest performing GPU for HPC.

Those interested in learning more about Tachyum’s unique approach in solving the most challenging barriers to AI can download the “Tachyum demonstrates supercharged LLM training in only 4 bits” white paper.

Follow Tachyum

About Tachyum

Tachyum is transforming the economics of AI, HPC, public and private cloud workloads with Prodigy, the world’s first Universal Processor. Prodigy unifies the functionality of a CPU, a GPU, and a TPU in a single processor to deliver industry-leading performance, cost and power efficiency for both specialty and general-purpose computing. As global data center emissions continue to contribute to a changing climate, with projections of their consuming 10 percent of the world’s electricity by 2030, the ultra-low power Prodigy is positioned to help balance the world’s appetite for computing at a lower environmental cost. Tachyum received a major purchase order from a US company to build a large-scale system that can deliver more than 50 exaflops performance, which will exponentially exceed the computational capabilities of the fastest inference or generative AI supercomputers available anywhere in the world today. Tachyum has offices in the United States, Slovakia and the Czech Republic. For more information, visit https://www.tachyum.com/.

Tachyum Promotes its High-Performance Flash Storage Capabilities at 25th China International Fair for Investment & Trade

Wed, 03 Sep 2025 00:00:00 +0000

LAS VEGAS, Sept. 3, 2025 –Tachyum® today announced its participation in the 25th China International Fair for Investment & Trade (CIFIT), one of the world’s leading platforms for investment promotion by the Ministry of Commerce of the People’s Republic of China, Sept. 8-11 at the Xiamen International Conference & Exhibition Center.

Tachyum will be presenting progress of its Prodigy® Universal Processor development as part of the Slovak national group, organized in cooperation with the Slovak Investment and Trade Development Agency and the Consulate General of the Slovak Republic in Shanghai. Tachyum’s engagement with Chinese partners will center on enabling flash memory solutions, which have become increasingly critical for AI scaling and accessibility ventures.

With the trend of moving from hard disk drives (HDDs) to solid state drives (SSDs) increasingly globally, Tachyum will showcase at CIFIT its latest software distribution package and pre-configured applications, which have been optimized for 64KB pages for Linux and 16KB Indirection Unit for SSDs, in advance of QLC NAND flash replacing HDDs in the data center.

China is actively pursuing a national strategy of replacing imported HDDs and SSDs to fully in-house produced SSDs. As such, the country’s investment and development focus has pivoted to creating domestic etching equipment, which plays a key role in determining how many layers can be stacked on a chip to achieve higher storage capacity and performance at lower cost. This aligns well with flash memory development and manufacturing because flash is not limited by lithography but is more deposition-centric.

“Tachyum supports initiatives that work on high-performance flash solutions to keep pace with emerging AI development,” said Dr. Radoslav Danilak, founder and CEO of Tachyum. “We work with organizations planning production through Asia, the Middle East and Europe to help them optimize flash for AI applications. We see China as an important partner, with the Chinese government continuously increasing its investments to better leverage domestically produced flash with the goal to win the NAND, SSD and storage spaces.”

As a Universal Processor offering industry-leading performance for all workloads, Prodigy-powered data center servers can seamlessly and dynamically switch between computational domains (such as AI/ML, HPC, and cloud) with a single homogeneous architecture. By eliminating the need for expensive dedicated AI hardware and dramatically increasing server utilization, Prodigy reduces CAPEX and OPEX significantly while delivering unprecedented data center performance, power, and economics. Prodigy integrates 256 high-performance custom-designed 64-bit compute cores to deliver up to 18x the highest performing GPU for AI applications, 3x the performance of the highest-performing x86 processors for cloud workloads, and up to 8x that of the highest performing GPU for HPC.

Follow Tachyum

About Tachyum

Tachyum Enhances Prodigy Universal Processor Behind eBPF Port

Tue, 29 Jul 2025 00:00:00 +0000

LAS VEGAS, July 29, 2025 –Tachyum® today announced that it has successfully ported the eBPF Just-In-Time (JIT) compiler on its Prodigy® Universal Processor software emulation platform.

eBPF is a technology that can run sandboxed programs in a privileged context, such as the operating system kernel. It is used to extend the capabilities of the kernel safely and efficiently without requiring changes to the kernel source code or having to load kernel modules. Uses include kernel tracking, profiling and debugging; performance profiling; network packing filtering; security policies and events; and task scheduling. eBPF JIT is around 10 times faster than a generic eBPF interpreter.

Tachyum’s engineers ported Kprobes (Kernel Probes), which plays an important role in the eBPF JIT technology and serves as trigger for eBPF subroutines.

“Our software emulation system is an important part of ensuring that existing applications can run optimally on Prodigy processors and verifies that they are fully able to reap the benefits of high performance, low power and lower total cost of ownership when compared to running in traditional data centers,” said Dr. Radoslav Danilak, founder and CEO of Tachyum. “Porting the eBPF JIT compiler to our system is a valuable enhancement for Tachyum’s customers and an important step in unlocking the full potential of Prodigy.”

Tachyum’s demonstration, available in a video below, shows an Eunomia execsnoop example. Eunomia is an open-source organization dedicated to exploring and enhancing the eBPF ecosystem.

Follow Tachyum

About Tachyum

Tachyum is transforming the economics of AI, HPC, public and private cloud workloads with Prodigy, the world’s first Universal Processor. Prodigy unifies the functionality of a CPU, a GPU, and a TPU in a single processor to deliver industry-leading performance, cost and power efficiency for both specialty and general-purpose computing. As global data center emissions continue to contribute to a changing climate, with projections of their electricity demand worldwide set to more than double by 2030⁽¹⁾, the ultra-low power Prodigy is positioned to help balance the world’s appetite for computing at a lower environmental cost. Tachyum received a major purchase order from a US company to build a large-scale system that can deliver more than 50 exaflops performance, which will exponentially exceed the computational capabilities of the fastest inference or generative AI supercomputers available anywhere in the world today. Tachyum has offices in the United States, Slovakia and the Czech Republic. For more information, visit https://www.tachyum.com/.

https://www.iea.org/reports/energy-and-ai

Tachyum Radically Cuts the Cost of DeepSeek by Quantizing it to 2-bits

Tue, 03 Jun 2025 00:00:00 +0000

LAS VEGAS, June 3, 2025 –Tachyum® today announced the release of a new white paper detailing how it efficiently scales Large Language Model (LLM) training and inference through the Mixture of Experts (MoE) approach. The company’s method is further improved by a DeepSeekMoE architecture with 4-bit FP4 data types for activations quantization and 2-bit Tachyum AI (TAI2) sparse weights quantization.

The white paper, “Tachyum Successfully Quantized DeepSeek LLM to its 2-bit TAI2,” illustrates how Tachyum integrates MoE with low-bit data formats to unlock scalable AI with unmatched efficiency. The combination allows for the development of more powerful models while significantly lowering resource requirements.

MoEs can match the performance of dense models using approximately 4 times less computing and memory bandwidth, while only memory capacity needs to be increased by approximately 4 times. It is expected that that ratio will continue to grow. This architecture benefits from Tachyum’s proprietary high-performance memory, eliminating the need for costly high-bandwidth memory (HBM) solutions. Successfully quantizing DeepSeek LLM to 2-bit TAI2 further doubles benefit of DeepSeekMoE LLM compared to other architectures.

Tachyum’s AI researchers applied FP4 activation quantization and 2-bit TAI2 sparse weights quantization to DeepSeekMoE and Llama 3.1 models. Benchmark testing demonstrated up to 25x faster inference speeds and a 20x cost reduction per token, marking a major leap in LLM deployment efficiency.

“DeepSeek approach has shown the potential to make next-generation models 10 times more efficient at today’s costs, avoiding the exponential scaling challenges faced by organizations today,” said Dr. Radoslav Danilak, founder and CEO of Tachyum. “With the Prodigy platform, we’re enabling this kind of breakthrough efficiency for AI applications at global scale.”

The white paper also emphasizes the critical role of Tachyum’s hardware in facilitating this transformation, showcasing the Prodigy Universal Processor’s ability to support high-efficiency AI workloads with industry-leading performance.

Those interested in reading the “Tachyum Successfully Quantized DeepSeek LLM to its 2-bit TAI2” white paper can visit the company website to download.

Follow Tachyum

About Tachyum

Tachyum Demonstrates System Management on Prodigy Universal Processor

Tue, 06 May 2025 00:00:00 +0000

LAS VEGAS, May 6, 2025 – Tachyum® today announced that it has successfully demonstrated System Management on its Prodigy® Universal Processor, validating its ability to run system-wide functions such as power management, system hardware control, or proprietary OEM-designed code.

System Management covers privileged System Management Mode (SMM) level, dedicated System Management Interrupts (SMI) and SMM firmware. In x86-based processors, SMM exists as a special-purpose operating mode with a higher effective privilege level than the hypervisor. SMM is used only by SMM firmware, not by application software or general-purpose systems software.

Validating SMM shows that Tachyum’s Prodigy is capable of running in this distinct and isolated processor environment, independently of the operating system and transparently to executive and software applications. SMM uses separate on-chip protected static RAM (SRAM), and is installed by the Unified Extensible Firmware Interface (UEFI).

Among SMM drivers, SMI is critical for managing key hardware tasks and other system support functions like energy and thermal management, initializing memory, and adjusting clock speed.

“SMM and SMI are essential components of a mature processor architecture, and this demonstration signals that Prodigy is able to perform crucial management functions,” said Dr. Radoslav Danilak, founder and CEO of Tachyum. “As we advance towards commercialization of Prodigy, Tachyum’s engineers continue to validate that Prodigy offers the proper features for demanding enterprise and data center installations.”

The first version of Prodigy will implement an SMM-like environment with new features and improvements, and a software interface for SMM. Tachyum has also developed a separate privilege mode for SMM to provide availability, safety, and isolation for SMM and SMI.

Tachyum’s demonstration, available in a video below, shows Prodigy performing the SMM function of reading the CPU temperature and configuring thermal zones from Tachyum Linux OS. Tachyum’s SMM reads the value of CPU temperature from the embedded HW sensor, validates it, and takes action to protect the CPU. The Linux kernel can configure the temperature thresholds and read actual temperature via the SMM software interface, similar to OS system calls.

Follow Tachyum

About Tachyum

Tachyum Joins UALink to Advance the Future of Data Center AI Connectivity

Tue, 15 Apr 2025 00:00:00 +0000

LAS VEGAS, April 15, 2025 – Tachyum® today announced its membership in Ultra Accelerator Link™ (UALink™), a consortium of industry-leading hyperscalers and AI infrastructure providers, to ensure the Prodigy® Universal Processor is compatible with emerging accelerator ecosystems.

UALink is a high-speed accelerator interconnect technology that advances next-generation AI/ML cluster performance. The group’s goal is to define and establish an open, interoperable industry standard for high-performance computing connections in scale-up AI environments that enable AI accelerators to communicate more effectively.

Tachyum’s membership connects the company with key players in the industry driving the future of AI infrastructure. Tachyum looks to extend scale up of its systems in future designs through the incorporation of UALink switches.

“UALink promises high-bandwidth, low-latency connectivity for accelerators, which aligns with Tachyum’s vision for our Prodigy processor architecture,” said Dr. Radoslav Danilak, founder and CEO of Tachyum. “As a non-proprietary, open specification standard, UALink is expected to be adopted by many players across the entire industry making it ideal for addressing the need for efficient multi-accelerator communications as AI models grow. We look forward to working in concert with the consortium to further advance AI computing.”

Follow Tachyum

About Tachyum

Tachyum Demonstrates DRAM Failover for Large Scale AI on Prodigy FPGA prototype

Thu, 27 Mar 2025 00:00:00 +0000

LAS VEGAS, 27 March, 2024 –Tachyum® today announced that it has successfully enabled DRAM Failover correct system on its Prodigy Universal Processor, demonstrating enhanced reliability for even larger-scale AI and HPC applications even in the case of DRAM chip failures.

Tachyum’s DRAM Failover is an advanced memory error correction technology that improves the reliability of DRAM and provides a higher level of protection than traditional Error Correction Code (ECC). DRAM Failover can correct multi-bit errors within a single memory chip or across multiple memory chips, allowing continued memory operation in the event of device-level faults in memory. With DRAM Failover, even a whole DRAM chip failure can be tolerated without affecting the system and applications.

Because they help preserve customer data and maintain system availability, correct systems like DRAM Failover have become popular among HPC systems and high-end servers with large memory capacities. In the case of AI clusters reaching 100,000 accelerators, the time between failures is hours—scaling to even larger AI clusters presents a major reliability challenge.

A single Prodigy processor with 640 or 1280 DRAM chips attached would mean 64,000,000 DRAM chips, a significant scale. With DRAM Failover correct, a failing DRAM die per DIMM would not affect the operation of the system and will not cause failure with Prodigy, unlike GPU accelerators.

Tachyum’s DRAM Failover validation responds to the market’s interest in large-scale AI, including Cognitive AI and Artificial General Intelligence (AGI), and signals the company’s dedication to robust Reliability, Accessibility and Serviceability (RAS) features.

“This capability is essential to increase the scale of AI training as it moves from Large Language Models and Generative AI to much bigger systems needed for Cognitive AI and AGI,” said Dr. Radoslav Danilak, founder and CEO of Tachyum. “The importance of using DRAM Failover on Tachyum’s platform will be even more evident as we increase memory capacity per Prodigy processor with every generation.”

For example, AI innovator DeepSeek is enabling open-source LLMs to scale with DRAM capacity rather than bandwidth, making DeepSeek a compelling use case for Prodigy. DeepSeek’s efficiency makes its technology more akin to how a human brain works: only a fraction of neurons fire in response to stimuli. As this paradigm is increasingly adopted by the industry, and improves over time, the benefits of ever-bigger DRAM capacity—without the reliability challenges—will further establish the advantages of Prodigy.

A video demonstrating DRAM Failover on the Prodigy FPGA prototype is available below.

Follow Tachyum

About Tachyum

Tachyum Delivers a Release Candidate of its Prodigy Software Distribution Package Supporting 4x256 Cores

Tue, 18 Feb 2025 00:00:00 +0000

LAS VEGAS, February 18, 2025 –Tachyum® today announced that software distribution for the Prodigy® Universal Processor has entered the release candidate (RC1) testing phase. This completely integrated software stack and package is now available to selected early adopters and customers.

The company’s software distribution package combines key advancements to deliver optimal performance for Prodigy. Notable updates to RC1 include the inclusion of GCC 14.2, Linux 6.12, LLVM 19, FreeBSD 14.1 and QEMU 9.11. The package also provides users with an optimization guide and performance-tuning tools to ensure maximum efficiency.

Tachyum’s software distribution now supports up to 1,024 cores in 4-socket configurations with a default 64KB page size and 4KB page support as a backup solution for OEM partners. With this robust capability, Prodigy is optimized for demanding workloads, offering seamless scalability for various enterprise applications, including exascale supercomputing, big AI, cloud, data analytics, big data, databases, storage, edge computing and web hosting.

The Prodigy Series has been expanded to include multiple new SKUs, including a top-tier configuration featuring 256 high-performance, 64-bit cores; 16 DDR5 memory controllers running at DDR5-7200; and 96 lanes of PCIe 5.0.

“As Tachyum continues to advance Prodigy towards full-scale production and distribution, moving its software distribution into release candidate stage further solidifies our commitment to meeting the diverse needs of the next-generation computing landscape,” said Dr. Radoslav Danilak, founder and CEO of Tachyum. “We look forward to receiving feedback of the software distribution package from our early-adopter partners and customers during this phase and are eager to ensure Prodigy is ready to tackle the demands of hyperscale data centers, private cloud, AI and high-performance computing for generations to come.”

A video demonstrating the capabilities of Prodigy’s software distribution package is available below.

In the video, Tachyum shows various boot options of Linux and FreeBSD operating systems from a partitioned drive using UEFI firmware and with different settings, which allow crafting images for more specific usage scenarios. Tachyum demonstrates real-world package building like building Bash shell using GCC and Clang including debugging. Clang is a newer compiler than GCC benefiting from the LLVM project. It is essential to support newer compilers in software distributions as many projects benefit the LLVM ecosystem and open wide opportunities for porting modern software making Tachyum’s software platform future-proof. Further, Tachyum shows the running of scripting languages like Python and Perl, which are a must-have in today’s IT ecosystem.

Follow Tachyum

About Tachyum

Tachyum Enables Protection from Quantum Computer Cryptographic Attacks

Tue, 21 Jan 2025 00:00:00 +0000

LAS VEGAS, January 21, 2025 –Tachyum® today announced that it has ported all four of the National Institute of Standards and Technology’s (NIST) quantum-safe algorithms across Prodigy® software distributions to ensure data center deployments leveraging the company’s universal processor are quantum-resistant and future-proofed for data security.

Quantum computing is a powerful emerging technology that enables much higher performance for certain algorithms, such as optimization and molecular simulation. Quantum computers operate in fundamentally different ways from traditional computers and pose a threat to data security with Shor’s algorithm – a quantum-computing algorithm that is predicted to break current data security processes within the next 5 years.

Tachyum’s software engineering team has ported and verified the four quantum-resistant asymmetric algorithms selected by NIST to Prodigy – ML-KEM, ML-DSA, SLH-DSA and Falcon. Prodigy also supports the quantum-safe AES-256 standard, which has already been optimized. Tachyum’s post-quantum cryptography (PQC) will run on all Prodigy platforms.

“Tachyum takes security very seriously – from both a hardware and software perspective – so the development of quantum computer-proof data security methods is critical to us maintaining such a commitment,” said Dr. Radoslav Danilak, founder and CEO of Tachyum. “As such, we will continue to monitor future cryptograph standards to ensure that Prodigy-based systems remain capable of providing the highest level of security and optimum performance for customers and partners.”

As a Universal Processor offering industry-leading performance for all workloads, Prodigy-powered data center servers can seamlessly and dynamically switch between computational domains (such as AI/ML, HPC, and cloud) with a single homogeneous architecture. By eliminating the need for expensive dedicated AI hardware and dramatically increasing server utilization, Prodigy reduces CAPEX and OPEX significantly while delivering unprecedented data center performance, power, and economics. Prodigy integrates 192 high-performance custom-designed 64-bit compute cores, to deliver up to 4.5x the performance of the highest-performing x86 processors for cloud workloads, up to 3x that of the highest performing GPU for HPC, and 6x for AI applications.

Those interested in learning more about Tachyum’s approach to future-proofing data security can download the white paper “Tachyum Prodigy Solutions for Post-Quantum Cryptography”.

Follow Tachyum

,https://x.com/tachyum

About Tachyum

Tachyum Publishes Prodigy Performance Optimization Manual

Tue, 17 Dec 2024 00:00:00 +0000

LAS VEGAS, December 17, 2024 –Tachyum® today announced that it has released a 1,600-page Performance Optimization Manual for its Prodigy® Universal Processor FPGA hardware.

Tachyum’s Prodigy Performance Optimization Manual provides detailed information on how to fully benefit from the performance features that are built into Prodigy. It includes the required design guidelines for the development of high-performance software for a broad range of applications, including cloud, AI, and HPC.

The manual outlines Prodigy’s revolutionary new Universal Processor microarchitecture, and intrinsic functions as well as processor instructions, throughputs, and latencies. Tachyum has also included Prodigy’s performance counters, which enable performance monitoring and analysis across a wide array of run-time events.

The Prodigy Performance Optimization Manual describes special considerations for performance optimization that include dispatch constraints, load/store alignment, optimizing memory routines, branch instruction alignment, special register access, register forwarding hazards, cache maintenance operations, and complex instructions.

“Software programmers, test engineers, compiler developers, and systems and solutions engineers will appreciate the opportunity to take this deep dive into how Prodigy offers inherent performance benefits for efficient processing of AI, cloud, and HPC workloads,” said Dr. Radoslav Danilak, founder and CEO of Tachyum. “Prodigy’s integrated features will help users achieve industry-leading compute efficiency to derive insights faster, to perform research faster, to generate results faster.”

The Prodigy Instruction Set Architecture (ISA) includes a large number of vector and matrix instructions that optimize the performance and efficiency of vector and matrix operations. The Prodigy ISA is a mix of RISC and CISC but doesn’t include any complex and/or long variable length, inefficient instructions that many CISC processors have. All instructions are either 32 or 64-bits wide, and some instructions include memory accesses to optimize performance.

As a Universal Processor offering industry-leading performance for all workloads, Prodigy-powered data center servers can seamlessly and dynamically switch between computational domains (such as AI/ML, HPC, and cloud) with a single homogeneous architecture. By eliminating the need for expensive dedicated AI hardware and dramatically increasing server utilization, Prodigy reduces CAPEX and OPEX significantly while delivering unprecedented data center performance, power, and economics. Prodigy integrates 192 high-performance custom-designed 64-bit compute cores, to deliver up to 4.5x the performance of the highest-performing x86 processors for cloud workloads, up to 3x that of the highest performing GPU for HPC, and 6x for AI applications.

Follow Tachyum

About Tachyum

Tachyum is transforming the economics of AI, HPC, public and private cloud workloads with Prodigy, the world’s first Universal Processor. Prodigy unifies the functionality of a CPU, a GPU, and a TPU in a single processor to deliver industry-leading performance, cost and power efficiency for both specialty and general-purpose computing. As global data center emissions continue to contribute to a changing climate, with projections of their consuming 10 percent of the world’s electricity by 2030, the ultra-low power Prodigy is positioned to help balance the world’s appetite for computing at a lower environmental cost. Tachyum received a major purchase order from a U.S. company to build a large-scale system that can deliver more than 50 exaflops performance, which will exponentially exceed the computational capabilities of the fastest inference or generative AI supercomputers available anywhere in the world today. When complete in 2026, the Prodigy-powered system will deliver a 25x multiplier vs. the world’s fastest conventional supercomputer – built just this year – and will achieve AI capabilities 25,000x larger than models for ChatGPT4. Tachyum has offices in the United States, Slovakia and the Czech Republic. For more information, visit https://www.tachyum.com/.

Tachyum Runs AI Integer Matrix Operations on Prodigy FPGA Hardware

Tue, 03 Dec 2024 00:00:00 +0000

LAS VEGAS, December 3, 2024 –Tachyum® today announced that it has successfully validated integer matrix operations running on its Prodigy® Universal Processor FPGA hardware.

The Tachyum team tested and verified vector operations, 8-bit integer matrix operations for image classification using a Resnet model with custom convolution and linear operators on Prodigy.

Tachyum’s Prodigy was designed to handle matrix and vector processing from the ground up. Among Prodigy’s vector and matrix features are support for a range of data types (FP64, FP32, TF32, BF16, Int8, FP8, FP4 and TAI); 2x1024-bit vector units per core; AI sparsity and super-sparsity support; and no penalty for misaligned vector loads or stores when crossing cache lines. This built-in support offers high performance for AI training and inference workloads, increases performance, and reduces memory utilization.

“Tachyum’s AI software stack supports AI applications with demanding computational requirements out of the box, rather than as an afterthought, which will radically change the efficiency and productivity of AI,” said Dr. Radoslav Danilak, founder and CEO of Tachyum. “Data centers are among the most energy-hungry facilities in the world, and power infrastructures cannot accommodate the rapid adoption of AI without a platform like Prodigy that is capable of efficient AI processing, plus standard workloads.”

Tachyum will validate all supported data types. The next plan is to validate matrix operations for the FP8 data type this year on Prodigy FPGA hardware.

As a Universal Processor offering industry-leading performance for all workloads, Prodigy-powered data center servers can seamlessly and dynamically switch between computational domains (such as AI/ML, HPC, and cloud) with a single homogeneous architecture. By eliminating the need for expensive dedicated AI hardware and dramatically increasing server utilization, Prodigy reduces CAPEX and OPEX significantly while delivering unprecedented data center performance, power, and economics. Prodigy integrates 192 high-performance custom-designed 64-bit compute cores, to deliver up to 4.5x the performance of the highest-performing x86 processors for cloud workloads, up to 3x that of the highest performing GPU for HPC, and 6x for AI applications.

A video demonstration of 8-bit integer matrix operations running on Prodigy FPGA is available below.

Follow Tachyum

About Tachyum

Tachyum is transforming the economics of AI, HPC, public and private cloud workloads with Prodigy, the world’s first Universal Processor. Prodigy unifies the functionality of a CPU, a GPU, and a TPU in a single processor to deliver industry-leading performance, cost and power efficiency for both specialty and general-purpose computing. As global data center emissions continue to contribute to a changing climate, with projections of their consuming 10 percent of the world’s electricity by 2030, the ultra-low power Prodigy is positioned to help balance the world’s appetite for computing at a lower environmental cost. Tachyum received a major purchase order from a US company to build a large-scale system that can deliver more than 50 exaflops performance, which will exponentially exceed the computational capabilities of the fastest inference or generative AI supercomputers available anywhere in the world today. When complete in 2026, the Prodigy-powered system will deliver a 25x multiplier vs. the world’s fastest conventional supercomputer – built just this year – and will achieve AI capabilities 25,000x larger than models for ChatGPT4. Tachyum has offices in the United States, Slovakia and the Czech Republic. For more information, visit https://www.tachyum.com/.

Tachyum Joins the Slovak-Saudi Chamber of Commerce

Tue, 19 Nov 2024 00:00:00 +0000

LAS VEGAS, November 19, 2024 –Tachyum® today announced its membership in the Slovak-Saudi Chamber of Commerce (SSCC) as part of its commitment to create and expand business connections throughout Middle Eastern markets. This move closely follows several other association agreements that Tachyum has entered to further develop partnerships with various public and private sector entities that support economic growth, trade and investments with the respective regions.

SSCC is a non-profit organization established in cooperation with the Slovak Embassy in Riyadh – Saudi Arabia’s capital and main financial hub. The mission of SSCC is to develop and deepen bilateral trade relations between Slovakia and Saudi Arabia while contributing to improving annual economic relations between the EU and the MEA.

The Kingdom of Saudi Arabia (KSA) is the biggest country in the MEA and plays a key role in driving importance to the region. Within its government program, “Saudi Vision 2030,” and the intensive hard work led by His Royal Highness Crown Prince Mohammed bin Salman, KSA is seeing substantial gains across social and economic platforms. As part of KSA’s growing influence on the global stage, adoption and implementation of advanced technologies, including AI, is a top priority.

The Saudi Vision 2030 plan is to amplify AI and increase contributions to the economy by more than 12% of GDP. KSA has articulated its ambitions to become a hub for digital innovation and create jobs for software developers, data analysts, cybersecurity experts and AI specialists. Tachyum is positioned to help KSA achieve these goals behind its Prodigy® Universal Processor family, which encompasses multiple product SKUs that address a wide range of workloads, providing the same high performance and efficiency for applications that range from exascale supercomputers to Big AI, hyperspace data centers and edge servers.

“We launched our activities in KSA by presenting our Prodigy Universal Processor at the most prestigious tech event, LEAP, in Riyadh,” said Marwan Ajam Oghli, Business Development Manager at Tachyum. “Membership in the SSCC is a continuation of our efforts to develop relationships and partnerships within the region. We look forward to cooperating with the SSCC as well as with the newly appointed Slovak Honorary Counsel in KSA Mr. Abdulrahman Al Otaishan.”

As a Universal Processor offering industry-leading performance for all workloads, Prodigy-powered data center servers can seamlessly and dynamically switch between computational domains (such as AI/ML, HPC, and cloud) with a single homogeneous architecture. By eliminating the need for expensive dedicated AI hardware and dramatically increasing server utilization, Prodigy reduces CAPEX and OPEX significantly while delivering unprecedented data center performance, power, and economics. Prodigy integrates 192 high-performance custom-designed 64-bit compute cores, to deliver up to 4.5x the performance of the highest-performing x86 processors for cloud workloads, up to 3x that of the highest performing GPU for HPC, and 6x for AI applications.

Follow Tachyum

,https://x.com/tachyum

About Tachyum

Tachyum is transforming the economics of AI, HPC, public and private cloud workloads with Prodigy, the world’s first Universal Processor. Prodigy unifies the functionality of a CPU, a GPU, and a TPU in a single processor to deliver industry-leading performance, cost and power efficiency for both specialty and general-purpose computing. As global data center emissions continue to contribute to a changing climate, with projections of their consuming 10 percent of the world’s electricity by 2030, the ultra-low power Prodigy is positioned to help balance the world’s appetite for computing at a lower environmental cost. Tachyum received a major purchase order from a US company to build a large-scale system that can deliver more than 50 exaflops performance, which will exponentially exceed the computational capabilities of the fastest inference or generative AI supercomputers available anywhere in the world today. When complete in 2026, the Prodigy-powered system will deliver a 25x multiplier vs. the world’s fastest conventional supercomputer – built just this year – and will achieve AI capabilities 25,000x larger than models for ChatGPT4. Tachyum has offices in the United States, Slovakia and the Czech Republic. For more information, visit https://www.tachyum.com/.

Tachyum Demonstrated Hardware Running AutoFDO Optimization Flow

Tue, 12 Nov 2024 00:00:00 +0000

LAS VEGAS, November 12, 2024 –Tachyum® today announced that it has successfully demonstrated enhanced hardware-assisted sampling running on its Prodigy® Universal Processor FPGA Emulation System, marking a breakthrough in the application of advanced compiler optimizations.

Compiler optimizations are critical for enhancing performance, reducing response times, minimizing storage footprints, and improving the overall Total Cost of Ownership (TCO) across a wide variety of modern computational workloads. Among the most powerful techniques employed by modern compilers are link-time optimizations (LTO), profile-guided optimizations (PGO) and feedback-directed optimizations (AutoFDO). These techniques leverage runtime data to fine-tune software performance, achieving results beyond what static optimizations can deliver.

Tachyum’s latest demonstration shows that these advanced AutoFDO optimizations—enabled through its Prodigy hardware platform—deliver performance improvements of 10-15%. By collecting data during program execution, PGO and AutoFDO with LTO allow compilers to adjust code in real time, optimizing how software runs and improving efficiency.

PGO traditionally involves modifying generated code to collect detailed execution information, allowing for full profiling but at the cost of added complexity. In contrast, AutoFDO takes a more practical approach by using specialized hardware blocks within the Prodigy processor to collect performance data with minimal overhead. This method enables feedback from production binaries without the need for code modification, providing a flexible and efficient solution for real-world applications. Though AutoFDO may have slightly lower profile precision compared to PGO, this is mitigated by increasing sampling time and merging data from multiple instances.

Both PGO and AutoFDO require re-compilation to apply insights from the collected profiles. While the results of both techniques are typically similar, AutoFDO’s hardware-assisted approach offers significant advantages in terms of ease of implementation and scalability. Tachyum’s Prodigy Universal Processor platform supports out-of-box AutoFDO flow for hardware-assisted profile collection, offering customers a choice of optimization techniques depending on their specific needs.

“This latest milestone demonstrates our continued commitment to pushing the boundaries of computational performance,” said Dr. Radoslav Danilak, founder and CEO of Tachyum. “By enabling hardware-assisted sampling, we are not only simplifying the process of advanced compiler optimizations but also delivering tangible improvements to software performance, making it easier for developers to achieve faster, more efficient applications.”

As a Universal Processor offering industry-leading performance for all workloads, Prodigy-powered data center servers can seamlessly and dynamically switch between computational domains (such as AI/ML, HPC, and cloud) with a single homogeneous architecture. By eliminating the need for expensive dedicated AI hardware and dramatically increasing server utilization, Prodigy reduces CAPEX and OPEX significantly while delivering unprecedented data center performance, power, and economics. Prodigy integrates 192 high-performance custom-designed 64-bit compute cores, to deliver up to 4.5x the performance of the highest-performing x86 processors for cloud workloads, up to 3x that of the highest performing GPU for HPC, and 6x for AI applications.

A video demonstrating hardware support for profile collection used for AutoFDO is now available for viewing below.

Follow Tachyum

About Tachyum

Tachyum is transforming the economics of AI, HPC, public and private cloud workloads with Prodigy, the world’s first Universal Processor. Prodigy unifies the functionality of a CPU, a GPU, and a TPU in a single processor to deliver industry-leading performance, cost and power efficiency for both specialty and general-purpose computing. As global data center emissions continue to contribute to a changing climate, with projections of their consuming 10 percent of the world’s electricity by 2030, the ultra-low power Prodigy is positioned to help balance the world’s appetite for computing at a lower environmental cost. Tachyum received a major purchase order from a US company to build a large-scale system that can deliver more than 50 exaflops performance, which will exponentially exceed the computational capabilities of the fastest inference or generative AI supercomputers available anywhere in the world today. When complete in 2026, the Prodigy-powered system will deliver a 25x multiplier vs. the world’s fastest conventional supercomputer – built just this year – and will achieve AI capabilities 25,000x larger than models for ChatGPT4. Tachyum has offices in the United States, Slovakia and the Czech Republic. For more information, visit https://www.tachyum.com/.

Tachyum Joins European Chamber of Commerce in China as a European Participant of IC China

Tue, 29 Oct 2024 00:00:00 +0000

LAS VEGAS, October 29, 2024 – Tachyum® today announced its participation in the China International Semiconductor Expo (IC CHINA), an exclusive exhibition taking place in the China National Convention Center in Beijing November 18-20.

For Europe, China is the largest business partner in the world by a factor close to 2x compared to any other business partner of the EU. Tachyum has Tachyum s.r.o. in Slovakia and Tachyum Czech Republic, s.r.o. in Europe. Participating in EUCCC and IC China Tachyum fulfills stated support of European Union efforts in the semiconductor space and technological sovereignty as well as helping the Slovak government efforts to search for further opportunities for economic cooperation with China.

Tachyum will engage in networking and business opportunities at IC CHINA as a member of the European Chamber of Commerce in China (EUCCC), which is widely recognized as the official voice of European business in China. EUCCC helps European companies grow their businesses in China with a goal of making markets fair and equitable – a crucial condition to those in high-tech fields like chips, where following rules and protecting new ideas are paramount.

As part of the EUCCC, Tachyum has a competitive advantage over other international enterprises in Chinese markets. Membership provides access to up-to-the-minute information, policy updates and business contacts, offering tangible strategic benefits in successfully planning and achieving sales and growth operations within the world’s largest manufacturing economy.

“Being among the top enterprises invited to be a part of the EUCCC and securing the opportunity to participate in such prestigious events like IC CHINA opens up a wealth of business partnerships and networking opportunities that will lead to tremendous advantages in the semiconductor industry,” said Dr. Radoslav Danilak, founder and CEO of Tachyum. “I fully understand the words of the Slovak prime minister, who stated that the upcoming visit to China is the most important trip that he will take this year. Tachyum will proactively contribute to gatherings and talks organized by the Chamber that will ensure that industry will have the avenues necessary to grow and thrive business between the EU and China.”

As a Universal Processor offering industry-leading performance for all workloads, Prodigy-powered data center servers can seamlessly and dynamically switch between computational domains (such as AI/ML, HPC, and cloud) with a single homogeneous architecture. By eliminating the need for expensive dedicated AI hardware and dramatically increasing server utilization, Prodigy reduces CAPEX and OPEX significantly while delivering unprecedented data center performance, power, and economics. Prodigy integrates 192 high-performance custom-designed 64-bit compute cores, to deliver up to 4.5x the performance of the highest-performing x86 processors for cloud workloads, up to 3x that of the highest performing GPU for HPC, and 6x for AI applications.

Follow Tachyum

About Tachyum

Tachyum is transforming the economics of AI, HPC, public and private cloud workloads with Prodigy, the world’s first Universal Processor. Prodigy unifies the functionality of a CPU, a GPU, and a TPU in a single processor to deliver industry-leading performance, cost and power efficiency for both specialty and general-purpose computing. As global data center emissions continue to contribute to a changing climate, with projections of their consuming 10 percent of the world’s electricity by 2030, the ultra-low power Prodigy is positioned to help balance the world’s appetite for computing at a lower environmental cost. Tachyum received a major purchase order from a US company to build a large-scale system that can deliver more than 50 exaflops performance, which will exponentially exceed the computational capabilities of the fastest inference or generative AI supercomputers available anywhere in the world today. When complete in 2026, the Prodigy-powered system will deliver a 25x multiplier vs. the world’s fastest conventional supercomputer – built just this year – and will achieve AI capabilities 25,000x larger than models for ChatGPT4. Tachyum has offices in the United States, Slovakia, and the Czech Republic. For more information, visit https://www.tachyum.com/.

Tachyum Running Linux and Applications with 64KB Pages on FPGA

Tue, 15 Oct 2024 00:00:00 +0000

LAS VEGAS, October 15, 2024 – Tachyum® today announced that it has successfully achieved the desired performance and stability of the Linux operating system and applications running as part of QEMU emulations and will expand all testing and optimization processes of its Prodigy® software distribution to 64KB page size on the FPGA prototype.

Tachyum first announced its intentions to move to 64KB page size as the default of its Prodigy software distribution – a completely integrated software stack and package available to early adopters and customers as a pre-installed image as part of beta testing – in June. After completing testing of Linux 6.8 and applications in QEMU software emulation, Tachyum’s move to the 64KB page size is expected to yield up to 10-15% better performance than 4KB pages depending on the application. 4KB support is now only supported for special configurations and will be deprecated in Prodigy 2.

“For decades, the default translation page size on most CPU architectures was 4KB,” said Dr. Radoslav Danilak, founder and CEO of Tachyum. “With the amount of data being processed and with application sizes growing, adhering to 4KB size leads to increasingly larger overhead. For increased efficiency, Tachyum decided to make 64KB page size the default in its offerings. This successful transition will ensure that the performance of Prodigy’s software distribution is optimized to meet applications’ needs.”

As a Universal Processor offering industry-leading performance for all workloads, Prodigy-powered data center servers can seamlessly and dynamically switch between computational domains (such as AI/ML, HPC, and cloud) with a single homogeneous architecture. By eliminating the need for expensive dedicated AI hardware and dramatically increasing server utilization, Prodigy reduces CAPEX and OPEX significantly while delivering unprecedented data center performance, power, and economics. Prodigy integrates 192 high-performance custom-designed 64-bit compute cores, to deliver up to 4.5x the performance of the highest-performing x86 processors for cloud workloads, up to 3x that of the highest performing GPU for HPC, and 6x for AI applications.

To enable THP and mTHP, Tachyum is now running on hardware with Linux 6.10. A video demonstrating booting and running applications at 64KB page size on Prodigy FPGA prototype is available below.

Follow Tachyum

About Tachyum

Tachyum is transforming the economics of AI, HPC, public and private cloud workloads with Prodigy, the world’s first Universal Processor. Prodigy unifies the functionality of a CPU, a GPU, and a TPU in a single processor to deliver industry-leading performance, cost and power efficiency for both specialty and general-purpose computing. As global data center emissions continue to contribute to a changing climate, with projections of their consuming 10 percent of the world’s electricity by 2030, the ultra-low power Prodigy is positioned to help balance the world’s appetite for computing at a lower environmental cost. Tachyum received a major purchase order from a US company to build a large-scale system that can deliver more than 50 exaflops performance, which will exponentially exceed the computational capabilities of the fastest inference or generative AI supercomputers available anywhere in the world today. When complete in 2026, the Prodigy-powered system will deliver a 25x multiplier vs. the world’s fastest conventional supercomputer – built just this year – and will achieve AI capabilities 25,000x larger than models for ChatGPT4. Tachyum has offices in the United States, Slovakia and the Czech Republic. For more information, visit https://www.tachyum.com/.

Tachyum Expands European Hub with New Offices in Czech Republic

Tue, 24 Sep 2024 00:00:00 +0000

LAS VEGAS, September 24, 2024 – Tachyum® today announced the opening of new positions in Brno, Czech Republic, to support the company’s continued growth as it nears tape-out of the Prodigy Universal Processor Chip. The first hardware engineers who have been hired are starting on November 1.

Brno is in the heart of a burgeoning Central European technology region that continues to attract international companies, innovation and talent. As such, it is an ideal location for Tachyum to attract qualified IT professionals interested in unlocking unprecedented performance, power efficiency and cost advantages within the data center to solve the most complex problems in big data analytics, deep learning, mobile and large-scale computing.

The Czech Republic is known for hardware design and software engineering talent, especially around Brno. Tachyum brings capable hardware and software engineers in the Czech Republic an unparalleled opportunity to work and grow from the current level to high-speed leading edge universal microprocessor designs. This will propel the Czech Republic’s advanced semiconductor industry forward within the European Union.

Having progressed to the final stage of development of Prodigy, Tachyum has become a magnet for highly qualified, world-class engineers and scientists. The company is seeking to staff senior professionals and talent at its Brno location to serve as CPU hardware design and verification engineers; software engineers for various system software porting, performance optimization, software testing and test automation; and engineers for advanced AI topics.

Brno’s proximity to Tachyum’s European headquarters in Bratislava, Slovakia, ensures effective cooperation and collaboration on common tasks. The company’s Slovakia facilities feature a 7,000 square foot R& D campus of modern Class-A offices with an internal datacenter, laboratory, and QA infrastructure.

“Tachyum is headed by a group of executive visionaries with hundreds of years of combined experience developing advanced data, networking and storage processing chips and systems,” said Dr. Radoslav Danilak, Tachyum founder and CEO. “As such, it is an ideal place for both experienced and novice European talent to work. With the locale advantages that we gain with our expansion into Brno and recent turbulent upheaval at IT giants, we expect to find an exceptional pool of qualified professionals interested in joining our mission.”

Follow Tachyum

About Tachyum

Tachyum is transforming the economics of AI, HPC, public and private cloud workloads with Prodigy, the world’s first Universal Processor. Prodigy unifies the functionality of a CPU, a GPU, and a TPU in a single processor to deliver industry-leading performance, cost and power efficiency for both specialty and general-purpose computing. As global data center emissions continue to contribute to a changing climate, with projections of their consuming 10 percent of the world’s electricity by 2030, the ultra-low power Prodigy is positioned to help balance the world’s appetite for computing at a lower environmental cost. Tachyum received a major purchase order from a US company to build a large-scale system that can deliver more than 50 exaflops performance, which will exponentially exceed the computational capabilities of the fastest inference or generative AI supercomputers available anywhere in the world today. When complete in 2026, the Prodigy-powered system will deliver a 25x multiplier vs. the world’s fastest conventional supercomputer – built just this year – and will achieve AI capabilities 25,000x larger than models for ChatGPT4. Tachyum has offices in the United States, Slovakia, and the Czech Republic. For more information, visit https://www.tachyum.com/.

Tachyum Joined Turkish Chamber Las Vegas To Expand Its Reach

Tue, 10 Sep 2024 00:00:00 +0000

LAS VEGAS, September 10, 2024 –Tachyum® today announced that it has become a member of the Turkish Chamber Las Vegas, a global business hub fostering strong ties between Turkish enterprises and the business community in Las Vegas. This move is part of Tachyum’s continued commitment to expanding and creating strong business connections in the Turkish, Middle Eastern and Central Asian markets.

The mission of TCLV is to foster robust economic ties between the Turkish and American business communities by facilitating trade, investment and collaboration. By leveraging the extensive network and deep understanding of both markets, TCLV empowers its members to navigate challenges, seize opportunities and achieve sustainable success. Being part of the TCLV gives Tachyum a competitive advantage by positioning it among leading Turkish enterprises.

Türkiye supports the development of semiconductors, a market that has been experiencing significant growth within the country in both public and private sectors, and has become one of the key consumers for computing performance. Tachyum aims to be a valuable partner as part of this emerging opportunity with its Prodigy® Universal Processor.

Members of the TCLV are familiar with Tachyum’s mission to revolutionize the data center and have helped develop connections that led to several Turkish companies joining Tachyum’s early adopter program. This opportunity gives early adopters an advantage by ensuring any specific requirements they may have are implemented as part of the Prodigy platform prior to its production and distribution.

“Turkish companies and enterprises are well-represented in industries where AI is – and will be – playing a significant role,” said Dr. Radoslav Danilak, founder and CEO of Tachyum. “In addition to its own burgeoning technology industry, Türkiye serves an important role in being a gateway between Asia, Europe, the Middle East and Africa due to its unique location. We believe any partnerships and goodwill that grows from our participation in TCLV will provide exponential success in delivering Prodigy to the world.”

As a Universal Processor offering industry-leading performance for all workloads, Prodigy-powered data center servers can seamlessly and dynamically switch between computational domains (such as AI/ML, HPC, and cloud) with a single homogeneous architecture. By eliminating the need for expensive dedicated AI hardware and dramatically increasing server utilization, Prodigy reduces CAPEX and OPEX significantly while delivering unprecedented data center performance, power, and economics. Prodigy integrates 192 high-performance custom-designed 64-bit compute cores, to deliver up to 4.5x the performance of the highest-performing x86 processors for cloud workloads, up to 3x that of the highest performing GPU for HPC, and 6x for AI applications.

Follow Tachyum

About Tachyum

Tachyum is transforming the economics of AI, HPC, public and private cloud workloads with Prodigy, the world’s first Universal Processor. Prodigy unifies the functionality of a CPU, a GPU, and a TPU in a single processor to deliver industry-leading performance, cost and power efficiency for both specialty and general-purpose computing. As global data center emissions continue to contribute to a changing climate, with projections of their consuming 10 percent of the world’s electricity by 2030, the ultra-low power Prodigy is positioned to help balance the world’s appetite for computing at a lower environmental cost. Tachyum received a major purchase order from a US company to build a large-scale system that can deliver more than 50 exaflops performance, which will exponentially exceed the computational capabilities of the fastest inference or generative AI supercomputers available anywhere in the world today. When complete in 2026, the Prodigy-powered system will deliver a 25x multiplier vs. the world’s fastest conventional supercomputer – built just this year – and will achieve AI capabilities 25,000x larger than models for ChatGPT4. Tachyum has offices in the United States and Slovakia. For more information, visit https://www.tachyum.com/.

Tachyum Builds Last FPGA Prototypes Batch Ahead of Tape-Out

Tue, 13 Aug 2024 00:00:00 +0000

LAS VEGAS, August 13, 2024 –Tachyum® today announced the final build of its Prodigy® FPGA emulation system in advance of chip production and general availability next year. As part of the announcement, the company is also ending its purchase program for prototype systems that was previously offered to commercial and federal customers.

These last hardware FPGA prototype units will ensure Tachyum hits its extreme-reliability test targets of more than 10 quadrillion cycles prior to tape-out and before the first Prodigy chips hit the market. Tachyum’s software emulation system – and access to it – is expanding with additional availability of open-source software ported ahead of Prodigy’s upstreaming.

Additional modifications included in this final build include:

Adding signals between boards to support an increased core count of more than 128 after Tachyum increased the core count to 192 last year
Minor fixes to support large-capacity DIMMs
Additional debug improvements
Modified BMC-UEFI hardware to simplify communication
Replacement of board-to-board connectors for improved experience

“Reaching this point of our development journey prior to tape-out and volume production of Prodigy processors next year is extremely gratifying,” said Dr. Radoslav Danilak, founder and CEO of Tachyum. “Our commitment to delivering the world’s smallest, fastest and greenest general-purpose chip has remained unwavering. Ensuring this happens Day One of launch has been a priority for us and we are excited to be on the precipice of this industry-altering release.”

As a Universal Processor offering industry-leading performance for all workloads, Prodigy-powered data center servers can seamlessly and dynamically switch between computational domains (such as AI/ML, HPC, and cloud) with a single homogeneous architecture. By eliminating the need for expensive dedicated AI hardware and dramatically increasing server utilization, Prodigy reduces CAPEX and OPEX significantly while delivering unprecedented data center performance, power, and economics. Prodigy integrates 192 high-performance custom-designed 64-bit compute cores, to deliver up to 4.5x the performance of the highest-performing x86 processors for cloud workloads, up to 3x that of the highest performing GPU for HPC, and 6x for AI applications.

Follow Tachyum

About Tachyum

Tachyum is transforming the economics of AI, HPC, public and private cloud workloads with Prodigy, the world’s first Universal Processor. Prodigy unifies the functionality of a CPU, a GPU, and a TPU in a single processor to deliver industry-leading performance, cost and power efficiency for both specialty and general-purpose computing. As global data center emissions continue to contribute to a changing climate, with projections of their consuming 10 percent of the world’s electricity by 2030, the ultra-low power Prodigy is positioned to help balance the world’s appetite for computing at a lower environmental cost. Tachyum received a major purchase order from a US company to build a large-scale system that can deliver more than 50 exaflops performance, which will exponentially exceed the computational capabilities of the fastest inference or generative AI supercomputers available anywhere in the world today. When complete in 2026, the Prodigy-powered system will deliver a 25x multiplier vs. the world’s fastest conventional supercomputer – built just this year – and will achieve AI capabilities 25,000x larger than models for ChatGPT4. Tachyum has offices in the United States and Slovakia. For more information, visit https://www.tachyum.com/.

Tachyum Successfully Tests BF16 on Prodigy FPGA Hardware

Tue, 30 Jul 2024 00:00:00 +0000

LAS VEGAS, July 30, 2024 – Tachyum® today announced that BF16 has been successfully tested and verified operational on its Prodigy® FPGA, ensuring increased throughput for users’ high-performance processing needs.

BF16, or bfloat16, is a shortened floating point data type based on the IEEE 32-bit single-precision floating point data type (f32). It is used to accelerate machine learning by reducing storage requirements and increasing the calculation speed of ML algorithms. Tachyum provided support for BF16 for use with GCC 13.2 (GNU Compiler Collection) and tested the software integration of BF16. Tachyum supports the same floating-point BF16 operations as IEEE FP32 and FP64 in hardware.

Tachyum’s Prodigy was designed to handle matrix and vector processing from the ground up rather than as an afterthought. Among Prodigy’s vector and matrix features are support for a range of data types (FP64, FP32, TF32, BF16, Int8, FP8, FP4 and TAI); 2x1024-bit vector units per core; AI sparsity and super-sparsity support; and no penalty for misaligned vector loads or stores when crossing cache lines. This built-in support offers high performance for AI training and inference workloads, increases performance and reduces memory utilization.

“Two months ago, Tachyum successfully integrated the BF16 datatype into Prodigy’s GCC compiler and software distribution,” said Dr. Radoslav Danilak, founder and CEO of Tachyum. “Since there is no standard application test, we developed an AI inference BF16 test application that has since been vetted for use on the Prodigy FPGA. The BF16 test application on FPGA shows that it performs optimally as part of Tachyum AI’s tensor matrix operations.”

The BF16 matrix tensor operations will be showcased in Prodigy FPGA HW prototype next month, followed by FP8, FP4 and low-precision data types.

As a Universal Processor offering industry-leading performance for all workloads, Prodigy-powered data center servers can seamlessly and dynamically switch between computational domains (such as AI/ML, HPC, and cloud) with a single homogeneous architecture. By eliminating the need for expensive dedicated AI hardware and dramatically increasing server utilization, Prodigy reduces CAPEX and OPEX significantly while delivering unprecedented data center performance, power, and economics. Prodigy integrates 192 high-performance custom-designed 64-bit compute cores, to deliver up to 4.5x the performance of the highest-performing x86 processors for cloud workloads, up to 3x that of the highest performing GPU for HPC, and 6x for AI applications.

A video demonstration of the BF16 test application on Prodigy FPGA is available to view below.

Follow Tachyum

About Tachyum

Tachyum is transforming the economics of AI, HPC, public and private cloud workloads with Prodigy, the world’s first Universal Processor. Prodigy unifies the functionality of a CPU, a GPU, and a TPU in a single processor to deliver industry-leading performance, cost and power efficiency for both specialty and general-purpose computing. As global data center emissions continue to contribute to a changing climate, with projections of their consuming 10 percent of the world’s electricity by 2030, the ultra-low power Prodigy is positioned to help balance the world’s appetite for computing at a lower environmental cost. Tachyum received a major purchase order from a US company to build a large-scale system that can deliver more than 50 exaflops performance, which will exponentially exceed the computational capabilities of the fastest inference or generative AI supercomputers available anywhere in the world today. When complete in 2026, the Prodigy-powered system will deliver a 25x multiplier vs. the world’s fastest conventional supercomputer – built just this year – and will achieve AI capabilities 25,000x larger than models for ChatGPT4. Tachyum has offices in the United States and Slovakia. For more information, visit https://www.tachyum.com/.

Tachyum Targets 16KB QLC NAND Flash Page and Indirection Unit

Tue, 09 Jul 2024 00:00:00 +0000

LAS VEGAS, July 9, 2024 – Tachyum® today announced that the Prodigy® software distribution package and pre-configured applications will be optimized for 64KB pages for Linux and 16KB Indirection Unit (IU) for solid state drives in advance of QLC NAND flash eventual replacement for HDDs in the data center. This achievement ensures that Prodigy-enabled systems will support future QLC NAND flash, networking requirements and make 16KB storage networking stack much more efficient for 1.6T Ethernet in servers.

Tachyum storage technology supports CEPH, an open-source software storage platform that is being optimized for 64KB pages and 16KB IU SSDs. With future networking expected to enable jumbo frames (9,000-9,216 bytes) and potentially super jumbo frames (17KB), the company’s optimization satisfies the reliability and high-performance needs while reducing costs.

Real-world SSD IU testing shows that these workloads allow more efficient data granularity and enable very large SSD capacities. While SSD is only twice as expensive as HDD today, QLC reduces costs further but has lower endurance. Moving from 4KB Low-Density Parity Check (LDPC) error correction to 16KB improves QLC reliability considerably.

Flash pages have already moved to 16KB. The 16KB IU reduces DRAM need for high performance SSD by a factor of 4. The 32TB SSD would need 32GB of DRAM, which is large and expensive with 4KB IU but only 8GB on a single 32GB DDR5 dual-die package of 2 DRAM chips. Kioxia and WD just unveiled a 2 terabit (Tb) QLC die delivering 4 TB in a 11.5mm x 13.5mm x 1.5mm package. With next year’s 420-plus layer NAND die, Samsung will also likely be moving to a 2Tb die, and nothing prevents Micron from introducing a 2Tb die within the year, followed by SK Hynix and YMTC. The increased density will drive 16TB E1.S SSDs to become a volume product by next year.

The 2Tb QLC flash will double storage system capacity at the same power and in the same footprint. The 2Tb die will reduce the cost per GB of flash storage systems dramatically, since the unchanged cost of storage systems excluding flash will be amortized against the doubled flash storage capacity, delivering 2x higher capacity at the same cost. In addition, as we continue our analysis of industry trends, we expect 4Tb QLC die to be feasible next year, making today’s 4KB IU unsustainable and forcing the transition to 16KB IU, with Prodigy fully prepared for the transition.

“Tachyum has core SandForce and Skyera team members who are experts in NAND flash storage,” said Dr. Radoslav Danilak, founder and CEO of Tachyum. “The Tachyum engineering team previously delivered SandForce flash controllers, as well as storage appliances at Skyera with compression, so extending this technology to data center level is simply evolutionary for the team. This ensures Prodigy will be able to handle the toughest workloads not only today but into the future as well.”

As a Universal Processor offering industry-leading performance for all workloads, Prodigy-powered data center servers can seamlessly and dynamically switch between computational domains (such as AI/ML, HPC, and cloud) with a single homogeneous architecture. By eliminating the need for expensive dedicated AI hardware and dramatically increasing server utilization, Prodigy reduces CAPEX and OPEX significantly while delivering unprecedented data center performance, power, and economics. Prodigy integrates 192 high-performance custom-designed 64-bit compute cores, to deliver up to 4.5x the performance of the highest-performing x86 processors for cloud workloads, up to 3x that of the highest performing GPU for HPC, and 6x for AI applications.

Follow Tachyum

About Tachyum

Tachyum is transforming the economics of AI, HPC, public and private cloud workloads with Prodigy, the world’s first Universal Processor. Prodigy unifies the functionality of a CPU, a GPU, and a TPU in a single processor to deliver industry-leading performance, cost and power efficiency for both specialty and general-purpose computing. As global data center emissions continue to contribute to a changing climate, with projections of their consuming 10 percent of the world’s electricity by 2030, the ultra-low power Prodigy is positioned to help balance the world’s appetite for computing at a lower environmental cost. Tachyum received a major purchase order from a US company to build a large-scale system that can deliver more than 50 exaflops performance, which will exponentially exceed the computational capabilities of the fastest inference or generative AI supercomputers available anywhere in the world today. When complete in 2026, the Prodigy-powered system will deliver a 25x multiplier vs. the world’s fastest conventional supercomputer – built just this year – and will achieve AI capabilities 25,000x larger than models for ChatGPT4. Tachyum has offices in the United States and Slovakia. For more information, visit https://www.tachyum.com/.

Tachyum Tests Hardware For Donation to OSS Community Before Upstreaming

Tue, 18 Jun 2024 00:00:00 +0000

LAS VEGAS, June 18, 2024 – Tachyum® today announced that it will begin upstreaming code to the open-source community to encourage expanded evaluation opportunities with the Prodigy platform. As part of the testing process, Tachyum will donate computing resources to the OSS community for expanded evaluation opportunities with the Prodigy® platform.

In preparing to upstream to open-source software (OSS), Tachyum has identified and solved many existing issues to improve public utilization and allow collaboration across the community.

“It’s only when you are deep in the process that you learn what you didn’t know before, and that’s in large part why we planned to upstream,” said Dr. Radoslav Danilak, founder and CEO of Tachyum. “Cleaning up temporary and development code has already given us a higher-quality end result, and we look forward to further benefits of being active participants in the open-source community.”

Tachyum is currently extending beta access, and when completed, upstreaming will be performed. Once upstreamed, machines will be available first for close OSS partners of Tachyum.

As a Universal Processor offering industry-leading performance for all workloads, Prodigy-powered data center servers can seamlessly and dynamically switch between computational domains (such as AI/ML, HPC, and cloud) with a single homogeneous architecture. By eliminating the need for expensive dedicated AI hardware and dramatically increasing server utilization, Prodigy reduces CAPEX and OPEX significantly while delivering unprecedented data center performance, power, and economics. Prodigy integrates 192 high-performance custom-designed 64-bit compute cores, to deliver up to 4.5x the performance of the highest-performing x86 processors for cloud workloads, up to 3x that of the highest performing GPU for HPC, and 6x for AI applications.

Follow Tachyum

About Tachyum

Tachyum is transforming the economics of AI, HPC, public and private cloud workloads with Prodigy, the world’s first Universal Processor. Prodigy unifies the functionality of a CPU, a GPU, and a TPU in a single processor to deliver industry-leading performance, cost and power efficiency for both specialty and general-purpose computing. As global data center emissions continue to contribute to a changing climate, with projections of their consuming 10 percent of the world’s electricity by 2030, the ultra-low power Prodigy is positioned to help balance the world’s appetite for computing at a lower environmental cost. Tachyum received a major purchase order from a US company to build a large-scale system that can deliver more than 50 exaflops performance, which will exponentially exceed the computational capabilities of the fastest inference or generative AI supercomputers available anywhere in the world today. When complete in 2026, the Prodigy-powered system will deliver a 25x multiplier vs. the world’s fastest conventional supercomputer – built just this year – and will achieve AI capabilities 25,000x larger than models for ChatGPT4. Tachyum has offices in the United States and Slovakia. For more information, visit https://www.tachyum.com/.

Tachyum Demonstrates Instruction Profiling Unit on Prodigy FPGA

Tue, 11 Jun 2024 00:00:00 +0000

LAS VEGAS, June 11, 2024 – Tachyum® today announced that it enabled an Instruction Profiling Unit (IPU), a low overhead way to collect the profile of non-instrumented executed code, on its Prodigy® Universal Processor. IPU is used by hyperscalers to profile applications in production execution and recompile code to gain better performance.

This latest enhancement is part of the company’s focus on refining and performance optimization of the Tachyum Software Distribution Package upon its beta release. Using results collected by IPU in re-compiling applications can provide a 5-15% gain in performance depending on application, which would result in a huge financial benefit to users.

In February, Tachyum added a Performance Monitoring Unit (PMU) to the emulation system that empowers customers and partners with the ability to address bottlenecks and better optimize Prodigy performance for all applications and workloads. The PMU’s wide range of performance counters – supported by both software C-model and FPGA – facilitates both system debugging and performance tuning. The addition of IPU allows it to be used by Profile Directed Optimizations (PDO) and is important for Just In Time (JIT) compilers for optimizing hotspots.

“IPU is essential for large-scale operators and is now readily available as part of our FPGA,” said Dr. Radoslav Danilak, founder and CEO of Tachyum. “We believe this technology will also be used by smaller data center operators. This is important for meeting our goals of Prodigy supplying industry-leading performance at significantly lower power and lower cost to organizations of all sizes looking to supercharge their workloads while supporting the greatest breadth of applications.”

As a Universal Processor offering industry-leading performance for all workloads, Prodigy-powered data center servers can seamlessly and dynamically switch between computational domains (such as AI/ML, HPC, and cloud) with a single homogeneous architecture. By eliminating the need for expensive dedicated AI hardware and dramatically increasing server utilization, Prodigy reduces CAPEX and OPEX significantly while delivering unprecedented data center performance, power, and economics. Prodigy integrates 192 high-performance custom-designed 64-bit compute cores, to deliver up to 4.5x the performance of the highest-performing x86 processors for cloud workloads, up to 3x that of the highest performing GPU for HPC, and 6x for AI applications.

A video demonstrating IPU running on a Prodigy FPGA prototype is available for viewing below.

Follow Tachyum

About Tachyum

Tachyum is transforming the economics of AI, HPC, public and private cloud workloads with Prodigy, the world’s first Universal Processor. Prodigy unifies the functionality of a CPU, a GPU, and a TPU in a single processor architecture to deliver industry-leading performance, cost and power efficiency for both specialty and general-purpose computing. As global data center emissions continue to contribute to a changing climate, with projections that they will consume 10 percent of the world’s electricity by 2030, the ultra-low power Prodigy is positioned to help balance the world’s appetite for computing at a lower environmental cost. Tachyum has received a major purchase order from a US company to build a large-scale system that can deliver more than 50 exaflops performance, which will exponentially exceed the computational capabilities of the fastest inference or generative AI supercomputers available anywhere in the world today. When complete in 2026, the Prodigy-powered system will deliver a 25x multiplier vs. the world’s fastest conventional supercomputer – built just this year – and will achieve AI capabilities 25,000x larger than models for ChatGPT4. Tachyum has offices in the United States and Slovakia. For more information, visit https://www.tachyum.com/.

Tachyum Demonstrates Linux and Applications with 64KB Page Size

Tue, 04 Jun 2024 00:00:00 +0000

LAS VEGAS, June 4, 2024 –Tachyum® today announced that it has moved to 64KB page size as the default of its Prodigy® software distribution, a completely integrated software stack and package, which is now available to early adopters and customers as a pre-installed image as part of the beta phase of testing.

Tachyum completed testing with 4KB pages and 4KB support is now viewed as legacy. The company will instead focus on 64KB pages with 4KB pages only serving as a backup solution for those who have a special need. OEMs that require 4KB support will still be able to receive it from Tachyum based on their respective agreements.

Application testing with the 64KB page size has begun. Once stable, Tachyum will expand testing to the Prodigy FPGA prototype. Debugging applications on FPGA at this stage will better ensure that hardware will not be mistakenly suspected of any software issues that may arise.

“We are taking every opportunity to maximize the capabilities of Prodigy prior to its launch,” said Dr. Radoslav Danilak, founder and CEO of Tachyum. “By increasing the page size to 64KB as the default, we are better able to optimize performance of our software distribution while ensuring that applications work as designed out of the box. This is the latest small step that Tachyum has taken to ensure data centers will be able to advance by leaps and bounds after implementing Prodigy.”

As a Universal Processor offering industry-leading performance for all workloads, Prodigy-powered data center servers can seamlessly and dynamically switch between computational domains (such as AI/ML, HPC, and cloud) with a single homogeneous architecture. By eliminating the need for expensive dedicated AI hardware and dramatically increasing server utilization, Prodigy reduces CAPEX and OPEX significantly while delivering unprecedented data center performance, power, and economics. Prodigy integrates 192 high-performance custom-designed 64-bit compute cores, to deliver up to 4.5x the performance of the highest-performing x86 processors for cloud workloads, up to 3x that of the highest performing GPU for HPC, and 6x for AI applications.

A video demonstrating booting and running applications at 4KB and 64KB page size with the identification of the numbers of page faults decreasing is available for viewing below.

Follow Tachyum

About Tachyum

Tachyum is transforming the economics of AI, HPC, public and private cloud workloads with Prodigy, the world’s first Universal Processor. Prodigy unifies the functionality of a CPU, a GPU, and a TPU in a single processor to deliver industry-leading performance, cost and power efficiency for both specialty and general-purpose computing. As global data center emissions continue to contribute to a changing climate, with projections of their consuming 10 percent of the world’s electricity by 2030, the ultra-low power Prodigy is positioned to help balance the world’s appetite for computing at a lower environmental cost. Tachyum recently received a major purchase order from a US company to build a large-scale system that can deliver more than 50 exaflops performance, which will exponentially exceed the computational capabilities of the fastest inference or generative AI supercomputers available anywhere in the world today. When complete in 2025, the Prodigy-powered system will deliver a 25x multiplier vs. the world’s fastest conventional supercomputer – built just this year – and will achieve AI capabilities 25,000x larger than models for ChatGPT4. Tachyum has offices in the United States and Slovakia. For more information, visit https://www.tachyum.com/.

Tachyum Demonstrates Full BF16 AI Support in GCC and PyTorch

Tue, 28 May 2024 00:00:00 +0000

LAS VEGAS, May 28, 2024 – Tachyum® today announced that it has successfully integrated the BF16 data type into its Prodigy® compiler and software distribution, which is now available to early adopters and customers as a pre-installed image as part of beta testing.

BF16, or bfloat16, is a shortened floating point data type based on the IEEE 32-bit single-precision floating point data type (f32). It is used to accelerate machine learning by reducing storage requirements and increasing the calculation speed of ML algorithms. Tachyum now fully supports BF16 for use with GCC 13.2 (GNU Compiler Collection); HPC/linear algebra Eigen library optimized for Prodigy Universal Processor; and PyTorch AI framework.

“We continue to strengthen our software distribution package to ensure the greatest breadth of application, framework and library support for Prodigy in advance of its release,” said Dr. Radoslav Danilak, founder and CEO of Tachyum. “The use of BF16 improves hardware efficiency by improving performance. Our support of the format is consistent with our goals of having Prodigy provide the performance required of hyperscale, high-performance computing and AI workloads without modifications and affirms our commitment to transforming data centers around the world.”

As a Universal Processor offering industry-leading performance for all workloads, Prodigy-powered data center servers can seamlessly and dynamically switch between computational domains (such as AI/ML, HPC, and cloud) with a single homogeneous architecture. By eliminating the need for expensive dedicated AI hardware and dramatically increasing server utilization, Prodigy reduces CAPEX and OPEX significantly while delivering unprecedented data center performance, power, and economics. Prodigy integrates 192 high-performance custom-designed 64-bit compute cores, to deliver up to 4.5x the performance of the highest-performing x86 processors for cloud workloads, up to 3x that of the highest performing GPU for HPC, and 6x for AI applications.

A video demonstration of image classification using ResNet model utilizing native PyTorch implementation on Tachyum Linux on a Prodigy emulation system is available for viewing below. The demonstrated ResNet model has been quantized using the BF16 data type to take advantage of Prodigy BF16 vector instruction, particularly activation, loss and reduction functions. The next video will demonstrate the completion of FP8 testing.

Follow Tachyum

About Tachyum

Tachyum is transforming the economics of AI, HPC, public and private cloud workloads with Prodigy, the world’s first Universal Processor. Prodigy unifies the functionality of a CPU, a GPU, and a TPU in a single processor to deliver industry-leading performance, cost and power efficiency for both specialty and general-purpose computing. As global data center emissions continue to contribute to a changing climate, with projections of their consuming 10 percent of the world’s electricity by 2030, the ultra-low power Prodigy is positioned to help balance the world’s appetite for computing at a lower environmental cost. Tachyum recently received a major purchase order from a US company to build a large-scale system that can deliver more than 50 exaflops performance, which will exponentially exceed the computational capabilities of the fastest inference or generative AI supercomputers available anywhere in the world today. When complete in 2025, the Prodigy-powered system will deliver a 25x multiplier vs. the world’s fastest conventional supercomputer – built just this year – and will achieve AI capabilities 25,000x larger than models for ChatGPT4. Tachyum has offices in the United States and Slovakia. For more information, visit https://www.tachyum.com/.

Tachyum Software Distribution Package Enters Beta Testing

Tue, 21 May 2024 00:00:00 +0000

LAS VEGAS, May 21, 2024 –Tachyum® today announced that its Prodigy® software distribution, a completely integrated software stack and package, is now available to early adopters and customers as a pre-installed image as part of the beta phase of testing.

Applications included in the beta version of the distribution package have been tested to work straight out of the box so that customers and partners can immediately begin using the Prodigy reference design. Additionally, users can effortlessly restore to the original image in case any issues arise during the beta testing phase.

As part of the beta release, Tachyum upgraded all software and tested 57b addressing with applications on the Prodigy FPGA emulation system. The company also began performance optimization, moving to 64KB page size as the default of its software distribution. The 4KB pages that Intel copied from IBM 370 in 1971 are outdated with 4KB support now viewed as legacy. Because optimization towards 16KB storage block size is where modern high-capacity, cost-efficient SSD storage is headed, Tachyum will focus on SSD Indirection Unit (IU) 16KB with 4KB pages only serving as a backup solution for those who have a special need.

Additional advancements to the beta release include:

GCC was updated to the most recent version that supports BFloat16 data type.
4KB 48b VA was tested to indicate whether issues were more likely in hardware vs. software but now with cleanly running 57b VA on Prodigy FPGA emulation system, Tachyum switched to 57b VA and fixed bugs in software.
Tachyum is now switching to 64KB pages 55b VA in the QEMU software emulator as application software in these modes is less frequently tested. After testing in QEMU, Tachyum will test it on the FPGA prototype of its Prodigy.
Tachyum will achieve the same testing coverage on 64KB pages as with 4KB pages with 48b VA.

This latest progression in the Prodigy software distribution closely follows a series of progressive achievements as part of Tachyum’s advancement to production-ready status. These achievements include adding support for performance optimization using counters and instruction profiling, hardware debugging capabilities and a watchdog timer to help detect and recovery from device or system malfunctions.

Tachyum previously advanced the Unified Extensible Firmware Interface (UEFI) to beta and has added RAID 1 capabilities and UEFI manual; the FreeBSD operating system can now be compiled using LLVM; and Linux kernel in the beta release will have mTHP support.

“Hardware and software must be able to work in unison to produce the greatest results possible,” said Dr. Radoslav Danilak, founder and CEO of Tachyum. “With the successes we have had in advancing the hardware aspect of the Prodigy Universal Processor Chip, moving our software distribution package to beta is a tremendous step in achieving our goals of revolutionizing the data center. We are thankful to those partners and customers that have helped us achieve beta status and are excited to see what they can do with this latest, upgraded version of our software.”

As a Universal Processor offering industry-leading performance for all workloads, Prodigy-powered data center servers can seamlessly and dynamically switch between computational domains (such as AI/ML, HPC, and cloud) with a single homogeneous architecture. By eliminating the need for expensive dedicated AI hardware and dramatically increasing server utilization, Prodigy reduces CAPEX and OPEX significantly while delivering unprecedented data center performance, power, and economics. Prodigy integrates 192 high-performance custom-designed 64-bit compute cores, to deliver up to 4.5x the performance of the highest-performing x86 processors for cloud workloads, up to 3x that of the highest performing GPU for HPC, and 6x for AI applications.

Follow Tachyum

About Tachyum

Tachyum is transforming the economics of AI, HPC, public and private cloud workloads with Prodigy, the world’s first Universal Processor. Prodigy unifies the functionality of a CPU, a GPU, and a TPU in a single processor to deliver industry-leading performance, cost and power efficiency for both specialty and general-purpose computing. As global data center emissions continue to contribute to a changing climate, with projections of their consuming 10 percent of the world’s electricity by 2030, the ultra-low power Prodigy is positioned to help balance the world’s appetite for computing at a lower environmental cost. Tachyum recently received a major purchase order from a US company to build a large-scale system that can deliver more than 50 exaflops performance, which will exponentially exceed the computational capabilities of the fastest inference or generative AI supercomputers available anywhere in the world today. When complete in 2025, the Prodigy-powered system will deliver a 25x multiplier vs. the world’s fastest conventional supercomputer – built just this year – and will achieve AI capabilities 25,000x larger than models for ChatGPT4. Tachyum has offices in the United States and Slovakia. For more information, visit https://www.tachyum.com/.

Tachyum Moves UEFI to Beta, Adds RAID 1 Capabilities to Prodigy

Wed, 15 May 2024 00:00:00 +0000

LAS VEGAS, May 15, 2024 –Tachyum® today announced that it has advanced the Unified Extensible Firmware Interface (UEFI) to beta and has added RAID 1 capabilities. These latest improvements are a further indication that Tachyum’s Prodigy® hardware continues to mature through its development stages prior to its public release.

UEFI replaces the traditional BIOS as a next-generation interface between the operating system and platform firmware. It is used during startup to initialize the hardware and load the OS. Additionally, UEFI determines device boot priority and allows users to customize hardware and software settings. With this latest release, menus and submenus, screens and options received upgrades for better functionality and an improved user experience.

In conjunction with the advancement to beta, Tachyum released a user manual that provides details about how to access and use the UEFI, which is embedded in the system flash on all Prodigy platforms. The manual explains each of the UEFI menus, how to configure UEFI parameters and how to save new configurations. The documents are intended for users who install, maintain, troubleshoot and administer Prodigy platforms.

RAID’s ability to store data across multiple hard drives (HDDs) or solid-state drives (SSDs) allows for the protection of data in case of a drive failure by creating redundancy and improving performance. Tachyum’s approach to this virtual disk technology provides:

• RAID 1 boot for preinstalled Linux partition with preinstalled software

• RAID 1 read-only partition for installation packages and to restore images to the manufacturing original state if the customer runs into problems

• Partition where customers can install and configure their own Linux and apps configured by them

• RAID 0 for data

“We continue to see impressive advancements in the capabilities and functionality of Prodigy at every level of development,” said Dr. Radoslav Danilak, founder and CEO of Tachyum. “Moving UEFI from alpha to beta allows a better look and polishing of the UEFI. RAID 1 was added to provide a RAID boot without the need for expensive NVMe SSD RAID cards, helping to bring Prodigy-enabled systems to datacenter-level class. All of these moves, including the expectation that we will have a release candidate available next quarter, further signifies that we are nearing production of the Prodigy Universal Processor.”

As a Universal Processor offering industry-leading performance for all workloads, Prodigy-powered data center servers can seamlessly and dynamically switch between computational domains (such as AI/ML, HPC, and cloud) with a single homogeneous architecture. By eliminating the need for expensive dedicated AI hardware and dramatically increasing server utilization, Prodigy reduces CAPEX and OPEX significantly while delivering unprecedented data center performance, power, and economics. Prodigy integrates 192 high-performance custom-designed 64-bit compute cores, to deliver up to 4.5x the performance of the highest-performing x86 processors for cloud workloads, up to 3x that of the highest performing GPU for HPC, and 6x for AI applications.

A video demonstration of RAID 1 UEFI boot on a Prodigy emulation system is available for viewing below.

Follow Tachyum

About Tachyum

Tachyum is transforming the economics of AI, HPC, public and private cloud workloads with Prodigy, the world’s first Universal Processor. Prodigy unifies the functionality of a CPU, a GPU, and a TPU in a single processor to deliver industry-leading performance, cost and power efficiency for both specialty and general-purpose computing. As global data center emissions continue to contribute to a changing climate, with projections of their consuming 10 percent of the world’s electricity by 2030, the ultra-low power Prodigy is positioned to help balance the world’s appetite for computing at a lower environmental cost. Tachyum recently received a major purchase order from a US company to build a large-scale system that can deliver more than 50 exaflops performance, which will exponentially exceed the computational capabilities of the fastest inference or generative AI supercomputers available anywhere in the world today. When complete in 2025, the Prodigy-powered system will deliver a 25x multiplier vs. the world’s fastest conventional supercomputer – built just this year – and will achieve AI capabilities 25,000x larger than models for ChatGPT4. Tachyum has offices in the United States and Slovakia. For more information, visit https://www.tachyum.com/.

Tachyum Adds Watchdog Timers to Prodigy FPGA

Tue, 07 May 2024 00:00:00 +0000

LAS VEGAS, May 7, 2024 –Tachyum® today announced that it has added a watchdog timer to its Prodigy® FPGA emulation system to help detect and recovery from device or system malfunction. The mechanism facilitates automatic correction of temporary hardware faults and prevents errant or malevolent software from disrupting system operations.

Available as both a hardware and virtualized time, Prodigy’s watchdog is part of a self-healing infrastructure within datacenters. It allows the automatic recovery of servers that hang without human intervention. This functionality is important for organizations with hundreds to thousands of servers.

Prodigy’s hardware watchdog timer must be cleared by the operating system based on a programmable timeout value or it will trigger a non-maskable interrupt. The watchdog performs corrective actions by initiating a soft reboot. If unsuccessful, it will attempt to boot the OS on a single core, save the kernel dump and then perform a hard reboot.

For a virtualized environment, the watchdogs are virtualized themselves so that each VM maintains an individual timer.

“At the most basic level, a watchdog is a mechanism that resets a computer if something goes wrong,” said Dr. Radoslav Danilak, founder and CEO of Tachyum. “This can encompass detecting any number of issues within the system, making a watchdog that will not only identify the problem but automatically initiate corrective issues is a valuable feature. As we continue to progress towards Prodigy’s commercial release, it makes sense that our FPGA performs with watchdogs in place to ensure success of systems that incorporate our Universal Processor chip.”

As a Universal Processor offering industry-leading performance for all workloads, Prodigy-powered data center servers can seamlessly and dynamically switch between computational domains (such as AI/ML, HPC, and cloud) with a single homogeneous architecture. By eliminating the need for expensive dedicated AI hardware and dramatically increasing server utilization, Prodigy reduces CAPEX and OPEX significantly while delivering unprecedented data center performance, power, and economics. Prodigy integrates 192 high-performance custom-designed 64-bit compute cores, to deliver up to 4.5x the performance of the highest-performing x86 processors for cloud workloads, up to 3x that of the highest performing GPU for HPC, and 6x for AI applications.

A video demonstration of a watchdog timer working on a Prodigy FPGA emulation system is available for viewing below.

Follow Tachyum

About Tachyum

Tachyum is transforming the economics of AI, HPC, public and private cloud workloads with Prodigy, the world’s first Universal Processor. Prodigy unifies the functionality of a CPU, a GPU, and a TPU in a single processor to deliver industry-leading performance, cost and power efficiency for both specialty and general-purpose computing. As global data center emissions continue to contribute to a changing climate, with projections of their consuming 10 percent of the world’s electricity by 2030, the ultra-low power Prodigy is positioned to help balance the world’s appetite for computing at a lower environmental cost. Tachyum recently received a major purchase order from a US company to build a large-scale system that can deliver more than 50 exaflops performance, which will exponentially exceed the computational capabilities of the fastest inference or generative AI supercomputers available anywhere in the world today. When complete in 2025, the Prodigy-powered system will deliver a 25x multiplier vs. the world’s fastest conventional supercomputer – built just this year – and will achieve AI capabilities 25,000x larger than models for ChatGPT4. Tachyum has offices in the United States and Slovakia. For more information, visit https://www.tachyum.com/.

Tachyum Is Testing Open Sourcing of Prodigy Software Through Git

Wed, 24 Apr 2024 00:00:00 +0000

LAS VEGAS, April 24, 2024 –Tachyum® today announced that it is testing Git access to Tachyum software repositories. This latest advancement is part of the company’s continued commitment to releasing all open-source software in order to provide customers and partners with an easy, straightforward platform for early Prodigy® testing ahead of silicon availability while reducing support requirements.

Git is an open-source distributed version control system designed to track modifications in source code during software development. Providing Git access helps enable quick, easy, out-of-the-box testing and evaluation for customers and partners using their existing ISA prior to running their applications natively on Prodigy.

Tachyum’s complete software distribution package, which is currently in its alpha stage, continues to receive significant updates. The Tachyum Software Distribution package is scheduled to enter the beta stage in the coming months, which will be available to early adopters and testers. This process ensures that Tachyum’s ported software has been tested in real-world scenarios and ensures that the ported software is capable of handling actual customer use cases.

Until Tachyum OSS is upstream, the company will maintain Git repositories for public access, initially for its close partners. Currently, only the most recent version will be available. After production release, Tachyum will maintain multiple software versions. As part of the testing process, Tachyum will donate hardware to the OSS community for expanded evaluation opportunities.

“This latest news re-affirms Tachyum’s commitment to open-source software development and maintenance,” said Dr. Radoslav Danilak, founder and CEO of Tachyum. “As Tachyum prepares for tape-out, we will continue to add and test a variety of supporting infrastructure, like Git access, ahead of time so that we are able to unleash the full potential of the Tachyum Prodigy Universal Processor from day one with a rich ecosystem of applications, system software and frameworks and libraries that are ported to run natively on Prodigy.”

As a Universal Processor offering industry-leading performance for all workloads, Prodigy-powered data center servers can seamlessly and dynamically switch between computational domains (such as AI/ML, HPC, and cloud) with a single homogeneous architecture. By eliminating the need for expensive dedicated AI hardware and dramatically increasing server utilization, Prodigy reduces CAPEX and OPEX significantly while delivering unprecedented data center performance, power, and economics. Prodigy integrates 192 high-performance custom-designed 64-bit compute cores, to deliver up to 4.5x the performance of the highest-performing x86 processors for cloud workloads, up to 3x that of the highest performing GPU for HPC, and 6x for AI applications.

A video demonstration of using Git in the development of the Prodigy native software is available for viewing below.

Follow Tachyum

About Tachyum

Tachyum is transforming the economics of AI, HPC, public and private cloud workloads with Prodigy, the world’s first Universal Processor. Prodigy unifies the functionality of a CPU, a GPU, and a TPU in a single processor to deliver industry-leading performance, cost and power efficiency for both specialty and general-purpose computing. As global data center emissions continue to contribute to a changing climate, with projections of their consuming 10 percent of the world’s electricity by 2030, the ultra-low power Prodigy is positioned to help balance the world’s appetite for computing at a lower environmental cost. Tachyum recently received a major purchase order from a US company to build a large-scale system that can deliver more than 50 exaflops performance, which will exponentially exceed the computational capabilities of the fastest inference or generative AI supercomputers available anywhere in the world today. When complete in 2025, the Prodigy-powered system will deliver a 25x multiplier vs. the world’s fastest conventional supercomputer – built just this year – and will achieve AI capabilities 25,000x larger than models for ChatGPT4. Tachyum has offices in the United States and Slovakia. For more information, visit https://www.tachyum.com/.

Tachyum Completes Testing of Debugger on its Prodigy FPGA Prototype

Tue, 09 Apr 2024 00:00:00 +0000

LAS VEGAS, April 9, 2024 – Tachyum® today announced that it has completed testing of full-fledged hardware debugging features of the FPGA prototype of its Prodigy® Universal Processor. This latest capability has undergone testing with GNU Debugger (gdb), a widely used Linux tool for debugging software, to ensure successful operations and is a key component of Prodigy’s continued march towards production.

As part of the hardware development process, debuggers are used to search for and identify components that are not operating correctly or are incorrectly configured. The Prodigy platform supports four debug registers for hardware breakpoints and four registers for hardware watchpoints for memory operations. Four hardware PC breakpoints can even debug ROM code where software breakpoints are unable to be used.

“Having hardware debugging features incorporated into the Prodigy FPGA allows our customers to debug directly on the prototype itself instead of through less effective software processes,” said Dr. Radoslav Danilak, founder and CEO of Tachyum. “Previous debugging, which ran on Prodigy software c-model, is now running on FPGA and ensures that hardware issues can be resolved prior to tape-out. This is the latest feature we’ve incorporated into the Prodigy FPGA tests to ensure our customers have all the necessary tools and capabilities implemented into our Universal Processor on Day 1 of launch.”

As a Universal Processor offering industry-leading performance for all workloads, Prodigy-powered data center servers can seamlessly and dynamically switch between computational domains (such as AI/ML, HPC, and cloud) with a single homogeneous architecture. By eliminating the need for expensive dedicated AI hardware and dramatically increasing server utilization, Prodigy reduces CAPEX and OPEX significantly while delivering unprecedented data center performance, power, and economics. Prodigy integrates 192 high-performance custom-designed 64-bit compute cores, to deliver up to 4.5x the performance of the highest-performing x86 processors for cloud workloads, up to 3x that of the highest performing GPU for HPC, and 6x for AI applications.

A video demonstration of Prodigy’s FPGA hardware debugger using hardware debug registers showcased with gdb is available for viewing below.

Follow Tachyum

About Tachyum

Tachyum is transforming the economics of AI, HPC, public and private cloud workloads with Prodigy, the world’s first Universal Processor. Prodigy unifies the functionality of a CPU, a GPU, and a TPU in a single processor to deliver industry-leading performance, cost and power efficiency for both specialty and general-purpose computing. As global data center emissions continue to contribute to a changing climate, with projections of their consuming 10 percent of the world’s electricity by 2030, the ultra-low power Prodigy is positioned to help balance the world’s appetite for computing at a lower environmental cost. Tachyum recently received a major purchase order from a US company to build a large-scale system that can deliver more than 50 exaflops performance, which will exponentially exceed the computational capabilities of the fastest inference or generative AI supercomputers available anywhere in the world today. When complete in 2025, the Prodigy-powered system will deliver a 25x multiplier vs. the world’s fastest conventional supercomputer – built just this year – and will achieve AI capabilities 25,000x larger than models for ChatGPT4. Tachyum has offices in the United States and Slovakia. For more information, visit https://www.tachyum.com/.

Tachyum Solidifies Reliability, Availability and Serviceability of Prodigy Universal Processor

Tue, 26 Mar 2024 00:00:00 +0000

LAS VEGAS, March 26, 2024 – Tachyum® today added to its extensive white paper library with the publication of an overview examining its Reliability, Availability and Serviceability (RAS) strategy, including a detailed look at the key RAS features being built into Prodigy®, the world’s first Universal Processor, which will help satisfy the demands of today’s data centers.

RAS is a set of related attributes that must be considered when designing, manufacturing, purchasing and utilizing a computer product or component. Designed from the ground up, Tachyum’s comprehensive RAS strategy encompasses multiple facets at the silicon, platform and system levels to ensure Prodigy deployments provide high performance along with high reliability and availability at all levels.

Prodigy’s RAS strategy is comprised of Device RAS, which includes advanced error detection and correction in all functional blocks; System RAS, which includes critical features such as machine check and recovery working with the Linux EDAC driver; and Platform RAS, encompassing features such as redundant power supplies and ease of serviceability.

Prodigy’s memory hierarchy provides robust error detection and correction for all memory subsystems. Both the L1 I-Cache and D-Cache are protected with SECDED (single error correction double error detection), and the L2/L3 block utilizes DECTED (double error correction triple error detection), exceeding Arm’s current parity offerings. In addition to Prodigy’s memory hierarchy, other functional blocks integrate significant amounts of memory that require protection to ensure Prodigy runs error-free and maintains high RAS standards.

Additional RAS features built into Prodigy include:

Error correcting codes (ECC) scrubbing and data poisoning
Watchdog timer
RAID for booting
PCIe 5.0 RAS features

Tachyum has incorporated redundant power supply unit (PSU) fans, network interface card (NIC) and efficient maintenance into its Prodigy evaluation platform. When launched, Tachyum will offer data center family SKUs with a 5-year warranty/support period and 10-year warranty/support periods for enterprise/telco family SKUs.

“A comprehensive approach to RAS becomes increasingly important as process shrinks drive higher density for components and platforms,” said Dr. Radoslav Danilak, founder and CEO of Tachyum. “In addition to the ever-increasing density, manufacturing chips on shrinking process nodes increases the risk of soft errors. Prodigy addresses these increased risks with a thorough approach to device and system reliability, ensuring that Prodigy-based systems function with maximum uptime to address the performance and demands of today’s data centers.”

As part of their recent keynote for GTC 24, Nvidia stressed the importance of RAS in their latest product introduction, spending valuable keynote time to include RAS features as part of their new products and features overview, and the importance of RAS was highlighted as they showed a large potential data center deployment that would provide 645 EF of AI performance.

The Tachyum RAS paper complements an earlier Tachyum white paper which showcased a large Prodigy lead customer data center designed to run 8,000 EF of AI performance where RAS will be a critical component.

As a Universal Processor offering industry-leading performance for all workloads, Prodigy-powered data center servers can seamlessly and dynamically switch between computational domains (such as AI/ML, HPC, and cloud) with a single homogeneous architecture. By eliminating the need for expensive dedicated AI hardware and dramatically increasing server utilization, Prodigy reduces CAPEX and OPEX significantly while delivering unprecedented data center performance, power, and economics. Prodigy integrates 192 high-performance custom-designed 64-bit compute cores, to deliver up to 4.5x the performance of the highest-performing x86 processors for cloud workloads, up to 3x that of the highest performing GPU for HPC, and 6x for AI applications.

Those interested in learning more about Tachyum’s RAS strategy and how it extends reliability, availability and serviceability in the data center can download the full white paper.

Follow Tachyum

About Tachyum

Tachyum is transforming the economics of AI, HPC, public and private cloud workloads with Prodigy, the world’s first Universal Processor. Prodigy unifies the functionality of a CPU, a GPU, and a TPU in a single processor to deliver industry-leading performance, cost and power efficiency for both specialty and general-purpose computing. As global data center emissions continue to contribute to a changing climate, with projections of their consuming 10 percent of the world’s electricity by 2030, the ultra-low power Prodigy is positioned to help balance the world’s appetite for computing at a lower environmental cost. Tachyum recently received a major purchase order from a US company to build a large-scale system that can deliver more than 50 exaflops performance, which will exponentially exceed the computational capabilities of the fastest inference or generative AI supercomputers available anywhere in the world today. When complete in 2025, the Prodigy-powered system will deliver a 25x multiplier vs. the world’s fastest conventional supercomputer – built just this year – and will achieve AI capabilities 25,000x larger than models for ChatGPT4. Tachyum has offices in the United States and Slovakia. For more information, visit https://www.tachyum.com/.

Tachyum Demonstrates Machine Check and Recovery on Prodigy FPGA

Tue, 12 Mar 2024 00:00:00 +0000

LAS VEGAS, March 12, 2024 – Tachyum® today announced the addition of Machine Check and Recovery (MCR) capabilities with the Linux Error Detection and Correction (EDAC) subsystem to the Prodigy® Universal Processor with successful deployment demonstrated as part of the FPGA emulation system.

MCR with Linux EDAC driver is essential for data center applications, with the pair working together to provide critical information to predict and mitigate failures in the field. By detecting and seamlessly correcting errors caused by external events in the CPU’s internal memory blocks and attached DDR modules, Prodigy can run prolonged workflows without interruption to maintain and improve uptime of systems deployed at scale. When the degree of Static Random-Access Memory (SRAM) damage is beyond repair, the error detection allows affected computations to be abandoned rather than provide incorrect results.

Error injection is an essential part of testing. Prodigy contains an error-injection module that can inject both correctable and uncorrectable errors into relevant CPU blocks and either a limited number or continuous stream of errors with programmable intervals to ensure the Prodigy architecture meets and exceeds data center requirements. Prodigy provides Double Error Correction and Triple Error Detection (DECTED), which is a key feature to improving uptime, and is complemented by EDAC to enable preventative maintenance.

“Today’s demanding data center applications require a level of reliability and availability previously unseen in order to complete complex functions while mitigating errors,” said Dr. Radoslav Danilak, founder and CEO of Tachyum. “Organizations choosing to deploy Prodigy-enabled datacenters will be able to rest comfortably knowing that we have fortified their system by fully integrating and testing the MCR system with the Linux EDAC driver as part of our FPGA emulator, which will ensure optimal performance when the processor is commercially available in the near future.”

As a Universal Processor offering industry-leading performance for all workloads, Prodigy-powered data center servers can seamlessly and dynamically switch between computational domains (such as AI/ML, HPC, and cloud) with a single homogeneous architecture. By eliminating the need for expensive dedicated AI hardware and dramatically increasing server utilization, Prodigy reduces CAPEX and OPEX significantly while delivering unprecedented data center performance, power, and economics. Prodigy integrates 192 high-performance custom-designed 64-bit compute cores, to deliver up to 4.5x the performance of the highest-performing x86 processors for cloud workloads, up to 3x that of the highest performing GPU for HPC, and 6x for AI applications.

A video demonstrating Prodigy’s ability to correct memory errors caused by its error-injection modules and how the EDAC subsystem reacts to those events is available for viewing below.

Follow Tachyum

About Tachyum

Tachyum is transforming the economics of AI, HPC, public and private cloud workloads with Prodigy, the world’s first Universal Processor. Prodigy unifies the functionality of a CPU, a GPU, and a TPU in a single processor to deliver industry-leading performance, cost and power efficiency for both specialty and general-purpose computing. As global data center emissions continue to contribute to a changing climate, with projections of their consuming 10 percent of the world’s electricity by 2030, the ultra-low power Prodigy is positioned to help balance the world’s appetite for computing at a lower environmental cost. Tachyum recently received a major purchase order from a US company to build a large-scale system that can deliver more than 50 exaflops performance, which will exponentially exceed the computational capabilities of the fastest inference or generative AI supercomputers available anywhere in the world today. When complete in 2025, the Prodigy-powered system will deliver a 25x multiplier vs. the world’s fastest conventional supercomputer – built just this year – and will achieve AI capabilities 25,000x larger than models for ChatGPT4. Tachyum has offices in the United States and Slovakia. For more information, visit https://www.tachyum.com/.

Tachyum Has Disclosed 1U and 2U Server/HPC/AI Reference Designs

Tue, 05 Mar 2024 00:00:00 +0000

LAS VEGAS, March 5, 2024 – Tachyum® today announced that it is bringing 1U and 2U platform solutions to market behind a strategy that ensures customers and partners will be able to quickly and easily test, benchmark and deploy Prodigy® solutions across a broad range of supported applications and workloads.

Tachyum’s platform strategy includes offering evaluation platforms for early testing with OEMs and ODMs able to incorporate Prodigy into their own designs. The 2U evaluation platform, optimized to address the high-performance needs of HPC and Big AI, will be the first to sample in Q1 of 2025. The 1U platform, targeting applications such as AI inference and a wide range of cloud applications, follows in the second quarter.

Tachyum has chosen Chenbro as the chassis partner for the Prodigy evaluation platforms. Chenbro’s standard chassis products provide solutions for both 1U and 2U that address Prodigy’s high-performance requirements, allowing Tachyum to focus on the device and motherboard development. Both platforms will use the same 2-socket motherboard supporting 32 DDR5 DIMMs, as well as supporting storage integration for 16 E1.S SSDs. The evaluation platforms will initially launch as air-cooled infrastructures, allowing for fast, easy evaluations, with a 4-socket liquid-cooled platform arriving later.

Chenbro’s 1U and 2U chassis complement Tachyum’s Prodigy, ensuring seamless integration and unmatched performance. Prodigy promises unparalleled efficiency and scalability, reshaping computing landscapes. “We’re excited to partner with Tachyum,” says Eric Hui, President of Chenbro. “Our chassis solutions, coupled with Prodigy’s innovation, will drive new standards in AI and HPC.”

“Chenbro’s quality aligns with our mission to revolutionize computing. Together, we’re set to redefine computational boundaries,” said Dr. Radoslav Danilak, founder and CEO of Tachyum. “As we achieve success at the system level, it is imperative that we prepare to progress to conquering other elements of the data center. We are not just a chip company but an innovator in revolutionizing the entire data center for those seeking better performance in hyperscale, cloud and AI environments. We are well positioned to continue on this quest through the creation of advanced motherboard designs complete with all the critical components.”

As a Universal Processor offering industry-leading performance for all workloads, Prodigy-powered data center servers can seamlessly and dynamically switch between computational domains (such as AI/ML, HPC, and cloud) with a single homogeneous architecture. By eliminating the need for expensive dedicated AI hardware and dramatically increasing server utilization, Prodigy reduces CAPEX and OPEX significantly while delivering unprecedented data center performance, power, and economics. Prodigy integrates 192 high-performance custom-designed 64-bit compute cores, to deliver up to 4.5x the performance of the highest-performing x86 processors for cloud workloads, up to 3x that of the highest performing GPU for HPC, and 6x for AI applications.

Details of Tachyum’s data center strategy in delivering 1U and 2U platforms to market can be found in the company’s latest white paper, “Tachyum’s Prodigy Evolves to Next Level: 1U/2U Solutions for Data Centers”.

Follow Tachyum

About Tachyum

Tachyum is transforming the economics of AI, HPC, public and private cloud workloads with Prodigy, the world’s first Universal Processor. Prodigy unifies the functionality of a CPU, a GPU, and a TPU in a single processor to deliver industry-leading performance, cost and power efficiency for both specialty and general-purpose computing. As global data center emissions continue to contribute to a changing climate, with projections of their consuming 10 percent of the world’s electricity by 2030, the ultra-low power Prodigy is positioned to help balance the world’s appetite for computing at a lower environmental cost. Tachyum recently received a major purchase order from a US company to build a large-scale system that can deliver more than 50 exaflops performance, which will exponentially exceed the computational capabilities of the fastest inference or generative AI supercomputers available anywhere in the world today. When complete in 2025, the Prodigy-powered system will deliver a 25x multiplier vs. the world’s fastest conventional supercomputer – built just this year – and will achieve AI capabilities 25,000x larger than models for ChatGPT4. Tachyum has offices in the United States and Slovakia. For more information, visit https://www.tachyum.com/.

Tachyum Demonstrates PMU Running on Prodigy FPGA Emulation System

Tue, 20 Feb 2024 00:00:00 +0000

LAS VEGAS, February 20, 2024 – Tachyum® today announced that it has added a Performance Monitoring Unit (PMU) to its Prodigy® FPGA emulation system, empowering customers and partners with the ability to address bottlenecks and better optimize Prodigy performance for all applications and workloads.

The PMU is an essential tool for collecting information about performance bottlenecks. It offers the ability to record a wide range of events that encompass every aspect of the Prodigy Universal Processor without slowing down the application itself. Tools like perf then present this information after the application is finished, enabling the identification and characterization of performance bottlenecks that may exist in the processor core, full mesh interconnect fabric, memory, and I/O subsystems. Perf is a go-to instrument for everybody working on performance assessment and tuning under Linux. The PMU’s wide range of performance counters – supported by both software C-model and FPGA – facilitates both system debugging and performance tuning.

Tachyum’s PMU enables faster time to market by allowing customers and partners to quickly identify performance issues and rapidly converge to a solution for all phases of go-to-market, including evaluation, development and final production testing. It provides an invaluable tool suite for customers spanning a broad array of markets, including AI, HPC and cloud computing.

“As the Prodigy Universal Processor continues towards successful product release, it is important that we focus on performance measurement and optimization metrics to ensure it is fully tuned before going into initial production later this year,” said Dr. Radoslav Danilak, founder and CEO of Tachyum. “In a production system, dozens of counters will be available to customer performance teams enabling them to achieve maximum efficiency with their applications running at scale within Prodigy equipped data centers.”

As a Universal Processor offering industry-leading performance for all workloads, Prodigy-powered data center servers can seamlessly and dynamically switch between computational domains (such as AI/ML, HPC, and cloud) with a single homogeneous architecture. By eliminating the need for expensive dedicated AI hardware and dramatically increasing server utilization, Prodigy reduces CAPEX and OPEX significantly while delivering unprecedented data center performance, power, and economics. Prodigy integrates 192 high-performance custom-designed 64-bit compute cores, to deliver up to 4.5x the performance of the highest-performing x86 processors for cloud workloads, up to 3x that of the highest performing GPU for HPC, and 6x for AI applications.

In the next few weeks, Tachyum’s engineers will focus on bringing error detection and correction capabilities to the Prodigy FPGA to enable server-class RAS (reliability, availability and serviceability) features that exceed other server processors.

A video demonstrating a small subset of the total Prodigy performance counters on Prodigy FPGA is available below.

Follow Tachyum

About Tachyum

Tachyum is transforming the economics of AI, HPC, public and private cloud workloads with Prodigy, the world’s first Universal Processor. Prodigy unifies the functionality of a CPU, a GPU, and a TPU in a single processor to deliver industry-leading performance, cost and power efficiency for both specialty and general-purpose computing. As global data center emissions continue to contribute to a changing climate, with projections of their consuming 10 percent of the world’s electricity by 2030, the ultra-low power Prodigy is positioned to help balance the world’s appetite for computing at a lower environmental cost. Tachyum recently received a major purchase order from a US company to build a large-scale system that can deliver more than 50 exaflops performance, which will exponentially exceed the computational capabilities of the fastest inference or generative AI supercomputers available anywhere in the world today. When complete in 2025, the Prodigy-powered system will deliver a 25x multiplier vs. the world’s fastest conventional supercomputer – built just this year – and will achieve AI capabilities 25,000x larger than models for ChatGPT4. Tachyum has offices in the United States and Slovakia. For more information, visit https://www.tachyum.com/.

Tachyum Upgrades Software Package in Advance of Beta Release of the Prodigy Universal Processor

Wed, 14 Feb 2024 00:00:00 +0000

LAS VEGAS, February 14, 2024 – Tachyum® today announced that it has upgraded the software stack for the Prodigy® Universal Processor before the anticipated launch of its beta version around the end of quarter. Quality completion of Prodigy’s software stack is a key component as the company continues to advance towards chip production and distribution.

Tachyum software engineers have worked hard to enable the full potential of Prodigy with the development of an ecosystem of applications, system software, frameworks and libraries that are ported to run natively on Prodigy hardware. Once the software package completes its testing and runs cleanly on the FPGA, the company can fully transition to advancing the Universal Processor into production.

The Prodigy software distribution is a completely integrated software stack and package that is ready for deployment “as is.” It is available as a single pre-installed image for Tachyum’s early adopters and customers. Applications have been tested to work out of the box so that customers can immediately start using the reference design. If users encounter any issues during deployment, the software can be quickly and easily restored to its original image.

Included in the software distribution package as part of alpha testing are:

Latest versions of the QEMU emulator 8.2
GCC 13.2 (GNU Compiler Collection) and glibc 2.39 (GNU C Library)
Linux 6.6 LTS (Long Term Support), which contains a large number of changes, updates and improvements

The company also announced plans to switch to the LLVM 18 release once it is available to download. LLVM plays a significant role in every major AI framework. Additionally, it is in the process of adding RAS (Reliability Accessibility Serviceability) in the form of an EDAC (Error Detection and Correction) driver in the next few weeks. Based on customer requests for server applications Tachyum agreed to add bootable SSD RAID next quarter to its UEFI.

As a Universal Processor offering industry-leading performance for all workloads, Prodigy-powered data center servers can seamlessly and dynamically switch between computational domains (such as AI/ML, HPC, and cloud) with a single homogeneous architecture. By eliminating the need for expensive dedicated AI hardware and dramatically increasing server utilization, Prodigy reduces CAPEX and OPEX significantly while delivering unprecedented data center performance, power, and economics. Prodigy integrates 192 high-performance custom-designed 64-bit compute cores, to deliver up to 4.5x the performance of the highest-performing x86 processors for cloud workloads, up to 3x that of the highest performing GPU for HPC, and 6x for AI applications.

“Having a robust software stack tested and ready to go upon the launch of the Prodigy Universal Processor chip is key to rapid adoption by data centers around the world looking to leverage their existing applications while achieving industry-leading performance for hyperscale, high-performance computing and artificial intelligence workloads,” said Dr. Radoslav Danilak, founder and CEO of Tachyum. “With each new enrichment we are able to incorporate into Prodigy’s software stack, we magnify the ability of a Prodigy platform ready to revolutionize the world.”

Follow Tachyum

About Tachyum

Tachyum is transforming the economics of AI, HPC, public and private cloud workloads with Prodigy, the world’s first Universal Processor. Prodigy unifies the functionality of a CPU, a GPU, and a TPU in a single processor to deliver industry-leading performance, cost and power efficiency for both specialty and general-purpose computing. As global data center emissions continue to contribute to a changing climate, with projections of their consuming 10 percent of the world’s electricity by 2030, the ultra-low power Prodigy is positioned to help balance the world’s appetite for computing at a lower environmental cost. Tachyum recently received a major purchase order from a US company to build a large-scale system that can deliver more than 50 exaflops performance, which will exponentially exceed the computational capabilities of the fastest inference or generative AI supercomputers available anywhere in the world today. When complete in 2025, the Prodigy-powered system will deliver a 25x multiplier vs. the world’s fastest conventional supercomputer – built just this year – and will achieve AI capabilities 25,000x larger than models for ChatGPT4. Tachyum has offices in the United States and Slovakia. For more information, visit https://www.tachyum.com/.

Tachyum Democratizes AI for All with $5000 Prodigy ATX Platform

Thu, 08 Feb 2024 00:00:00 +0000

LAS VEGAS, February 8, 2024 – Tachyum®, creator of Prodigy®, the world’s first Universal Processor, today released a white paper that details how its Prodigy ATX Platform will democratize AI for those who may not normally have access to sophisticated AI models. The Prodigy ATX Platform allows everyone to run cutting-edge AI models for as low as $5,000.

Built from the ground up to provide leading-edge AI features that address the emerging demand for AI across a wide range of applications and workloads, Prodigy’s AI subsystem incorporates innovative features that deliver the high performance and efficiency required of AI environments. The white paper shows how a single Prodigy system with 1 Terabyte (TB) of memory can run a ChatGPT4 model with 1.7 trillion parameters, whereas it requires 52 NVIDIA H100 GPUs to run the same model at significantly higher cost and power consumption.

Since LLMs (Large Language Models) are so memory capacity intensive, determining the memory footprint for an LLM is critical. Just as critical is the use of the latest technology that optimizes the memory footprint for state-of-the-art LLMs, which can have trillions of parameters. Prodigy benefits from its advanced AI subsystem that supports leading-edge data types such as 4-bit TAI and effective 2-bit weights with FP8 per activation that greatly reduces the memory footprint required for LLMs.

Since the Prodigy ATX Platform is intended to leverage pre-trained models and focus on inference, Tachyum reviews the assumptions for the memory footprint required for inference. Assuming an LLM with 1 trillion parameters running with FP8, the memory required for weights is 1 TB. There is additional overhead for inference memory that is typically 0.2x the model size, or 200 GB, that is added for runtime calculations for activations. For FP8, the total memory required for a 1 trillion parameter model is approximately 1.2 TB.

Considering Tachyum’s 4-bit TAI with 4-bit weights, the memory needed for weights is reduced to 500 GB and the runtime inference memory is fixed at 200 GB for a total requirement of 700 GB. Now, considering running TAI sparse with 2-bit weights the memory required for weights is further reduced to 250 GB. With the 200 GB runtime inference memory, the total requirement is 450 GB.

If we repeat these steps for the 1.7 trillion parameter ChatGPT4 LLM: the inference memory is 0.2 x 1.7 trillion = 340 GB, and the total memory needed for FP8 is 1.7 TB for weights + 340 GB = 2.04 TB. Going from FP8 to 4-bit TAI the weights require 2x less memory, so the memory consumed by the weights is 1.7/2 = 850 GB, and the total memory requirement is 850 GB + 340 GB = 1.19 TB.

If we go to TAI sparse with 2-bit weights, the memory requirement for weights is reduced to 425 GB. If we now add the 425 GB to the memory required for inference, 425 GB + 340 GB, we come up with a total memory footprint requirement of 765 GB, which fits well below the 1 TB of commodity system memory that the Prodigy ATX Platform offers, and there is plenty of headroom, so even larger LLMs can be supported.

Key architectural components of the Prodigy ATX Platform outlined in the white paper include:

Single-socket 96-core Prodigy Universal Processor running up to 5.7 GHz with 8 DDR5 memory controllers
16 64 GB commodity DIMMs (2 DIMMs/channel) running up to DDR5-6400 with a total memory capacity of 1 TB
3 PCIe 5.0 slots that support up to full height and full-length form factors:
- 1 slot x16 with 16 lanes
- 2 slots x16 with 8 lanes
3 M.2 NVMe slots supporting 22x80mm form factor
1200W power supply

The platform benefits from Prodigy’s unique “half die” solution that allows a full 192-core device to function as two separate 96-core devices. This architecture provides Tachyum with increased yield for 96-core devices, lowering platform costs and helping make the Prodigy ATX Platform even more affordable.

The Prodigy ATX Platform addresses an extensive range of use cases, such as language generation, language translation, code generation, virtual tutoring, content summarization, sentiment analysis, fraud or cyber-attack detection, and content filtering. The platform benefits the many pre-trained LLMs available today with support for both proprietary and open-source models.

“Generative AI will be widely used far faster than anyone originally anticipated,” said Dr. Radoslav Danilak, founder and CEO of Tachyum. “In a year or two, AI will be a required component on websites, chatbots and other critical productivity components to ensure a good user experience. Prodigy’s powerful AI capabilities enable LLMs to run much easier and more cost-effectively than existing CPU+GPGPU-based systems, empowering organizations of all sizes to compete in AI initiatives that otherwise would be dominated by the largest players in their industry.”

Prodigy provides both the high performance required for cloud and HPC/AI workloads within a single architecture. As a Universal Processor offering utility for all workloads, Prodigy-powered data center servers can seamlessly and dynamically switch between computational domains. By eliminating the need for expensive dedicated AI hardware and dramatically increasing server utilization, Prodigy reduces CAPEX and OPEX significantly while delivering unprecedented data center performance, power, and economics. Prodigy integrates 192 high-performance custom-designed 64-bit compute cores, to deliver up to 4x the performance of the highest-performing x86 processors for cloud workloads, up to 3x that of the highest performing GPU for HPC, and 6x for AI applications.

Those interested in reading the “Tachyum’s Prodigy ATX Platform Democratizing AI for Everyone” white paper can download a copy from the company’s website.

Follow Tachyum

About Tachyum

Tachyum is transforming the economics of AI, HPC, public and private cloud workloads with Prodigy, the world’s first Universal Processor. Prodigy unifies the functionality of a CPU, a GPU, and a TPU in a single processor to deliver industry-leading performance, cost and power efficiency for both specialty and general-purpose computing. As global data center emissions continue to contribute to a changing climate, with projections of their consuming 10 percent of the world’s electricity by 2030, the ultra-low power Prodigy is positioned to help balance the world’s appetite for computing at a lower environmental cost. Tachyum recently received a major purchase order from a US company to build a large-scale system that can deliver more than 50 exaflops performance, which will exponentially exceed the computational capabilities of the fastest inference or generative AI supercomputers available anywhere in the world today. When complete in 2025, the Prodigy-powered system will deliver a 25x multiplier vs. the world’s fastest conventional supercomputer – built just this year – and will achieve AI capabilities 25,000x larger than models for ChatGPT4. Tachyum has offices in the United States and Slovakia. For more information, visit https://www.tachyum.com/.

Tachyum Well Positioned for an Exceptional 2024

Tue, 09 Jan 2024 00:00:00 +0000

LAS VEGAS, January 9, 2024 – With volume production of Prodigy®, the world’s first Universal Processor, scheduled for 2024, Tachyum® officials expect to build off the momentum it gained through a series of successes last year while capitalizing on the current semiconductor upcycle to revolutionize the data center.

Anticipated company highlights of the coming year include:

Advancements in Product Development – Reference design is expected to soon be revealed once the Tachyum team finishes final debugging and cleaning of the chip.
Increased Business Development – With the final reference chip available, Tachyum will target early adopter markets, including high performance computing and artificial intelligence applications.

Tachyum diligently worked throughout 2023 to ensure that it is positioned at the precipice of success for the coming year. Achievements in hardware and software, advancements in IP component technology and a major purchase order to build a large-scale system were key contributing factors that are expected to benefit the company’s upcoming year.

In 2023, Tachyum received a major purchase order from a U.S. company to build a large-scale system based on its 5nm Prodigy Universal Processor chip. With performance of 50 EF of FP64 and 8 ZF of AI training for large language models, the announced system has gained a lot of attention from other interested parties looking to build a similar scale system for their AI applications and workloads.

The company made available its Tachyum Processing Unit (TPU®) IP as a licensable core, allowing developers to take full advantage of intelligent, datacenter-trained AI when making IoT and Edge devices.

Details for a new approach to designing HPC/AI supercomputer data centers have been released as part of a series of whitepapers in order to prepare developers for what they can expect when Prodigy is commercially available. Topics included how to use 4-bit Tachyum AI (TAI) and 2-bit effective per weight (TAI2) formats in Large Language Models (LMMs); how Prodigy overcomes hardware inefficiencies that make Generative AI cost prohibitive and energy excessive; and how Prodigy enables quantization using 8-bit floating point (FP8) to enable enormous model sizes.

To put Prodigy’s capabilities in perspective, a 1.6 trillion parameters Switch Transformer would require 52 NVIDIA H200 80GB GPUs at a cost of $41,789 each and seven Supermicro GPU servers at $25,000 each for a total cost of $2,348,028. In contrast, a single Prodigy socket system with 2TB DDR5 DRAM could fit and run such big models and bring them into the mainstream for generative AI applications for only $23,000 – 1/100th the cost of the first example.

Software achievements this past year included:

Development completion of the software stack for Prodigy, allowing for alpha testing to begin
Addition of LLVM for AI and Linux Rust support
Running of Xen hypervisor on QEMU software emulator
Running of Kubernetes natively on Prodigy ISA

Hardware achievements completed in 2023 include:

Completion of vector based High-Performance LINPACK (HPL) testing using 1kb vectors on the Prodigy FPGA
Running of FreeBSD OS on Prodigy FPGA
Seamless execution of non-native (x86_64) applications under Linux on FPGA
Integration of Baseboard Management Controller (BMC) and Unified Extensible Firmware Interface (UEFI) platforms into the Prodigy FPGA prototype system
Porting of other applications to test a wide range of workloads on the FPGA prototype

Additionally, Tachyum Prodigy 2 was selected by the Important Project of Common European Interests (IPCEI) program representing Slovakia to deliver exa-scale HPC and zetta-scale AI for Europe. The company hosted Slovak President Zuzana Čaputová at its Bratislava offices to demonstrate the progress of the Prodigy Universal Processor.

As a Universal Processor offering utility for all workloads, Prodigy-powered data center servers can seamlessly and dynamically switch between computational domains (such as AI/ML, HPC, and cloud) on a single architecture. By eliminating the need for expensive dedicated AI hardware and dramatically increasing server utilization, Prodigy reduces CAPEX and OPEX significantly while delivering unprecedented data center performance, power, and economics. Prodigy integrates 192 high-performance custom-designed 64-bit compute cores, to deliver up to 4,5x the performance of the highest-performing x86 processors for cloud workloads, up to 3x that of the highest performing GPU for HPC, and 6x for AI applications.

“Tachyum showed its resilience during a challenging 2023 when we experienced the worst semiconductor downturn in years and weathered the challenges faced with failed memory and interface IP blocks from a key vendor,” said Dr. Radoslav Danilak, founder and CEO of Tachyum. “The successes we were able to achieve while tweaking our product roadmap have brought us to a 2024 full of anticipation as we move towards volume production of Prodigy and the fulfillment of a multibillion-dollar sales pipeline. We look forward to fulfilling our commitment to transforming ordinary data centers into Universal Computing Centers in the near future.”

Follow Tachyum

About Tachyum

Tachyum is transforming the economics of AI, HPC, public and private cloud workloads with Prodigy, the world’s first Universal Processor. Prodigy unifies the functionality of a CPU, a GPU, and a TPU in a single processor to deliver industry-leading performance, cost and power efficiency for both specialty and general-purpose computing. As global data center emissions continue to contribute to a changing climate, with projections of their consuming 10 percent of the world’s electricity by 2030, the ultra-low power Prodigy is positioned to help balance the world’s appetite for computing at a lower environmental cost. Tachyum recently received a major purchase order from a US company to build a large-scale system that can deliver more than 50 exaflops performance, which will exponentially exceed the computational capabilities of the fastest inference or generative AI supercomputers available anywhere in the world today. When complete in 2025, the Prodigy-powered system will deliver a 25x multiplier vs. the world’s fastest conventional supercomputer – built just this year – and will achieve AI capabilities 25,000x larger than models for ChatGPT4. Tachyum has offices in the United States and Slovakia. For more information, visit https://www.tachyum.com/.

Tachyum 8 AI Zettaflops Blueprint to Solve OpenAI Capacity Limitation

Tue, 12 Dec 2023 00:00:00 +0000

LAS VEGAS, December 12, 2023 – Tachyum®, creator of Prodigy®, the world’s first Universal Processor, today released a white paper that presents how Tachyum’s customers planning to build new HPC/AI supercomputer data centers far exceeding not only the performance of existing supercomputers but also the target performance for next-generation systems.

Built from the ground up to provide the highest performance and efficiency, Prodigy’s revolutionary new architecture enables supercomputers to be deployed in fully homogeneous environments, providing simple development, deployment and maintenance. The solution is ideally suited for OpenAI, others like Microsoft Azure, CoreWeave, Ori, etc. and research facilities needing AI datacenters that today do not have the system architecture capable of serving all interested customers.

Developed by Tachyum’s world-class systems, solutions and software engineering teams, the Prodigy-enabled supercomputer, commissioned by a U.S. company this year, delivers the unprecedented performance of 50 exaflops of IEEE double-precision 64-bit floating-point operation and 8 zettaflops of AI training for large language models.

For the supercomputer solution referenced in the white paper, Prodigy has a custom 46RU rack with a liquid-cooled reference design. The rack supports 33 four-socket 1U servers for a total of 132 Prodigy processors. The racks have a modular architecture with the ability to combine them into a two-rack cabinet to optimize floor space.

Tachyum’s HPC/AI software stack provides a complete software environment to enable Prodigy family HPC/AI deployments, delivering full support for all aspects of HPC/AI clusters from low-level firmware to complete HPC/AI applications, and incorporating leading edge software environments for networking and storage. Tachyum’s software team has already integrated a software package for HPL LINPACK and other software for HPC with AI running on Prodigy FPGA soon.

“After our announcement of the purchase order we received this year, we attracted a lot of attention from other interested parties, several of them large organizations, looking to build a similar scale system for their AI applications and workloads,” said Dr. Radoslav Danilak, founder and CEO of Tachyum. “Prodigy’s system architecture fits well into a wide range of deployments, including those that need data center scale-out once the infrastructure for it is already in place. The scale of machines enabled by this new HPC/AI supercomputer data center likely will determine who will win the fight for compute and AI supremacy in the world.”

Tachyum’s latest white paper follows previous releases detailing how to use 4-bit Tachyum AI (TAI) and 2-bit effective per weight (TAI2) formats in Large Language Models (LLM) quantization without accuracy degradation, reducing the cost of LMMs by up to 100x and bringing them to the mainstream.

Those interested in reading the “Tachyum 50EF/8ZF Datacenter Can Solve OpenAI and Other Problems” white paper can download a copy.

Follow Tachyum

About Tachyum

Tachyum is transforming the economics of AI, HPC, public and private cloud workloads with Prodigy, the world’s first Universal Processor. Prodigy unifies the functionality of a CPU, a GPU, and a TPU in a single processor to deliver industry-leading performance, cost and power efficiency for both specialty and general-purpose computing. As global data center emissions continue to contribute to a changing climate, with projections of their consuming 10 percent of the world’s electricity by 2030, the ultra-low power Prodigy is positioned to help balance the world’s appetite for computing at a lower environmental cost. Tachyum recently received a major purchase order from a US company to build a large-scale system that can deliver more than 50 exaflops performance, which will exponentially exceed the computational capabilities of the fastest inference or generative AI supercomputers available anywhere in the world today. When complete in 2025, the Prodigy-powered system will deliver a 25x multiplier vs. the world’s fastest conventional supercomputer – built just this year – and will achieve AI capabilities 25,000x larger than models for ChatGPT4. Tachyum has offices in the United States and Slovakia. For more information, visit https://www.tachyum.com/.

Tachyum Runs Supercomputing Vector LINPACK on Prodigy FPGA

Wed, 06 Dec 2023 00:00:00 +0000

LAS VEGAS, December 6, 2023 – Tachyum®, creator of Prodigy®, the world’s first Universal Processor, today announced it has successfully completed vector-based High-Performance LINPACK (HPL) testing using 1kb vectors on the Prodigy FPGA.

LINPACK, a widely used benchmarking program for supercomputers, measures a system’s floating-point computing power by solving a random, dense system of linear equations to determine performance and accuracy. Following scalar LINPACK benchmarks using Prodigy’s IEEE-compliant scalar Floating-Point Unit (FPU), Tachyum has now advanced to vector LINPACK.

Prodigy’s vector unit was designed with a range of cutting-edge features to provide industry-leading performance. Prodigy has two vector pipelines with a 1024b-wide data path. Executing 2x1K SIMD (Single Instruction/Multiple Data) operations per cycle, Prodigy delivers 32 double precision FMA (Fused IEEE floating-point multiply-add) operations per cycle, delivering 64 double precision floating-point operations, because the FMA consists of multiplication and addition per cycle and per core.

Prodigy’s memory access micro-architecture uses an innovative approach to support unaligned data, meaning that unaligned operations are processed without incurring the performance loss that many other architectures in the market encounter. These features, along with high clock rates, enable Tachyum to deliver the highest vectorized data processing performance in the industry. Additionally, masking, unaligned memory access and loop control operations of the Prodigy architecture allow for efficient auto-vectorization in compilers. Compilers from Tachyum’s software ecosystem already fully utilize these features.

Vector instructions have been verified running cleanly on the Prodigy FPGA, including vectorization in GNU Compiler Collection (GCC), libraries, Linux-supporting vectors, as well as running applications and delivering the correct results.

With vector LINPACK completed, Tachyum is now focused on the final stage of vector unit verification and testing with FPU: AI matrix operations.

“Supercomputing is more than merely an exponent of standard computing. There are complex processes to measure performance and capability: ensuring that all bits in vectors are correctly wired, that IEEE flags are correctly reported, and many cases of data-shuffling vector operations, but we got there,” said Dr. Radoslav Danilak, founder and CEO of Tachyum. “We can see the light at the end of the tunnel as we move towards volume production in 2024, and towards fulfilling our multibillion-dollar sales pipeline.”

A video demonstrating the vector LINPACK on Prodigy FPGA is available for viewing below.

Follow Tachyum

About Tachyum

Tachyum is transforming the economics of AI, HPC, public and private cloud workloads with Prodigy, the world’s first Universal Processor. Prodigy unifies the functionality of a CPU, a GPU, and a TPU in a single processor to deliver industry-leading performance, cost and power efficiency for both specialty and general-purpose computing. As global data center emissions continue to contribute to a changing climate, with projections of their consuming 10 percent of the world’s electricity by 2030, the ultra-low power Prodigy is positioned to help balance the world’s appetite for computing at a lower environmental cost. Tachyum recently received a major purchase order from a US company to build a large-scale system that can deliver more than 50 exaflops performance, which will exponentially exceed the computational capabilities of the fastest inference or generative AI supercomputers available anywhere in the world today. When complete in 2025, the Prodigy-powered system will deliver a 25x multiplier vs. the world’s fastest conventional supercomputer – built just this year – and will achieve AI capabilities 25,000x larger than models for ChatGPT4. Tachyum has offices in the United States and Slovakia. For more information, visit https://www.tachyum.com/.

Tachyum Prodigy Complete Software Package Enters Testing

Tue, 28 Nov 2023 00:00:00 +0000

LAS VEGAS, November 28, 2023 – Tachyum® today announced that it has completed development of the software stack for the Prodigy® Universal Processor and has now entered the alpha testing phase in preparation of chip production and distribution planned for next year.

Tachyum software engineers have worked hard to enable the full potential of Prodigy with the development of a rich ecosystem of applications, system software, frameworks and libraries that are ported to run natively on Prodigy hardware. Once the software package completes its testing and runs cleanly on FPGA, the company can fully transition to advancing the Universal Processor into production.

Included in the software distribution package as part of alpha testing are:

System software including UEFI, Linux, SE-Linux, FreeBSD, KVM, XEN, GCC, LLVM, TensorFlow and PyTorch. From UEFI (Unified Extensible Firmware Interface), a next-generation interface between operating systems and platform firmware, customers can select Linux, SE-Linux, FreeBSD, KVM, Xen and others.
Middleware like containers, etc.
Applications such as databases, web servers, email servers and more
AI software and HPC software
Out-of-box binary compatibility with x86, ARM and RISC-V through dynamic binary translation
Demo applications and tests of applications

As a Universal Processor offering industry-leading performance for all workloads, Prodigy-powered data center servers can seamlessly and dynamically switch between computational domains (such as AI/ML, HPC, and cloud) with a single homogeneous architecture. By eliminating the need for expensive dedicated AI hardware and dramatically increasing server utilization, Prodigy reduces CAPEX and OPEX significantly while delivering unprecedented data center performance, power, and economics. Prodigy integrates 192 high-performance custom-designed 64-bit compute cores, to deliver up to 4.5x the performance of the highest-performing x86 processors for cloud workloads, up to 3x that of the highest performing GPU for HPC, and 6x for AI applications.

“We recognized early on that if we wanted to revolutionize the data center by producing the world’s smallest, fastest and greenest general-purpose chip that it would be imperative to deliver a stack capable of handling the full gamut of software that would run on it,” said Dr. Radoslav Danilak, founder and CEO of Tachyum. “As our Universal Processor chip nears production-ready status, we are proud to have developed an expansive portfolio of software ready to run natively on Prodigy. Having it available for alpha testing is a massive step towards revolutionizing the data center.”

A video demonstrating the alpha release of Tachyum’s Prodigy software distribution package is available for viewing below.

Follow Tachyum

About Tachyum

Tachyum is transforming the economics of AI, HPC, public and private cloud workloads with Prodigy, the world’s first Universal Processor. Prodigy unifies the functionality of a CPU, a GPGPU, and a TPU in a single processor that delivers industry-leading performance, cost, and power efficiency for both specialty and general-purpose computing. When hyperscale data centers are provisioned with Prodigy, all AI, HPC, and general-purpose applications can run on the same infrastructure, saving companies billions of dollars in hardware, footprint, and operational expenses. As global data center emissions contribute to a changing climate, and consume more than four percent of the world’s electricity—projected to be 10 percent by 2030—the ultra-low power Prodigy Universal Processor is a potential breakthrough for satisfying the world’s appetite for computing at a lower environmental cost. Prodigy, now in its final stages of testing and integration before volume manufacturing, is being adopted in prototype form by a rapidly growing customer base, and robust purchase orders signal a likely IPO in late 2024. Tachyum has offices in the United States and Slovakia. For more information, visit https://www.tachyum.com/.

Tachyum Cuts Cost of Large Language Models Up to 100x Bringing them to Mainstream

Tue, 14 Nov 2023 00:00:00 +0000

LAS VEGAS, November 14, 2023 – Tachyum® today announced the release of a white paper detailing how to use 4-bit Tachyum AI (TAI) and 2-bit effective per weight (TAI2) formats in Large Language Models (LLM) quantization without accuracy degradation. Tachyum hardware also enables workable LLM with 1-bit per weight with higher degradation than TAI2 and its AI scientists are continuing to further improve performance to reduce degradation as Tachyum looks to bring it to mainstream.

Tachyum addresses massive LLM models with capabilities that have dramatically increased by more than a thousand times over the past few years. Examples of these increases include ChatGPT-3.5 LLM with 175 billion parameters, the PALM LLM with 530 billion dense parameters and the Switch Transformer with 1.6 trillion sparse parameters.

For example, a 1.6 trillion parameters Switch Transformer would require 52x NVIDIA H100 80GB GPUs at $41,789 each + 7 x $25,000 for Supermicro GPU servers = $2,348,028. In contrast, a $23,000 single Prodigy socket system with 2TB DDR5 DRAM could fit and run such big models and bring them into the mainstream for generative AI applications.

AI systems built on Prodigy universal chips with 256PB DDR5 DRAM (Dynamic Access Random Memory) using FP8 (8-bit floating point) and 4-bit Tachyum AI (TAI) data formats can fit up to 100 quadrillion parameter models. It can serve more than 150,000x ChatGPT models or 610,000x PALM2 models and represents huge possibilities for using LLM as a mainstream technology in various industries from retail and e-commerce, marketing, finance, cyber security, military to healthcare including faster drug development or practical implementation of personalized medicine in hospitals.

Effective deployment of LLMs requires low-bit quantization to minimize model size and inference cost. Low-bit integer formats, like INT8 and INT4, have been the conventional choice, however, the emerging low-bit exponential formats offer a compelling alternative. At reasonable costs, LLMs could be deployed by enterprises small to large across a variety of industries. LLMs could be an integral part of an organization’s web presence to provide an interactive experience, such as enabling the ability of visitors to ask questions naturally vs. entering search terms.

“By combining TAI 4-bit and effective 2-bit weights with FP8 per activation, we are capable of quantizing LLMs without much accuracy degradation,” said Dr. Radoslav Danilak, founder and CEO of Tachyum. “Our techniques avoid expensive multiplication while simultaneously reducing the size of the model by 4x to 8x, enabling generative AI models that can be applied in use cases from complex language modelling tasks, text generation, drug and chip design, few-shot learning and reasoning to protein sequence modelling. Whole new avenues of calculations can be opened with Tachyum AI.”

As a Universal Processor offering industry leading performance for all workloads, Prodigy-powered data center servers can seamlessly and dynamically switch between computational domains (such as AI/ML, HPC, and cloud) with a single homogeneous architecture. By eliminating the need for expensive dedicated AI hardware and dramatically increasing server utilization, Prodigy reduces CAPEX and OPEX significantly while delivering unprecedented data center performance, power, and economics. Prodigy integrates 192 high-performance custom-designed 64-bit compute cores, to deliver up to 4.5x the performance of the highest-performing x86 processors for cloud workloads, up to 3x that of the highest performing GPU for HPC, and 6x for AI applications.

Those interested in reading “Mainstreaming Large Language Models With 2-bit TAI Weights” can download a copy from the Tachyum website.

Follow Tachyum

About Tachyum

Tachyum is transforming the economics of AI, HPC, public and private cloud workloads with Prodigy, the world’s first Universal Processor. Prodigy unifies the functionality of a CPU, a GPU, and a TPU in a single processor to deliver industry-leading performance, cost and power efficiency for both specialty and general-purpose computing. As global data center emissions continue to contribute to a changing climate, with projections of their consuming 10 percent of the world’s electricity by 2030, the ultra-low power Prodigy is positioned to help balance the world’s appetite for computing at a lower environmental cost. Tachyum recently received a major purchase order from a US company to build a large-scale system that can deliver more than 50 exaflops performance, which will exponentially exceed the computational capabilities of the fastest inference or generative AI supercomputers available anywhere in the world today. When complete in 2025, the Prodigy-powered system will deliver a 25x multiplier vs. the world’s fastest conventional supercomputer – built just this year – and will achieve AI capabilities 25,000x larger than models for ChatGPT4. Tachyum has offices in the United States and Slovakia. For more information, visit https://www.tachyum.com/.

Tachyum Successfully Runs FreeBSD on Prodigy FPGA Prototype

Tue, 07 Nov 2023 00:00:00 +0000

LAS VEGAS, November 7, 2023 – Tachyum® today announced that the FreeBSD operating system is now running on the Prodigy® FPGA prototype as part of efforts to ensure that customer requirements are met prior to tape-out of the world’s first universal processor. This announcement follows Tachyum successfully running the Linux OS as part of its emulation system.

FreeBSD powers modern servers, desktops and embedded platforms in environments that value performance, stability and security. It is the platform of choice for many of the busiest websites and the most pervasive embedded networking and storage devices.

As Tachyum readies a rich and robust collection of applications, system software, frameworks and libraries in advance of Prodigy chip production, increased customer demand for FreeBSD made testing it under FPGA a priority. The company’s ability to prioritize customer needs is one of the key reasons Tachyum accelerated support for LLVM and is a reflection of how its software engineers continue to work to enable the full potential of Prodigy.

Tachyum previously announced that it had completed FreeBSD porting. Applications for FreeBSD have now been added to the Quality Assurance (QA) testing timeline, which have been made a priority with the addition of FreeBSD to the FPGA prototype, in order to further satisfy customer demand.

“For Prodigy to truly be considered a universal processor, it is imperative that we are able to provide capabilities to any number of OS environments, software applications and system architectures,” said Dr. Radoslav Danilak, founder and CEO of Tachyum. “With the completion of testing of FreeBSD, we now have all planned operating systems successfully vetted under Prodigy FPGA hardware and gets us one step closer to revolutionizing the data center as we know it when we begin shipping products next year.”

A video demonstration of FreeBSD running on Prodigy FPGA is available for viewing below.

Follow Tachyum

About Tachyum

Tachyum is transforming the economics of AI, HPC, public and private cloud workloads with Prodigy, the world’s first Universal Processor. Prodigy unifies the functionality of a CPU, a GPU, and a TPU in a single processor to deliver industry-leading performance, cost and power efficiency for both specialty and general-purpose computing. As global data center emissions continue to contribute to a changing climate, with projections of their consuming 10 percent of the world’s electricity by 2030, the ultra-low power Prodigy is positioned to help balance the world’s appetite for computing at a lower environmental cost. Tachyum recently received a major purchase order from a US company to build a large-scale system that can deliver more than 50 exaflops performance, which will exponentially exceed the computational capabilities of the fastest inference or generative AI supercomputers available anywhere in the world today. When complete in 2025, the Prodigy-powered system will deliver a 25x multiplier vs. the world’s fastest conventional supercomputer – built just this year – and will achieve AI capabilities 25,000x larger than models for ChatGPT4. Tachyum has offices in the United States and Slovakia. For more information, visit https://www.tachyum.com/.

Tachyum Details 4-bit and 2-bit New AI Frontier Formats

Tue, 24 Oct 2023 00:00:00 +0000

LAS VEGAS, October 24, 2023 – Tachyum^®, creator of Prodigy®, the world’s first Universal Processor, today released a new white paper on the use of 4-bit Tachyum AI (TAI) format, including 2-bit effective per weight (TAI2) format, in both inference and Deep Neural Network (DNN) learning.

“Image AI Processing at the Next Level With 4b TAI & 2b Effective Weights” presents the 4-bit TAI format, an extremely silicon- and power-efficient approach that limits memory and bandwidth for even large models.

To reduce the computational complexity of neural networks, a process of quantization and pruning is used to reduce the number of parameters (weights) of a training model. Low-bit floating point formats have recently emerged as promising for Deep Neural Network (DNN) quantization. While INT8-level quantization is currently used, INT4—which would double throughput compared to INT8—is currently being evaluated by industry for feasibility and potential loss of accuracy and quality.

The TAI format is capable of significantly exceeding INT4 with a logarithmic 4-bit format and 2-bit effective weight format. Tachyum’s AI team has published experimental results showing that TAI2 effectively runs on Prodigy.

To verify the usability of the TAI formats with 4-bits and 2-bits effective per weight, Tachyum selected several models from the field of image processing and computer vision. From image, classifiers selected were ResNet20, ResNet32, ResNet34 and SWIN transformer; from segmentation models UNet, FastSCNN and ConvMixer; and from detectors SSD. The team tested well-known datasets such as CIFAR10, CIFAR100, and Imagenet for image classification tasks; Cityscapes and Kits19 for segmentation tasks; and VOC and COCO for object detection.

“The TAI format is far more powerful and efficient than INT4, and 2-bit per weight is ahead of any other technology of our time for reduced bandwidth and compute requirements,” said Dr. Radoslav Danilak, founder and CEO of Tachyum. “We are seeing the future of AI now, and it’s truly a new frontier, even for large training models and complex image processing.”

Tachyum experiments with training models in TAI with 2b effective weight formats have shown that it is a usable format for general models that are not specially optimized for specific tasks or for specific datasets. The degradation of the models was at an acceptable level, even in the case of post-training pruning. TAI format can also be used as a format suitable for pre-training the model.

Large models perform well even with 2b per weight; smaller models do not experience performance issues, so they are not a primary target.

Tachyum TPU® (Tachyum Processing Unit) Inference intellectual property is available as a licensable core for models trained on Tachyum’s Prodigy Universal Processor chip.

“Image AI Processing at the Next Level With 4b TAI & 2b Effective Weights” white paper is now available on Tachyum’s website.

As a Universal Processor offering utility for all workloads, Prodigy-powered data center servers can seamlessly and dynamically switch between computational domains (such as AI/ML, HPC, and cloud) on a single architecture. By eliminating the need for expensive dedicated AI hardware and dramatically increasing server utilization, Prodigy reduces CAPEX and OPEX significantly while delivering unprecedented data center performance, power, and economics. Prodigy integrates 192 high-performance custom-designed 64-bit compute cores, to deliver up to 4.5x the performance of the highest-performing x86 processors for cloud workloads, up to 3x that of the highest performing GPU for HPC, and 6x for AI applications.

Follow Tachyum

About Tachyum

Tachyum is transforming the economics of AI, HPC, public and private cloud workloads with Prodigy, the world’s first Universal Processor. Prodigy unifies the functionality of a CPU, a GPU, and a TPU in a single processor to deliver industry-leading performance, cost and power efficiency for both specialty and general-purpose computing. As global data center emissions continue to contribute to a changing climate, with projections of their consuming 10 percent of the world’s electricity by 2030, the ultra-low power Prodigy is positioned to help balance the world’s appetite for computing at a lower environmental cost. Tachyum recently received a major purchase order from a US company to build a large-scale system that can deliver more than 50 exaflops performance, which will exponentially exceed the computational capabilities of the fastest inference or generative AI supercomputers available anywhere in the world today. When complete in 2025, the Prodigy-powered system will deliver a 25x multiplier vs. the world’s fastest conventional supercomputer – built just this year – and will achieve AI capabilities 25,000x larger than models for ChatGPT4. Tachyum has offices in the United States and Slovakia. For more information, visit https://www.tachyum.com/.

Tachyum Books Purchase Order to Build System with 25,000x ChatGPT4 Capacity and 25x Faster than Current Supercomputers

Tue, 03 Oct 2023 00:00:00 +0000

LAS VEGAS, October 3, 2023 – Tachyum® announced that it has accepted a major purchase order from a US company to build a large-scale system, based on its 5nm Prodigy® Universal Processor chip, which delivers more than 50 exaflops performance that will exponentially exceed the computational capabilities of the fastest inference or generative AI supercomputers available anywhere in the world today.

Prodigy, the world’s first Universal Processor, is engineered to transform the capacity, efficiency and economics of datacenters through its industry-leading performance for hyperscale, high-performance computing and AI workloads. When complete, the Prodigy-powered system will deliver a 25x multiplier vs. the world’s fastest conventional supercomputer built just this year, and will achieve AI capabilities 25,000x larger than models for ChatGPT4.

Exascale supercomputers can quickly analyze and process massive amounts of data to solve complex problems that had previously been intractable. Prodigy overdelivers on traditional exascale capabilities by providing industry-leading performance, significantly reduced energy consumption, improved server utilization and space efficiency. Prodigy’s exponential increases in memory, storage and compute to enable breakthroughs across datacenter, AI and HPC workloads in government, research and academia, business, manufacturing and other industries.

The human brain consists of around 100 billion neurons and approximately 200 trillion synaptic connections. Assuming a few bytes per synaptic connection, it would require 100’s of terabytes of memory. Therefore, the Prodigy system with hundreds of petabytes of DRAM will have 100x more memory than needed.

Installation of this Prodigy-enabled solution will begin in 2024 and reach full capacity in 2025. Among the deliverables are:

8 Zettaflops AI training for big language models
16 Zettaflops of image and video processing
Ability to fit more than 100,000x PALM2 530B parameter models OR 25,000x ChatGPT4 1.7T parameter models with base memory and 100,000x ChatGPT4 with 4x of base DRAM
Upgradable memory of the base model system
Hundreds of petabytes of DRAM and exabytes of flash-based primary storage
4-socket, liquid-cooled nodes connected to 400G RoCE ethernet, with the capability to double to an 800G all non-blocking and non-overprovisioned switching fabric

Tachyum’s proprietary TPU® AI Inference IP supports Tachyum AI (TAI) data type and provides even more; breakthrough efficiency for video and large language model data formats that otherwise would require excessive power and expensive multipliers in matrix multiplication to achieve.

“The unprecedented scale and computational power required as part of this installation simply could not be provided by any chip manufacturer on the market today,” said Dr. Radoslav Danilak, founder and CEO of Tachyum. “While there are startups receiving billions of dollars, based on their promise of achieving similar capabilities sometime in the future, only Tachyum is positioned to deliver the capability to economically build order-of-magnitude bigger machines that potentially enable the transition to cognitive AI, beginning later this year. This purchase order is a testament to our first-to-market position and our ability to provide a positive impact to worldwide AI markets.”

Follow Tachyum

About Tachyum

Tachyum is transforming the economics of AI, HPC, public and private cloud workloads with Prodigy, the world’s first Universal Processor. Prodigy unifies the functionality of a CPU, a GPGPU, and a TPU in a single processor that delivers industry-leading performance, cost, and power efficiency for both specialty and general-purpose computing. When hyperscale data centers are provisioned with Prodigy, all AI, HPC, and general-purpose applications can run on the same infrastructure, saving companies billions of dollars in hardware, footprint, and operational expenses. As global data center emissions contribute to a changing climate, and consume more than four percent of the world’s electricity—projected to be 10 percent by 2030—the ultra-low power Prodigy Universal Processor is a potential breakthrough for satisfying the world’s appetite for computing at a lower environmental cost. Prodigy, now in its final stages of testing and integration before volume manufacturing, is being adopted in prototype form by a rapidly growing customer base, and robust purchase orders signal a likely IPO in late 2024. Tachyum has offices in the United States and Slovakia. For more information, visit https://www.tachyum.com/.

Tachyum Runs x86-64 Binaries on Prodigy FPGA as Its Key Milestone

Tue, 26 Sep 2023 00:00:00 +0000

LAS VEGAS, September 26, 2023 – Tachyum® today announced that it has successfully demonstrated a seamless execution of a non-native (x86_64) application under Linux running on the Prodigy FPGA emulation system. This capability will ensure the availability of a wide range of software applications to customers from day one when the Prodigy Universal Processor becomes generally available next year.

x86 (also known as x64, x86_64, AMD64 or Intel 64) is a family of Instruction Set Architectures (ISA) that accounts for the widest installation base for datacenter workloads today.

Because x86’s complexity and power consumption create a barrier to very high performance in nanometer-class processors, Tachyum’s revolutionary architecture does not include any x86 specific hardware, as this would limit Prodigy’s performance. Instead, a standard dynamic binary translator efficiently provides the ability to run unmodified Linux x86 binaries right out of the box.

Tachyum customers who want to run applications beyond those provided by Tachyum in Prodigy Linux distribution image, will initially utilize a dynamic binary translator while they port their legacy x86 applications to Prodigy native ISA, usually within 12-18 months. Prodigy allows users to mix x86 applications with native Prodigy applications, as previously demonstrated, by seamlessly running Prodigy native Apache web servers combined with x86 Linux binary databases.

Until customers migrate to fully native software environments, Prodigy will deliver 2x performance instead of the anticipated 3x performance versus Intel high-end CPUs on SPECint2017 rate. However, for most customers, only 20% of runtime will be legacy x86 applications, with the remaining 80% of runtime dedicated to native applications, such as TensorFlow, PyTorch, databases, and more, which Tachyum has already ported to native Prodigy ISA, typically delivering 80% x 3x + 20% x 2x = 2.8x more performance from day one.

Prodigy’s x86 performance outdistances other implementations, including the recently introduced Intel® Advanced Performance Extensions (Intel® APX) in Intel’s attempt to match new and modern ISAs; ARM or RISC-V processors that are far too slow for server applications; or Apple’s Rosetta dynamic binary translations that have sufficient performance to run x86 applications on laptops where low power is more important than the high performance needed for servers.

“Demonstrating the ability to run x86-64 binary applications on the Prodigy processor emulation is a key milestone for Tachyum and further validates our architecture before tape out,” said Dr. Radoslav Danilak, founder and CEO of Tachyum. “Proving all these existing binary applications on our FPGA emulator is critical in order to get Prodigy ready for volume production next year, allowing us to begin fulfilling the billions of dollars of orders we already have in our sales pipeline.”

A video demonstration of x86-64 code running on Prodigy is available for viewing below.

Follow Tachyum

About Tachyum

Tachyum is transforming the economics of AI, HPC, public and private cloud workloads with Prodigy, the world’s first Universal Processor. Prodigy unifies the functionality of a CPU, a GPGPU, and a TPU in a single processor that delivers industry-leading performance, cost, and power efficiency for both specialty and general-purpose computing. When hyperscale data centers are provisioned with Prodigy, all AI, HPC, and general-purpose applications can run on the same infrastructure, saving companies billions of dollars in hardware, footprint, and operational expenses. As global data center emissions contribute to a changing climate, and consume more than four percent of the world’s electricity—projected to be 10 percent by 2030—the ultra-low power Prodigy Universal Processor is a potential breakthrough for satisfying the world’s appetite for computing at a lower environmental cost. Prodigy, now in its final stages of testing and integration before volume manufacturing, is being adopted in prototype form by a rapidly growing customer base, and robust purchase orders signal a likely IPO in late 2024. Tachyum has offices in the United States and Slovakia. For more information, visit https://www.tachyum.com/.

Tachyum Adds LLVM for AI and Linux Rust Support

Tue, 19 Sep 2023 00:00:00 +0000

LAS VEGAS, September 19, 2023 – Tachyum® today announced the expansion of its Prodigy software ecosystem with the addition of LLVM for AI and Linux Rust support. This is the latest addition to an already rich and robust collection of applications, system software, frameworks and libraries running under QEMU emulation and tested by FPGA emulation in advance of Prodigy chip production.

The LLVM Project is a collection of modular and reusable compiler and toolchain technologies. Rust is a multi-paradigm, general-purpose programming language with an emphasis on performance, type safety and concurrency, and has become the second language officially accepted for Linux kernel development.

LLVM plays a large role in every major AI framework, including PyTorch and TensorFlow, with their AI compilers based on LLVM for native instruction generation. Additionally, standalone AI compilers, like Apache TVM, are also based on the LLVM compiler infrastructure. An LLVM backend supports the Tachyum ISA, particularly vector and tensor instructions in both full and low precisions, allowing it to take full advantage of Prodigy’s hardware features that are specifically designed to boost AI workload performance. This AI compiler backend provides the ultimate performance for inference applications.

With Rust, Linux developers can create new functionality faster and with greater confidence. It is both fast and memory efficient. Its rich type system and ownership model guarantee memory and thread safety. Rust is being incorporated into Prodigy’s Linux kernel, relying on LLVM. Support for GCC Rust will come later when the language becomes readily available.

“As impressive as the hardware architecture of the Tachyum Prodigy Universal Processor is, it is the software that we support that will allow customers and partners to unlock the full potential of the chip,” said Dr. Radoslav Danilak, founder and CEO of Tachyum. “With initial availability of LLVM planned for the first quarter of 2024, I am proud of our software engineering team’s efforts to have it available ahead of schedule. Our team is now hard at work building Tachyum software distribution images, and a quarter after the alpha release, a beta version will be available for our customers to test.”

As a Universal Processor offering utility for all workloads, Prodigy-powered data center servers can seamlessly and dynamically switch between computational domains (such as AI/ML, HPC, and cloud) on a single architecture. By eliminating the need for expensive dedicated AI hardware and dramatically increasing server utilization, Prodigy reduces CAPEX and OPEX significantly while delivering unprecedented data center performance, power, and economics. Prodigy integrates 192 high-performance custom-designed 64-bit compute cores, to deliver up to 4.5x the performance of the highest-performing x86 processors for cloud workloads, up to 3x that of the highest performing GPU for HPC, and 6x for AI applications.

A video demonstration of LLVM running on Prodigy is available below.

Follow Tachyum

About Tachyum

Tachyum is transforming the economics of AI, HPC, public and private cloud workloads with Prodigy, the world’s first Universal Processor. Prodigy unifies the functionality of a CPU, a GPGPU, and a TPU in a single processor that delivers industry-leading performance, cost, and power efficiency for both specialty and general-purpose computing. When hyperscale data centers are provisioned with Prodigy, all AI, HPC, and general-purpose applications can run on the same infrastructure, saving companies billions of dollars in hardware, footprint, and operational expenses. As global data center emissions contribute to a changing climate, and consume more than four percent of the world’s electricity—projected to be 10 percent by 2030—the ultra-low power Prodigy Universal Processor is a potential breakthrough for satisfying the world’s appetite for computing at a lower environmental cost. Prodigy, now in its final stages of testing and integration before volume manufacturing, is being adopted in prototype form by a rapidly growing customer base, and robust purchase orders signal a likely IPO in late 2024. Tachyum has offices in the United States and Slovakia. For more information, visit https://www.tachyum.com/.

Tachyum Offers Its TPU Inference IP to Edge and Embedded Markets

Tue, 12 Sep 2023 00:00:00 +0000

LAS VEGAS, September 12, 2023 – Tachyum® today announced that it is expanding the unique value proposition of its Tachyum Prodigy by offering its Tachyum TPU® (Tachyum Processing Unit) intellectual property as a licensable core, allowing developers to take full advantage of intelligent, datacenter-trained AI when making IoT and Edge devices.

Tachyum’s Prodigy is the first Universal Processor combining General Purpose Processors, High Performance Computing (HPC), Artificial Intelligence (AI), Deep Machine Learning, Explainable AI, Bio AI and other AI disciplines with a single chip. With the tremendous growth of the AI chipset market for edge inference, Tachyum is looking to extend its proprietary Tachyum AI data type beyond the datacenter by providing its internationally registered and trademarked IP to outside developers.

Key features of the TPU inference and generative AI/ML IP architecture include architectural transactional and cycle accurate simulators; tools and compilers support; and hardware licensable IP, including RTL in Verilog, UVM Testbench and synthesis constraints. Tachyum has 4b per weight working for AI training and 2b per weight as part of the proprietary Tachyum AI (TAI) data type, which will be announced later this year.

“Inference and generative AI is coming to almost every consumer product and we believe that licensing TPU is a key avenue for Tachyum to proliferate our world-leading AI into this marketplace for models trained on Tachyum’s Prodigy Universal Processor chip,” said Dr. Radoslav Danilak, founder and CEO of Tachyum. “As Tachyum is the only owner of the TPU trademark within the AI space, it is a valuable corporate asset to not only Tachyum but to all the vendors who respect that trademark and ensure that they properly license its use as part of their products.”

As a Universal Processor offering utility for all workloads, Prodigy-powered data center servers can seamlessly and dynamically switch between computational domains (such as AI/ML, HPC, and cloud) on a single architecture. By eliminating the need for expensive dedicated AI hardware and dramatically increasing server utilization, Prodigy reduces CAPEX and OPEX significantly while delivering unprecedented data center performance, power, and economics. Prodigy integrates 192 high-performance custom-designed 64-bit compute cores, to deliver up to 4.5x the performance of the highest-performing x86 processors for cloud workloads, up to 3x that of the highest performing GPU for HPC, and 6x for AI applications.

For licensing opportunities of Tachyum’s TPU IP, interested vendors are invited to contact the company.

Follow Tachyum

About Tachyum

Tachyum is transforming the economics of AI, HPC, public and private cloud workloads with Prodigy, the world’s first Universal Processor. Prodigy unifies the functionality of a CPU, a GPGPU, and a TPU in a single processor that delivers industry-leading performance, cost, and power efficiency for both specialty and general-purpose computing. When hyperscale data centers are provisioned with Prodigy, all AI, HPC, and general-purpose applications can run on the same infrastructure, saving companies billions of dollars in hardware, footprint, and operational expenses. As global data center emissions contribute to a changing climate, and consume more than four percent of the world’s electricity—projected to be 10 percent by 2030—the ultra-low power Prodigy Universal Processor is a potential breakthrough for satisfying the world’s appetite for computing at a lower environmental cost. Prodigy, now in its final stages of testing and integration before volume manufacturing, is being adopted in prototype form by a rapidly growing customer base, and robust purchase orders signal a likely IPO in late 2024. Tachyum has offices in the United States and Slovakia. For more information, visit https://www.tachyum.com/.

Tachyum FP8 Super-Sparsity Is Showing Path to Efficient Generative AI

Tue, 22 Aug 2023 00:00:00 +0000

LAS VEGAS, August 22, 2023 – Tachyum® announced today the release of a new research paper to address how Prodigy®, the world’s first Universal Processor, will transform the quality, efficiency, and economics of generative AI (GenAI).

“Unprecedented Scale and Efficiency in Generative AI with FP8 8:3 Super-Sparsity” offers technical information on how Prodigy can more effectively meet the computation and scale requirements of generative AI, which trains on massive data sets to create original results, rather than identifying or analyzing known data. The larger the training data, the better and more accurate the output of GenAI. ChatGPT 3.5, a quintessential example of a generative AI model, has 175 billion trainable parameters, and ChatGPT 4.0 increases this by a factor of 10 to 1.76 trillion parameters, with another 10x increase possible in the near future.

Language models like ChatGPT, vision models, and other GenAI tools have improved dramatically due to successful scale-up, resulting in impressive few-shot capabilities close to that of a human being. These growing numbers of parameters require corresponding increases in computational power to train AI systems: high memory capacity, high processing performance, and high memory bandwidth to optimize the efficiency of large and dense models. Today, the scale of the largest AI computation is doubling every six months, outpacing Moore’s Law by 7x; and moving from generative AI to cognitive AI is expected to require 100-1000x more capacity.

To address memory and energy consumption, quantization reduces the precision of parameters as a means of compressing deep-neural networks (DNNs). Similarly, pruning removes redundant/insensitive parameters to reduce density. While density is often necessary to successfully train the model, once trained, many parameters can be removed without any quality degradation.

In this paper Tachyum shows how Prodigy overcomes the hardware inefficiencies that make GenAI cost-prohibitive and energy-excessive. Prodigy enables quantization using 8-bit floating point (FP8) with 8:3 block pruning, improving performance, power, and memory bandwidth to enable enormous model sizes. Tachyum’s recommendations significantly increase training speed, and reduce the memory footprint of the model after training. Super-sparsity FP8 8:3 greatly reduces the model sizes, important for language models, as well as power and area—important for edge and IoT applications.

“GenAI is a truly transformational technology, but its value cannot be realized, nor can it be widely adopted, without solving the hardware challenges of running such large models,” said Dr. Radoslav Danilak, founder and CEO of Tachyum. “With Prodigy poised to become the mainstream cost-efficient high performance processor in 2024, these compression approaches, together with hardware support, will enable even small to midsized enterprises and academic users to work with large, dense deep learning models.”

Because Prodigy offers increased memory over currently available AI processors—2TB using low-cost DRAM and 32TB/socket, with a 4-socket Prodigy platform supporting low cost 8TB and up to 128TB of TSV DDR5 DRAM — a single Prodigy chip can replace more than 10 competitor units, delivering unprecedented performance, scalability and efficiency.

FP8 8:3 models must be trained on Tachyum chips to achieve the proper computational efficiency. FP8 8:3 inference and generative AI IP is available now to partners and customers; a license includes all necessary software, which is process-independent.

As a Universal Processor offering utility for all workloads, Prodigy-powered data center servers can seamlessly and dynamically switch between computational domains (such as AI/ML, HPC, and cloud) on a single architecture. By eliminating the need for expensive dedicated AI hardware and dramatically increasing server utilization, Prodigy reduces CAPEX and OPEX significantly while delivering unprecedented data center performance, power, and economics. Prodigy integrates 192 high-performance custom-designed 64-bit compute cores, to deliver up to 4.5x the performance of the highest-performing x86 processors for cloud workloads, up to 3x that of the highest performing GPU for HPC, and 6x for AI applications.

“Unprecedented Scale and Efficiency in Generative AI with FP8 8:3 Super-Sparsity” is now available for download.

Follow Tachyum

About Tachyum

Tachyum is transforming the economics of AI, HPC, public and private cloud workloads with Prodigy, the world’s first Universal Processor. Prodigy unifies the functionality of a CPU, a GPGPU, and a TPU in a single processor that delivers industry-leading performance, cost, and power efficiency for both specialty and general-purpose computing. When hyperscale data centers are provisioned with Prodigy, all AI, HPC, and general-purpose applications can run on the same infrastructure, saving companies billions of dollars in hardware, footprint, and operational expenses. As global data center emissions contribute to a changing climate, and consume more than four percent of the world’s electricity—projected to be 10 percent by 2030—the ultra-low power Prodigy Universal Processor is a potential breakthrough for satisfying the world’s appetite for computing at a lower environmental cost. Prodigy, now in its final stages of testing and integration before volume manufacturing, is being adopted in prototype form by a rapidly growing customer base, and robust purchase orders signal a likely IPO in late 2024. Tachyum has offices in the United States and Slovakia. For more information, visit https://www.tachyum.com/.

Tachyum Achieves 192-Core Chip After Switch to New EDA Tools

Tue, 15 Aug 2023 00:00:00 +0000

LAS VEGAS, August 15, 2023 – Tachyum® today announced that new EDA tools, utilized during the physical design phase of the Prodigy® Universal Processor, have allowed the company to achieve significantly better results with chip specifications than previously anticipated, after the successful change in physical design tools – including an increase in the number of Prodigy cores to 192.

After RTL design coding, Tachyum began work on completing the physical design (the actual placement of transistors and wires) for Prodigy. After the Prodigy design team had to replace IPs, it also had to replace RTL simulation and physical design tools. Armed with a new set of EDA tools, Tachyum was able to optimize settings and options that increased the number of cores by 50 percent, and SERDES from 64 to 96 on each chip. Die size grew minimally, from 500mm2 to 600mm2 to accommodate improved physical capabilities. While Tachyum could add more of its very efficient cores and still fit into the 858mm2 reticle limit, these cores would be memory bandwidth limited, even with 16 DDR5 controllers running in excess of 7200MT/s. Tachyum cores have much higher performance than any other processor cores.

Other improvements realized during the physical design stage are:

Increase of the chip L2/L3 cache from 128MB to 192MB
Support of DDR5 7200 memory in addition to DDR5 6400
More speed with 1 DIMM per channel
Larger package accommodates additional 32 serial links and as many as 32 DIMMs connected to a single Prodigy chip

“At every step of the process in bringing Prodigy to market, our innovation allows us to push beyond the limits of traditional design and continue to exceed even our lofty design goals,” said Dr. Radoslav Danilak, founder and CEO of Tachyum. “We have achieved better results and timing with our new EDA PD tools. They are so effective that we wish we had used them from the beginning of the process but as the saying goes, ‘Better now than never.’ While we did not have any choice but to change EDA tools, our physical design (PD) team worked hard to redo physical design and optimizations with the new set of PD tools, as we approach volume-level production.”

As a universal processor, the patented Prodigy architecture enables it to switch seamlessly and dynamically from normal CPU tasks to AI/ML workloads, so it delivers high AI/ML performance in both training and inference. AI/ML is increasingly important in the banking industry, and used to identify fraud and cyberattacks before serious financial damage can be done.

Prodigy delivers unprecedented data center performance, power, and economics, reducing CAPEX and OPEX significantly. Because of its utility for both high-performance and line-of-business applications, Prodigy-powered data center servers can seamlessly and dynamically switch between workloads, eliminating the need for expensive dedicated AI hardware and dramatically increasing server utilization. Tachyum’s Prodigy delivers performance up to 4x that of the highest performing x86 processors (for cloud workloads) and up to 3x that of the highest performing GPU for HPC and 6x for AI applications.

With the achievement of this latest Prodigy milestone, Tachyum’s next steps are to complete the substrate package and socket design to accommodate more SERDES lines. Delivery of the first Prodigy high-performance processors remains on track by the end of the year.

Follow Tachyum

About Tachyum

Tachyum is transforming the economics of AI, HPC, public and private cloud workloads with Prodigy, the world’s first Universal Processor. Prodigy unifies the functionality of a CPU, a GPGPU, and a TPU in a single processor that delivers industry-leading performance, cost, and power efficiency for both specialty and general-purpose computing. When hyperscale data centers are provisioned with Prodigy, all AI, HPC, and general-purpose applications can run on the same infrastructure, saving companies billions of dollars in hardware, footprint, and operational expenses. As global data center emissions contribute to a changing climate, and consume more than four percent of the world’s electricity—projected to be 10 percent by 2030—the ultra-low power Prodigy Universal Processor is a potential breakthrough for satisfying the world’s appetite for computing at a lower environmental cost. Prodigy, now in its final stages of testing and integration before volume manufacturing, is being adopted in prototype form by a rapidly growing customer base, and robust purchase orders signal a likely IPO in late 2024. Tachyum has offices in the United States and Slovakia. For more information, visit https://www.tachyum.com/.

Tachyum Demonstrates Support for Xen Hypervisor

Wed, 02 Aug 2023 00:00:00 +0000

LAS VEGAS, August 2, 2023 – Tachyum® today announced that it has completed porting of all the software necessary for chip validation on the Prodigy FPGA prototype before it is sent for tape out with the successful running of the Xen hypervisor on a QEMU software emulator.

Xen is a type-1 hypervisor providing services that allow multiple virtual machines to run on a single physical host system, sharing resources such as memory, processing and devices. Its bare metal capabilities allow the Xen hypervisor to communicate directly from the host’s hardware, acting like a lightweight operating system. It supports four types of virtualization: paravirtualization, full hardware virtualization, full virtualization with a set of paravirtualization drivers, and paravirtualization with hardware virtualization support.

Tachyum demonstrated its support of XEN running in three different modes:

Standard “dom0” mode – the most privileged domain and the only virtual machine that has direct access to hardware by default. From dom0, the hypervisor can be managed and unprivileged domains (domU) can be launched.
Hyperlaunch – enabled seamless transition for existing systems that require a dom0. Provides a new general capability to build and launch alternative configurations of VMs, including support for static partitioning and accelerated start of VMs during host boot while adhering to the principles of least privileged.
Dom0less – a set of Xen features that enable the deployment of a Xen system without a control domain (dom0). It is a hyperlaunch-like configuration with static partitioning without using dom0. This means that there are no virtual devices and hardware resources are statically assigned to a specific virtual machine using the passthrough method.

“In order to assure that Prodigy truly becomes the means for transforming data centers into Universal Computing Centers, we need to ensure that it can properly handle any number of components, software and technology available now or in the future,” said Dr. Radoslav Danilak, founder and CEO of Tachyum. “By successfully running Xen in our Prodigy software emulator, we can confidently assure organizations that they will reap the full benefits of Prodigy when it becomes commercially available. I’m proud of the work of our highly skilled software team as they continue to complete the final components needed before tape out.”

To showcase its latest milestone, Tachyum has made available a video demonstrating the hyperlaunch of two VMs – dom0 and domU1 – running on QEMU. The video is available below.

Follow Tachyum

About Tachyum

Tachyum is transforming the economics of AI, HPC, public and private cloud workloads with Prodigy, the world’s first Universal Processor. Prodigy unifies the functionality of a CPU, a GPGPU, and a TPU in a single processor that delivers industry-leading performance, cost, and power efficiency for both specialty and general-purpose computing. When hyperscale data centers are provisioned with Prodigy, all AI, HPC, and general-purpose applications can run on the same infrastructure, saving companies billions of dollars in hardware, footprint, and operational expenses. As global data center emissions contribute to a changing climate, and consume more than four percent of the world’s electricity—projected to be 10 percent by 2030—the ultra-low power Prodigy Universal Processor is a potential breakthrough for satisfying the world’s appetite for computing at a lower environmental cost. Prodigy, now in its final stages of testing and integration before volume manufacturing, is being adopted in prototype form by a rapidly growing customer base, and robust purchase orders signal a likely IPO in late 2024. Tachyum has offices in the United States and Slovakia. For more information, visit https://www.tachyum.com/.

Tachyum Unveils Path to Sustainable Green Datacenters

Tue, 25 Jul 2023 00:00:00 +0000

LAS VEGAS, July 25, 2023 – Tachyum™ announced today the release of the white paper, “Tachyum Prodigy – Solution for Data Centers that are Hungry for Energy,” explaining how Prodigy^®, the world’s first Universal Processor, is ideally suited to help overcome increasingly excessive energy use in data centers.

Data centers are among the most energy-demanding facilities in the world. A report by the International Energy Agency (IEA) indicates that 3% of global electricity use comes from data centers and data transmission networks. Data centers are expected to consume 20% of the world’s energy supply by 2025 as the demand for digital services continues to grow.

Tachyum’s white paper details how Prodigy delivers disruptive innovation for data centers with unprecedented power efficiency by unifying the CPU, GPGPU and TPU into a single processor die, enabling groundbreaking performance while removing the need for power-hunger, costly accelerators. The white paper further shows how this compares to other data center solutions with higher performance/watt than both CPUs and GPUs; up to 4.4x higher performance/W than Intel 8490-H for cloud computing; and up to 13.4x higher performance/W than Nvidia H100 for generative AI.

“Keeping pace with the energy demands of today’s hyperscale data centers is simply not attainable,” said Dr. Radoslav Danilak, founder and CEO of Tachyum. “Additionally, massive growth from the rapid adoption of AI can only be accomplished by enabling a single hardware platform, like Prodigy, that can handle multiple workloads and avoids the time and energy from separate, custom-built solutions. Those reading the white paper will see that this approach makes the most sense and will become more evident as power demand increases.”

Prodigy provides both the high performance required for cloud and HPC/AI workloads within a single architecture. As a Universal Processor offering utility for all workloads, Prodigy-powered data center servers can seamlessly and dynamically switch between computational domains. By eliminating the need for expensive dedicated AI hardware and dramatically increasing server utilization, Prodigy reduces CAPEX and OPEX significantly while delivering unprecedented data center performance, power, and economics. Prodigy integrates 128 high-performance custom-designed 64-bit compute cores, to deliver up to 4x the performance of the highest-performing x86 processors for cloud workloads, up to 3x that of the highest performing GPU for HPC, and 6x for AI applications.

Those interested in reading the “Tachyum Prodigy – Solution for Data Centers that are Hungry for Energy” white paper can download a copy.

Follow Tachyum

About Tachyum

Tachyum is transforming the economics of AI, HPC, public and private cloud workloads with Prodigy, the world’s first Universal Processor. Prodigy unifies the functionality of a CPU, a GPGPU, and a TPU in a single processor that delivers industry-leading performance, cost, and power efficiency for both specialty and general-purpose computing. When hyperscale data centers are provisioned with Prodigy, all AI, HPC, and general-purpose applications can run on the same infrastructure, saving companies billions of dollars in hardware, footprint, and operational expenses. As global data center emissions contribute to a changing climate, and consume more than four percent of the world’s electricity—projected to be 10 percent by 2030—the ultra-low power Prodigy Universal Processor is a potential breakthrough for satisfying the world’s appetite for computing at a lower environmental cost. Prodigy, now in its final stages of testing and integration before volume manufacturing, is being adopted in prototype form by a rapidly growing customer base, and robust purchase orders signal a likely IPO in late 2024. Tachyum has offices in the United States and Slovakia. For more information, visit https://www.tachyum.com/.

Tachyum Presents Paper on Prodigy Processors in Blockchain for Banking and FinTech

Tue, 18 Jul 2023 00:00:00 +0000

LAS VEGAS, July 18, 2023 – Tachyum® today released a paper “Credit Unions, Blockchain, CBDC, FinTech and Tachyum Prodigy®” explaining how Prodigy, the world’s first Universal Processor, can transform transactional banking while minimizing the monetary and environmental costs of blockchain’s electrical energy requirements.

With Blockchain, transactions are transmitted directly from payer to payee (person to person) without an intermediary. According to PricewaterhouseCoopers, 77 percent of FinTech businesses are either currently using blockchain or transitioning to blockchain-based products and services. Credit Unions in particular are leading the trend toward blockchain-based banking products and services, just as the industry once led the transition to internet banking as well as other technologies widely adopted later by the commercial and consumer banking sector.

“Blockchain and cryptocurrency already raise legitimate concerns about unsustainable energy consumption, but as the technology becomes more widespread in financial services, we’re going to see consumption increase to an alarming level,” said Dr. Radoslav Danilak, founder and CEO of Tachyum. “We cannot possibly build thousands of power plants worldwide to supply electricity for blockchain, so the real solution is greater compute efficiency—in other words, faster processing that is far more energy-efficient—such as Prodigy.”

Blockchain uses vastly more energy than centrally controlled banking systems. Currently, electricity use for crypto-assets is 120-240 billion kilowatt-hours per year, which exceeds the total annual electricity consumption of entire nations, such as Argentina or Australia.

Prodigy offers a compute fabric to proliferate blockchain with optional [Central Bank] Digital Currency (CBDC) banking. A 128-core processor, Prodigy runs at clock speeds over 5GHz, outperforming all other CPU platforms in normal data center workloads and delivering extremely high hashing function performance required for blockchain banking. Prodigy achieves this performance using one-tenth the power (core vs. core), which slashes the TCO of hyperscale data center operations to one-fourth that of its closest competitors.

Those interested in reading the full paper can download a copy at the Blockchain white paper page.

Follow Tachyum

About Tachyum

Tachyum is transforming AI, HPC, public and private cloud data center markets with its recently launched flagship product. Prodigy, the world’s first Universal Processor, unifies the functionality of a CPU, a GPU, and a TPU into a single processor that delivers industry-leading performance, cost, and power efficiency for both specialty and general-purpose computing. When Prodigy processors are provisioned in a hyperscale data center, they enable all AI, HPC, and general-purpose applications to run on one hardware infrastructure, saving companies billions of dollars per year. With data centers currently consuming over 4% of the planet’s electricity, predicted to be 10% by 2030, the ultra-low power Prodigy Universal Processor is critical to continue doubling worldwide data center capacity every four years. Tachyum, co-founded by Dr. Radoslav Danilak is building the world’s fastest AI supercomputer (128 AI exaflops) in the EU based on Prodigy processors. Tachyum has offices in the United States and Slovakia. For more information, visit https://www.tachyum.com/.

Tachyum Demonstrated Beta Version of BMC/UEFI on Prodigy FPGA

Tue, 11 Jul 2023 00:00:00 +0000

LAS VEGAS, July 11, 2023 – Tachyum™ announced today it is nearing completion of software packages for customers and OEM/ODM partners on Prodigy, the world’s first Universal Processor.

Following porting and demonstrating standalone applications, Tachyum is progressing to full software distribution. It completed integration of BMC (Baseboard Management Controller) and UEFI (Unified Extensible Firmware Interface) platforms and now it entered beta version. These firmware packages are now functional and integrated into the Prodigy FPGA prototype system.

UEFI is a next-generation interface between operating systems and platform firmware that has replaced legacy BIOS interfaces. The newer UEFI standard delivers security, modularity, and faster booting. Tachyum provides UEFI firmware as a primary method of initializing the Prodigy platform, and Prodigy’s BIOS configuration can be changed with UEFI services.

BMC is a service processor embedded in a server that can monitor the physical state of the server and provides administrators with full remote control for maintenance.

BMC and UEFI integration are critical milestones, and signals Prodigy’s integrated system software stack has reached the alpha stage. After one quarter of alpha release, it advances to beta and finally to production release with chip sampling.

“Since completing testing processes, our engineering team has focused on integration and interoperability to drive the increasing system-level maturity of Prodigy,” said Dr. Radoslav Danilak, founder and CEO of Tachyum. “These software packages are important boxes to tick for deployment in customer environments and OEM partner products.”

For a video demonstration of the UEFI/BMC running on the Prodigy FPGA prototype, visit our Presentations page, or watch the video below.

Prodigy delivers a revolutionary new architecture that unifies the functionality of CPU, GPGPU, and TPU into a single chip. As a Universal Processor, Prodigy provides both the high performance required for cloud and HPC/AI workloads within a single architecture. Because of its utility for all workloads, Prodigy-powered data center servers can seamlessly and dynamically switch between computational domains.

By eliminating the need for expensive dedicated AI hardware and dramatically increasing server utilization, Prodigy reduces CAPEX and OPEX significantly while delivering unprecedented data center performance, power, and economics. Prodigy integrates 128 high-performance custom-designed 64-bit compute cores, to deliver up to 4x the performance of the highest-performing x86 processors for cloud workloads, up to 3x that of the highest performing GPU for HPC, and 6x for AI applications.

Follow Tachyum

About Tachyum

Tachyum is transforming the economics of AI, HPC, public and private cloud workloads with Prodigy, the world’s first Universal Processor. Prodigy unifies the functionality of a CPU, a GPGPU, and a TPU in a single processor that delivers industry-leading performance, cost, and power efficiency for both specialty and general-purpose computing. When hyperscale data centers are provisioned with Prodigy, all AI, HPC, and general-purpose applications can run on the same infrastructure, saving companies billions of dollars in hardware, footprint, and operational expenses. As global data center emissions contribute to a changing climate, and consume more than four percent of the world’s electricity—projected to be 10 percent by 2030—the ultra-low power Prodigy Universal Processor is a potential breakthrough for satisfying the world’s appetite for computing at a lower environmental cost. Prodigy, now in its final stages of testing and integration before volume manufacturing, is being adopted in prototype form by a rapidly growing customer base, and robust purchase orders signal a likely IPO in late 2024. Tachyum has offices in the United States and Slovakia. For more information, visit https://www.tachyum.com/.

Tachyum Concludes FPGA Prototype Orders As Tape-Out Approaches

Tue, 27 Jun 2023 00:00:00 +0000

LAS VEGAS, June 27, 2023 – Tachyum™ announced today it will cease taking orders for its Prodigy Universal Processor Field Programmable Gate Array (FPGA) emulation system boards effective immediately. The company’s release of the final Prodigy build for tape-out is imminent.

New partners and customers who wish to work with Prodigy FPGAs for product evaluation, performance measurements, software development, debugging and compatibility testing can arrange for private testing in Tachyum’s facility. As these are shared systems, they can’t be used for classified or proprietary data, or data subject to regulatory governance.

The Prodigy hardware emulator consists of multiple FPGA and IO boards connected by cables in a rack. A single board with four FPGAs emulates eight Prodigy processor cores (a small fraction of the final Prodigy product design, which consists of 128 cores) including vector and matrix fixed and floating-point processing units. Deploying more FPGAs will improve test cycles by orders of magnitude to achieve target quality, a risk reduction mechanism for early adopters.

“We thank those who have worked with our prototypes, put Prodigy to the test in their environments, validated compatibility with their hardware and software systems, and provided feedback to our team,” said Radoslav Danilak, founder and CEO of Tachyum. “Concluding this stage of Prodigy’s development brings us closer to the finish line, closer to volume production, and closer to transforming the industry with unprecedented processing power and efficiency.”

Prodigy delivers unprecedented data center performance, power, and economics, reducing CAPEX and OPEX significantly. Because of its utility for both high-performance and line-of-business applications, Prodigy-powered data center servers can seamlessly and dynamically switch between workloads, eliminating the need for expensive dedicated AI hardware and dramatically increasing server utilization. Tachyum’s Prodigy integrates 128 high-performance custom-designed 64-bit compute cores, to deliver up to 4x the performance of the highest-performing x86 processors for cloud workloads, up to 3x that of the highest performing GPU for HPC, and 6x for AI applications.

Follow Tachyum

About Tachyum

Tachyum is transforming AI, HPC, public and private cloud data center markets with its recently launched flagship product. Prodigy, the world’s first Universal Processor, unifies the functionality of a CPU, a GPU, and a TPU into a single processor that delivers industry-leading performance, cost, and power efficiency for both specialty and general- purpose computing. When Prodigy processors are provisioned in a hyperscale data center, they enable all AI, HPC, and general-purpose applications to run on one hardware infrastructure, saving companies billions of dollars per year. With data centers currently consuming over 4% of the planet’s electricity, predicted to be 10% by 2030, the ultra-low power Prodigy Universal Processor is critical to continue doubling worldwide data center capacity every four years. Tachyum, co-founded by Dr. Radoslav Danilak is building the world’s fastest AI supercomputer (128 AI exaflops) in the EU based on Prodigy processors. Tachyum has offices in the United States and Slovakia. For more information, visit https://www.tachyum.com/.

Tachyum Passes all Tests for FP Compliance with IEEE Standards

Tue, 20 Jun 2023 00:00:00 +0000

LAS VEGAS, June 20, 2023 – Tachyum™ announced today it has completed Floating-Point Unit (FPU) verification for Prodigy, the world’s first Universal Processor, having passed all tests to ensure Prodigy adheres to industry computation standards of Institute of Electrical and Electronic Engineers (IEEE). Prodigy passed the package of tests—more than 24,000,000 in all—in RTL simulation and on Prodigy FPGA prototype hardware.

Tachyum built its IEEE-compliant FPU from the ground up to be one of the most advanced in the world, performing at remarkably high clock speeds. Prodigy’s design was verified with the most recent version of the industry-standard Berkeley TestFloat, a widely used test for supercomputers that checks the fundamental soundness of a floating-point implementation against the standard Berkeley SoftFloat, testing all modes and special cases of the IEEE Standard. Tachyum will also perform continuous regression testing to ensure compliance with new releases.

“Establishing compliance with IEEE-prescribed FPU standards clears the way for final integration and testing and gives partners and customers additional assurance of Prodigy’s prowess,” said Dr. Radoslav Danilak, founder and CEO of Tachyum. “FPU computation is an essential foundation of modern computers and supercomputers, so this accomplishment is meaningful for future applications.”

Prodigy provides both the high performance required for cloud and HPC/AI workloads within a single architecture. As a Universal Processor offering utility for all workloads, Prodigy-powered data center servers can seamlessly and dynamically switch between computational domains. By eliminating the need for expensive dedicated AI hardware and dramatically increasing server utilization, Prodigy reduces CAPEX and OPEX significantly while delivering unprecedented data center performance, power, and economics. Prodigy integrates 128 high-performance custom-designed 64-bit compute cores, to deliver up to 4x the performance of the highest-performing x86 processors for cloud workloads, up to 3x that of the highest performing GPU for HPC, and 6x for AI applications.

A video showcasing the Prodigy FPU verification running on Prodigy FPGA hardware emulation system is available below.

Follow Tachyum

About Tachyum

Tachyum is transforming the economics of AI, HPC, public and private cloud workloads with Prodigy, the world’s first Universal Processor. Prodigy unifies the functionality of a CPU, a GPGPU, and a TPU in a single processor that delivers industry-leading performance, cost, and power efficiency for both specialty and general-purpose computing. When hyperscale data centers are provisioned with Prodigy, all AI, HPC, and general-purpose applications can run on the same infrastructure, saving companies billions of dollars in hardware, footprint, and operational expenses. As global data center emissions contribute to a changing climate, and consume more than four percent of the world’s electricity—projected to be 10 percent by 2030—the ultra-low power Prodigy Universal Processor is a potential breakthrough for satisfying the world’s appetite for computing at a lower environmental cost. Prodigy, now in its final stages of testing and integration before volume manufacturing, is being adopted in prototype form by a rapidly growing customer base, and robust purchase orders signal a likely IPO in late 2024. Tachyum has offices in the United States and Slovakia.

For more information, visit https://www.tachyum.com/.

Tachyum Among Recipients of 8.1 Billion Euro IPCEI Grant

Tue, 13 Jun 2023 00:00:00 +0000

LAS VEGAS, June 13, 2023 – Tachyum™ is one of 56 companies from 14 European Union countries selected to share in an up to 8.1 billion euro public grant as part of a recently announced Important Project of Common European Interest in Microelectronics and Communication Technologies (IPCEI ME/CT). As part of the funding, the European Commission has agreed with providing Tachyum with 26.4 million euros to accelerate the delivery of the Prodigy 2 Universal Processor.

The IPCEI ME/CT project supports research, innovation and the first industrial deployment of microelectronics and communication technologies across the value chain – from materials and tools to chip design and manufacturing processes. The IPCEI is part of a wider EC effort to ensure a greener, digital, more secure, resilient and sovereign European economy.

Tachyum is participating in IPCEI with the project “Accelerating Prodigy 2 HPC/AI, an EU processor,” which addresses a fundamental tenet of the multi-national program – the need for a competitive EU processor. This grant will accelerate Prodigy 2 processor for HPC/AI in a 3-nanometer fabrication process, which will enable exa-scale HPC and zetta-scale AI for Europe.

“Tachyum is pleased to be among the highly ambitious IPCEI projects chosen to help propel the EU beyond what the market currently offers and enable major improvements in the development of sovereign technologies,” said Dr. Radoslav Danilak, founder and CEO of Tachyum. “Strengthening Europe’s capacity to develop the next generation of processors and semiconductors is key to ensuring the region’s competitiveness while addressing key environmental and societal challenges.”

The IPCEI involves 68 projects from 56 companies. These direct participants will closely cooperate through more than 180 envisaged cross-border collaborations. In addition to the 8.1 billion euros in public funding, an additional 13.7 billion euros is expected in private investments. Additional information about the IPCEI ME/CT aid is available on the European Commission website

Follow Tachyum

About Tachyum

Tachyum is transforming the economics of AI, HPC, public and private cloud workloads with Prodigy, the world’s first Universal Processor. Prodigy unifies the functionality of a CPU, a GPGPU, and a TPU in a single processor that delivers industry-leading performance, cost, and power efficiency for both specialty and general-purpose computing. When hyperscale data centers are provisioned with Prodigy, all AI, HPC, and general-purpose applications can run on the same infrastructure, saving companies billions of dollars in hardware, footprint, and operational expenses. As global data center emissions contribute to a changing climate, and consume more than four percent of the world’s electricity—projected to be 10 percent by 2030—the ultra-low power Prodigy Universal Processor is a potential breakthrough for satisfying the world’s appetite for computing at a lower environmental cost. Prodigy, now in its final stages of testing and integration before volume manufacturing, is being adopted in prototype form by a rapidly growing customer base, and robust purchase orders signal a likely IPO in late 2024. Tachyum has offices in the United States and Slovakia. For more information, visit https://www.tachyum.com/.

Tachyum Testing Applications on Prodigy FPGA

Wed, 31 May 2023 00:00:00 +0000

LAS VEGAS, May 31, 2023 – Tachyum™ announced today it has achieved another milestone on the road to commercialization of Prodigy, the world’s first Universal Processor, as it begins testing key applications on Prodigy FPGA prototypes.

Having recently completed the porting and testing necessary for system-software like UEFI, BMC, bootloader, and operating systems through the toolchains, Tachyum has now completed porting of other applications to test a wide range of workloads on the FPGA prototype. When applications are verified to run cleanly on the Prodigy FPGA, the Prodigy chip will be ready for tape-out and volume manufacturing.

To accelerate application testing, Tachyum has invented and developed a new infrastructure for regression testing, to capture FPGA execution and bugs in the first instance. With this new system, engineers avoid the necessity of manually reproducing bugs, a very time-consuming process that can delay development schedules. This infrastructure is essential for bugs which occur after trillions of cycles, which would take a very long time to debug. Some bugs are so rare that they are very hard to reproduce, and sometimes cannot be manually reproduced. Catching bugs on their first occurrence is the key to getting the chip ready for production quickly. Engineers can take traces from the Prodigy FPGA prototype, run trillions of cycles in a C-model called “cyclerun“ and then feed the state from the C-model to the RTL simulation, which can run tens of millions of cycles, or more, before the bug manifests in RTL. The engineers can then debug in the RTL simulation, instead of using a very time-consuming signal tap on the FPGA.

“Validating applications necessarily follows completing the physical design of the chip and will help us verify Prodigy’s performance in customer and partner deployments,” said Dr. Radoslav Danilak, founder and CEO of Tachyum. “The pace at which our team is hitting milestones is a model for the semiconductor industry and shows our laser-focus on reaching tape-out and ultimately transforming the economics of data center computing.”

Tachyum has also verified Prodigy’s design against the most recent version of the industry-standard Berkeley TestFloat/SoftFloat package.

GNU Compiler Collection (GCC) and toolchain are already running successfully on Prodigy FPGAs. For a video demonstrating GCC and toolchain on the Prodigy FPGA, and automated QA testing, visit our YouTube channel or watch the video below.

Follow Tachyum

About Tachyum

Tachyum is transforming the economics of AI, HPC, public and private cloud workloads with Prodigy, the world’s first Universal Processor. Prodigy unifies the functionality of a CPU, a GPGPU, and a TPU in a single processor that delivers industry-leading performance, cost, and power efficiency for both specialty and general-purpose computing. When hyperscale data centers are provisioned with Prodigy, all AI, HPC, and general-purpose applications can run on the same infrastructure, saving companies billions of dollars in hardware, footprint, and operational expenses. As global data center emissions contribute to a changing climate, and consume more than four percent of the world’s electricity—projected to be 10 percent by 2030—the ultra-low power Prodigy Universal Processor is a potential breakthrough for satisfying the world’s appetite for computing at a lower environmental cost. Prodigy, now in its final stages of testing and integration before volume manufacturing, is being adopted in prototype form by a rapidly growing customer base, and robust purchase orders signal a likely IPO in late 2024. Tachyum has offices in the United States and Slovakia. For more information, visit https://www.tachyum.com/.

Tachyum AI Mastered FP8 To Reach FP32 Precision in Updated Version of AI White Paper

Wed, 17 May 2023 00:00:00 +0000

LAS VEGAS, May 17, 2023 – Tachyum™ today released the second edition of the “Tachyum Prodigy on the Leading Edge of AI Industry Trends” whitepaper featuring updates such as the implementation of improved 8-bit floating point (FP8) quantization aware techniques with adaptive scaling achieving 32-bit floating point (FP32) accuracy.

In the white paper, Tachyum demonstrates the optimality of the FP8 format for quantizing deep neural networks including weights, activations and gradients, exploiting the fact that floating point numbers provide better coverage than 8-bit integer data type. The results show that FP8 quantized networks can maintain accuracy on par – or even exceed the accuracy – of baseline FP32 models.

FP8 is essential for being able to do more with less. It achieves much higher performance at much lower power consumption and chip area than legacy technology like BFLOAT16. FP8 not only reduces the cost of computation but also memory requirements for large and rapidly growing AI models. The white paper features analysis and quantization errors for different models and datasets as well as how Tachyum amplifies the benefits of FP8 in terms of twice the performance, power efficiency and bandwidth reduction.

Visual models, large language AI models and generative AI are increasingly being included in a number of software applications, making AI an essential part of data processing and requiring tighter and lower latency integration into software. With FP8 capable of performing mainstream AI functions, leaders like Tachyum, are poised to help accelerate the rapid evolution of AI hardware technology. This will lead to a unification of specialized HPC and AI hardware modules into a single processing engine rather than integrating disparate chips into one package, which is a more costly and less satisfactory solution.

“Our experimental results show that FP8 enables faster training and reduced power consumption without any degradation in accuracy for a range of deep learning models,” said Dr. Radoslav Danilak, founder and CEO of Tachyum. “This is one of the most significant AI milestones Tachyum wanted to achieve before tape-out to ensure we have what it takes to make FP8 with sparsity and super-sparsity mainstream AI technology.”

After achieving the milestone of completing infrastructure around FP8 to achieve nearly the same precision as FP32 for training and inference, Tachyum is shifting its engineering focus to a yet-to-be-announced Tachyum AI (TAI) infrastructure, which provides the next level of AI beyond FP8. TAI is part of Tachyum’s current hardware and will be introduced later this year with results obtained from today’s mainstream AI applications.

To read more about Prodigy’s capabilities in AI, including results from its implementation of FP8, interested parties can download Tachyum’s latest white paper.

Follow Tachyum

About Tachyum

Tachyum Completes Porting of Software for Prodigy Tape-Out

Tue, 02 May 2023 00:00:00 +0000

LAS VEGAS, May 2, 2023 – Tachyum™ today announced its latest milestone of officially entering the final phase of test and development for the Prodigy Universal Processor. This last stage of Quality Assurance (QA) testing of all necessary ported software on a Field Programmable Gate Array (FPGA) for final testing will ensure the chip is production worthy.

Prodigy is currently running ported software on its software emulation platform. Once all Quality Assurance testing is successfully completed on FPGA, the Prodigy chip will be ready for tape-out this year. This phase of porting and finalizing all necessary software for tape out will ensure that existing applications run seamlessly and deliver industry-leading performance for hyperscale, high-performance computing and artificial intelligence workloads once the Prodigy chip is received from the fabrication house.

Among the porting milestones that have already been completed including but not limited to the following software packages:

Ported and demonstrated Open BMC
Ported and demonstrated UEFI v2.70
Ported and integrated boot loaders system: EFI boot and GRUB 2
Running Linux 6.1 LTS with SE Linux, Free BSD 13, KVM 6.1, QEMU 7.2
GNU toolchain: binutils 2.40, GDB-13.1, GCC 12.2 and glibc 2.37
Support of C, C++, Fortran, Go, Erlang, Lua, Perl, PHP, R, Python, Ruby, Tcl
TensorFlow 2.11 and PyTorch 1.11 compilers
Docker and Kubernetes
Eigen, GEMM, NumPy (Numerical Python) libraries
Fully representative list of applications such as databases, web servers, SLURM, CEPH

Prodigy delivers unprecedented data center performance, power, and economics, reducing CAPEX and OPEX significantly. Because of its utility for both high-performance and line-of-business applications, Prodigy-powered data center servers can seamlessly and dynamically switch between workloads, eliminating the need for expensive dedicated AI hardware and dramatically increasing server utilization. Tachyum’s Prodigy integrates 128 high-performance custom-designed 64-bit compute cores, to deliver up to 4x the performance of the highest-performing x86 processors for cloud workloads, up to 3x that of the highest performing GPU for HPC, and 6x for AI applications.

“We want to declare this latest milestone as we are moving from the development phase to completion of the tasks needed for tape-out, of which the completion of the software porting is key,” said Radoslav Danilak, founder and CEO of Tachyum. “We are diligently working towards closing this phase of our production cycle by ensuring that the porting and preparation of all the necessary software for tape-out is behind us and this chapter is closed. I am very excited about the progress that Tachyum has made thus far and look forward to entering the next phase of success.”

Follow Tachyum

About Tachyum

Tachyum Chooses SENKO’s SN Connectivity for Prodigy System

Tue, 25 Apr 2023 00:00:00 +0000

LAS VEGAS, April 25, 2023 – Tachyum™ has entered into an agreement with Senko Advance Co., Ltd. for Optical Interconnect Solution for Prodigy® systems enabling high-speed data transfer in data centers. The criterion for selection was the robustness, reliability, and versatility of the SN® (Simplified Network) connectivity solutions.

SENKO’s SN connectivity solutions are known for their high data rates and robustness. The SN breakout solution is designed to provide high-speed data transmission at rates of up to 100 Gbps per port. The SN breakout solution is an ideal solution for high-performance computing, data center, and cloud applications that require high data rates, low latency, and high reliability.

“We are proud to have been chosen by Tachyum to provide the high-speed connectivity required by its Prodigy system,” said David Jenkins, Director of Product Management at SENKO. “Our SN connectivity solutions are designed to provide robust and reliable performance for the most demanding data center and cloud applications. The SN breakout solution is a cost-effective and versatile solution that enables easy migration to high-speed 100 Gbps networks, providing increased flexibility and scalability for our customers.”

SENKO’s SN breakout solution is designed to enable easy migration from existing 10 Gbps or 25 Gbps networks to high-speed 100 Gbps networks. The solution supports breakout cables that enable a single 400 Gbps port to be split into four 100 Gbps ports, providing increased flexibility for switch-to-switch connectivity in the data center and cloud applications.

“We chose SENKO’s SN connectivity solutions for their high data rates, reliability, and versatility. It enabled Tachyum to develop its own networking infrastructure to support HPC/AI deployments for the Prodigy Universal Processor. Tachyum’s flattened networking infrastructure is optimized for performance, cost, efficiency, and ease of maintenance,” said Pini Herman, Senior Director of Solutions Engineering of Tachyum. “The Prodigy system requires a high-speed and reliable connectivity solution to deliver the high-performance computing required by modern data centers and AI applications. SENKO’s SN connectivity solutions meet this requirement and enable us to deliver unprecedented computing performance at a fraction of the power consumption of existing solutions.”

Follow Tachyum

About Tachyum

About Senko

Senko Advanced Components is a wholly owned subsidiary of SENKO Advance, which is headquartered in Yokkaichi, Japan. With 16 locations globally and dozens of design and manufacturing facilities providing local support to customers all around the globe. SENKO was incorporated in the United States in the early nineties and has since been recognized as one of the industry’s specialists in passive fiber optic interconnects and optical components. To date, SENKO has deployed over 800 million connectors. 150+ patents awarded, with more than 300 patents pending. For more information, visit https://www.senko.com/.

Tachyum Prodigy is Exceeding Original Design Goals With New EDA Tools

Tue, 18 Apr 2023 00:00:00 +0000

LAS VEGAS, April 18, 2023 – Tachyum™ announced today that with Register Transfer Level (RTL) design wrapping up, the Tachyum design team is now focused on completing the physical design for Prodigy, the world’s first Universal Processor.

An RTL design validates how data flows within the chip’s integrated circuits and addresses levels of inputs and controls. New CAD tools are enabling Tachyum to work on completing the physical design (the actual placement of transistors and wires). With the transition from RTL to physical design, the Prodigy processor is approaching chip layout freeze, prior to tape out this year.

The Prodigy design achieves unprecedented processing density, allowing the design team to increase from 64 to 96 silicon IP cores (112G/PCIE 5.0 SerDes) on a single Prodigy chip. Tachyum engineers are now focusing on detailed physical design.

“Thanks to our engineers’ superhuman efforts to quickly adapt to using new physical design toolsets, even after replacing the design IP and the physical design EDA software, we have now delivered the new physical design of the chip with the new tools in record time, exceeding performance targets for Prodigy,” said Dr. Radoslav Danilak, founder and CEO of Tachyum. “This is a critical milestone before chip layout freeze — providing a frozen physical design of the chip, to our partners to complete substrate, package, and socket design — prior to tape-out and volume production.”

Follow Tachyum

About Tachyum

Tachyum Unveils 20 Exa-Flop / 10 AI Zetta-Flop Supercomputer Design

Wed, 12 Apr 2023 00:00:00 +0000

LAS VEGAS, April 12, 2023 – Tachyum™ today published a new white paper presenting HPC and AI supercomputer data center designs using the Prodigy Universal Processor Family, Prodigy and Prodigy 2.

Tachyum Prodigy 2 was selected by Important Project of Common European Interests (IPCEI) program for Slovakia to deliver exa-scale HPC and zetta-scale AI for Europe. European Commission has accepted the funding gap of 26.4 million EUR for Tachyum, which is currently in the notification process.

Developed by Tachyum’s systems, solutions, and software engineering teams, these reference designs transform data centers into universal computing centers in which HPC and AI workloads can run on the same architecture. Tachyum has developed thorough data center designs incorporating state-of-the-art solutions for computing, networking, storage, software, and cooling to address the next generation of HPC/AI applications.

The designs are based on Tachyum’s submission in response to a U.S. Department of Energy RFI for supercomputing systems for both HPC and AI. Data center designs based on Prodigy 2, expected to be available in 2025, will deliver 20 EF of FP64 vector performance, the top of the DOE’s target, within the target 60MW power envelope in a footprint of 6,000 square feet.

This same data center can deliver over 10 ZF of AI performance, more than 30x the top end of the DOE’s target. FP64 performance up to 25 EF can be achieved within the same 6,000 square-foot area, depending on the flexibility of the power envelope.

Prodigy 2 data centers include next-generation infrastructure for networking and storage to utilize the industry’s most advanced technology and ensure that the entire cluster runs at maximum performance and efficiency.

First-generation Prodigy data centers will offer 3.3 EF of FP64, three times the HPC performance of existing supercomputers, and also deliver around 1.8 ZF of AI processing in just 6,000 square feet at power consumption of only 45 MW.

“Tachyum provides leading-edge solutions from silicon to complete systems to address the ever-increasing demands for both HPC and AI,” said Radoslav Danilak, founder and CEO of Tachyum. “Tachyum-designed supercomputers push the forefront of HPC performance while crossing the zetta-scale barrier for AI, transforming data centers into universal computing centers.”

Those interested in learning more about Tachyum’s data center designs for HPC and AI can download Tachyum Prodigy Universal Processor Enabling Next Generation 20+EF Supercomputers.

Prodigy delivers unprecedented data center performance, power, and economics, reducing CAPEX and OPEX significantly. Because of its utility for both high-performance and line-of-business applications, Prodigy-powered data center servers can seamlessly and dynamically switch between workloads, eliminating the need for expensive dedicated AI hardware and dramatically increasing server utilization. Tachyum’s Prodigy integrates 128 high-performance custom-designed 64-bit compute cores, to deliver up to 4x the performance of the highest-performing x86 processors for cloud workloads, up to 3x that of the highest performing GPU for HPC, and 6x for AI applications.

Follow Tachyum

About Tachyum

Tachyum To Use UCIe Interconnect Standards In Prodigy 2

Tue, 28 Mar 2023 00:00:00 +0000

LAS VEGAS, March 28, 2023 – Tachyum™ announced today it has strengthened its presence in the Universal Chiplet Interconnect Express (UCIe) organization that develops and supports the chiplet ecosystem and in-package innovations. The UCIe community fosters collaboration among leaders in semiconductors, packaging, IP suppliers, foundries, and cloud services.

UCIe 1.0 is an open specification guiding the interconnection between chiplets within a package to ensure communication and functionality. Chiplet-based design increases density and capability compared to standard printed circuit-boards (PCBs), and enables much smaller, lower-cost solutions that consume less power.

UCIe specification addresses the die-to-die I/O physical layer, die-to-die protocols, and software stack currently served by PCI Express® (PCIe®) and Compute Express Link™ (CXL™) industry standards. UCIe ultimately extends the PCIe standard to in-package interconnects and allows the potential to bring CXL into in-package integration.

Tachyum will be using UCIe chip-to-chip interconnect in the development of its second-generation Prodigy processors, expected to be deployed in Tachyum-enabled products for edge applications. As part of the emerging chiplet ecosystem, Prodigy’s adherence to UCIe standards means developers can mix and match ideal components from multiple vendors and be assured of compatibility.

“In-package integration with UCIe can revolutionize the industry and overcome technical barriers by allowing unprecedented miniaturization of consumer devices, edge, and IoT,” said Dr. Radoslav Danilak, founder and CEO of Tachyum. “Since first joining UCIe last year, we’ve recognized the importance of this collaborative work for advancing the state of the art in semiconductors, and naturally took the opportunity to increase our level of participation to a higher tier.”

Prodigy delivers unprecedented data center performance, power, and economics, reducing CAPEX and OPEX significantly. Because of its utility for both high-performance and line-of-business applications, Prodigy-powered data center servers can seamlessly and dynamically switch between workloads, eliminating the need for expensive dedicated AI hardware and dramatically increasing server utilization. Tachyum’s Prodigy integrates 128 high-performance custom-designed 64-bit compute cores, to deliver up to 4x the performance of the highest-performing x86 processors for cloud workloads, up to 3x that of the highest performing GPU for HPC, and 6x for AI applications.

Follow Tachyum

About Tachyum

Tachyum Presents Software Stack for HPC/AI and Supercomputer Management

Tue, 21 Mar 2023 00:00:00 +0000

LAS VEGAS, March 21, 2023 – Tachyum™ today released a white paper, “Tachyum HPC/AI Software Stack Overview,” detailing the software packages offered by the company and supported by its software team that enable software developers to build HPC applications deployed as part of Prodigy®-empowered data centers.

Tachyum’s software designs help meet the high demands of cloud and HPC/AI workloads with Prodigy. The company’s software team cooperates with early customers and partners to create software packages ideally suited for HPC environments primarily using open-source software. With its software stack, Tachyum supports all key software libraries required for HPC applications and addresses current challenges and opportunities.

Tachyum software stack components are comprised of:

Tachyum firmware: BIOS (UEFI), BMC
Operating system level: LINUX Kernel, SELinux
Storage and file-systems options: Ceph
OS level virtualization: Docker and containerd, Podman based on Kubernetes pods
Communication stack and runtimes: OpenMPI, MPICH, gPRC, RabbitMQ
User space utilities and scripting: Tachyum Linux distribution provides all the necessary utilities and scripting languages like Bash, SED, Gawk, Grep, Wget, Rsync, Strace, Git, Subversion, PHP, Perl, Ruby, Tcl, Top, Htop, Iotop, Sar, Nc.
Parallelism standards: OpenMP, SHMEM
Programming languages and compilers: Tachyum has ported GNU gcc compiler and go compiler.
Debugging tools: GDB, KGDB, OpenOCD
Emulation tools: Tachyum has developed multiple emulators like GDB based emulator for internal use and QEMU based emulator for customers to translate X-86, ARM and RISC-V software to Tachyum’s Prodigy architecture.
Clustering, management, and job scheduling: SLURM for jobs management, Zabbix and Prometheus with Grafana for monitoring, Kubernetes for automating software deployment, scaling and management of containerized applications, Ansible for IT automation.
Networking capabilities: Tachyum has developed its own networking infrastructure to support HPC/AI deployments for Prodigy. SONIC was chosen as the network OS.
AI/HPC frameworks and libraries: BLAS, GEMM, NumPy, Eigen, PyTorchTensorFlow
AI applications: Resnet and ConvMixer for computer vision, SSD and Yolo for object detection, Fast-SCNN and UNET for instance segmentation, BERT for NPL
HPC applications: Quantum espresso, DeepMD, LAMMPS

“Multiple different processing technologies are combined in today’s HPC systems,” said Dr. Radoslav Danilak, founder and CEO of Tachyum. “Achieving the required performance and functionality at an affordable power envelope places a lot of challenges to HPC software environments – especially on scheduling and resource management. Tachyum’s Prodigy, the world’s first Universal Processor, integrates the functionality of CPU, GPGPU and TPU into a single homogeneous architecture so that there is no additional complexity needing to be solved on the software level between hardware components.”

Those interested in reading the “Tachyum HPC/AI Software Stack Overview” white paper can download a copy.

Follow Tachyum

https://twitter.com/tachyum
https://www.linkedin.com/company/tachyum
https://www.facebook.com/Tachyum/

About Tachyum

Tachyum Hosts Slovak President to Showcase Prodigy Innovation

Fri, 03 Mar 2023 00:00:00 +0000

LAS VEGAS, March 3, 2023 – Tachyum™ today announced that Slovak President Zuzana Čaputová visited its Bratislava offices to witness first-hand the progress of the Prodigy Universal Processor, which is in the final stage of development. President Čaputová met with Tachyum founder and CEO Radoslav Danilak and a team of engineers to discuss topics such as innovation, green economy, sustainability and attracting top regional talent.

“If we want to start catching up with the living standard in western European countries again, we need to develop cutting-edge technologies in Slovakia. We need companies with the ambition to push the boundaries of what is possible and to succeed on the world market. It is great that we already have companies with this vision in Slovakia,” said Zuzana Čaputová, President of the Slovak Republic.

The Slovakian government has made it a priority to transform the economy with pioneering technologies that align with global megatrends. The country’s economic development strongly encourages tech innovation and high value-added industries. Tachyum is committed to supporting a competitive and innovative Slovakia through the introduction of the world’s first universal processor.

“It was an honor to showcase Prodigy’s capabilities to Slovakian officials today, and demonstrate how our universal processor will provide an immediate impact to the world’s data centers by enabling efficient processing of increasingly needy applications,” said Danilak. “Tachyum is well positioned to help with the transformation of the economy into a sustainable one in large part through enabling more efficient use of resources. With a mantra of “do more for less,” Prodigy allows companies and governments to dramatically reduce electricity consumption, helping to control spiraling electricity costs and reducing the need to continually build power plants.”

With data centers currently consuming more than four percent of the planet’s electricity – predicted to be 10 percent by 2030 – the ultra-low power Prodigy Universal Processor is critical to minimizing the environmental impact, including emissions, of data centers.

During the visit, Tachyum representatives showed president Čaputová how Prodigy-powered supercomputers are poised to address some of the world’s most important and challenging problems. Faster and sustainable computation speed for AI is considered critical for advancement in all aspects of human life – from faster drug discovery and vaccine development to climate change impact assessment, weather forecasting, personalized medicine and more.

Tachyum has a huge potential to become a magnet for highly qualified world-class engineers and scientists. After Tachyum launched its Prodigy into the worldwide marketplace, it anticipates the company will be able to attract an even greater need for more talented individuals to join its world class team.

Follow Tachyum

About Tachyum

Tachyum Enters Kingdom of Saudi Arabia with Watad Energy & Communications Partnership

Tue, 21 Feb 2023 00:00:00 +0000

LAS VEGAS, February 21, 2023 – Tachyum™ today announced it has entered into a memorandum of understanding (MOU) agreement with Watad Energy & Communications Co. ltd, the leading OT Cybersecurity and AI Company in the KSA and GCC (Gulf Cooperation Council) region, to bring Prodigy technology to commercial, industrial and scientific research institutions throughout the Kingdom of Saudi Arabia and the GCC region. Prodigy, the world’s first universal processor, will enable high-performance computing (HPC), artificial intelligence (AI), edge computing, cloud and hyperscale data centers, and other IT segments and applications that demand cutting-edge performance.

Watad Energy & Communications Co. ltd (“WATAD”), implements and maintains tailored solutions for businesses operating in ICT, Telecommunications, General Contracting, Government, Defense & Security, Oil & Gas, Real Estate, Hospitality and the Industrial sectors. While these are the main sectors, the solutions are flexible enough to meet the demands of any business in the public or private sector. WATAD further provides a wide range of IT value-added services such as System Integration, IT Programme & Project Management, IT Procurement, IT Strategy & Architecture and IT Consulting.

Tachyum and WATAD plan to work together to advance an open platform for AI and Big Data, as well as support innovations in relevant industry sectors and engage with other leaders in the field in KSA and GCC region. Both companies expect these efforts to yield opportunities for high-tech investors and regional economies including employment growth, and more profitable research.

“Tachyum Prodigy processors can move the AI to a different level, and reach a new era, which will be a great value nowadays to the ambitious projects happening in KSA,” said Mohammed Almutairi, MD of Watad Energy & Communications Co. ltd.

Tachyum’s Prodigy integrates 128 high-performance custom designed 64-bit compute cores with the functionality of a CPU, a GPU, and a TPU in a single device within a homogeneous architecture. This allows Prodigy to deliver performance up to 4x that of the highest performing x86 processors (for cloud workloads) and up to 3x that of the highest performing GPU for HPC and 6x for AI applications, at the same power consumption.

Prodigy’s unique architecture delivers industry-leading performance in both data center and AI workloads. Therefore, during off peak hours, Prodigy-powered data center servers can be seamlessly and dynamically switched to AI workloads, eliminating the need for expensive dedicated AI hardware, and dramatically increasing server utilization. Prodigy delivers unprecedented data center performance, power, and economics, reducing CAPEX and OPEX significantly.

“WATAD is an ideal partner for Tachyum, with the kits’ designing and building ability as well as market knowledge, position it as a great system integrator for Tachyum Prodigy for interesting projects happening nowadays in KSA,” said Dr. Radoslav Danilak, founder and CEO of Tachyum. “Both KSA and the GCC region are among the world’s leaders in the implementation of new technologies and solutions, and we believe that our chips will be a great value added.”

Follow Tachyum

https://twitter.com/tachyum
https://www.linkedin.com/company/tachyum
https://www.facebook.com/Tachyum/

Follow WATAD

https://www.twitter.com/watadenergy
https://www.linkedin.com/company/watad-energy
https://www.watadenergy.com

About Tachyum

About Watad

Watad is a fast-growing information and communications technology company that prides itself on providing cutting-edge IT/OT/ICS Cybersecurity, Artificial Intelligence, Machine Learning, Industrial Automation and Smart City solutions.

At Watad, we design, develop, implement and maintain tailored solutions for businesses operating in ICT, Telecommunications, General Contracting, Government, Defense & Security, Oil & Gas, Real Estate, Hospitality and the Industrial sectors. While these are the main sectors we currently work with, our solutions are flexible enough to meet the demands of any business in the public or private sector. We further provide a wide range of IT value-added services such as System Integration, IT Programme & Project Management, IT Procurement, IT Strategy & Architecture and IT Consulting.

Tachyum Validates Prodigy Universal Processor with Kubernetes for High-Performance, High-Density Computing for Containers

Tue, 14 Feb 2023 00:00:00 +0000

LAS VEGAS, February 14, 2023 – Tachyum™ today announced it has completed testing and validation of its Prodigy Universal Processor with Kubernetes for hyperscale, high-performance container management. The success of Kubernetes on Prodigy assures Tachyum customers and partners of quick, easy, out-of-the-box testing and evaluation in containerized environments.

Tachyum’s software team deployed its Prodigy emulation with K3s server, a lightweight Kubernetes distribution ideal for testing; two agents; and NGINX, an open-source reverse proxy server for HTTP, HTTPS, SMTP, POP3, and IMAP protocols, as well as a load balancer, HTTP cache, and a web server.

“Demonstrating that Kubernetes can run natively on Prodigy is essential to customers and partners, such as those using Kubernetes behind performance-sensitive containerized applications,” said Radoslav Danilak, founder and CEO of Tachyum. “Tachyum’s growing software ecosystem is addressing the needs of an increasingly wide range of data center, hyperscale, and high-performance workloads that will benefit from Prodigy today and tomorrow.”

Clustering, management and job scheduling in containerized environments can be difficult at scale. Prodigy-based data centers will find increased ease and performance in automating deployment, scaling, and management of Kubernetes clusters, which can span across hosts on-premise, public, private, or hybrid clouds, scaling from one compute node up to massive deployments.

A video demonstrating Kubernetes on Prodigy is available below.

Prodigy delivers unprecedented data center performance, power, and economics, reducing CAPEX and OPEX significantly. Because of its utility for both high-performance and line-of-business applications, Prodigy-powered data center servers can seamlessly and dynamically switch between workloads, eliminating the need for expensive dedicated AI hardware and dramatically increasing server utilization. Tachyum’s Prodigy integrates 128 high-performance custom-designed 64-bit compute cores, to deliver up to 4x the performance of the highest-performing x86 processors for cloud workloads, up to 3x that of the highest performing GPU for HPC, and 6x for AI applications.

Follow Tachyum

About Tachyum

Tachyum Unveils Air Defense Superiority Using Prodigy AI White Paper

Tue, 07 Feb 2023 00:00:00 +0000

LAS VEGAS, February 7, 2023 – Tachyum™ today released details of Prodigy®, the world’s first universal processor, for advanced military and defense avionics applications in a new white paper “Air Dominance Powered by Prodigy“.

AI is fast becoming an indispensable part of military aircraft, and next-generation fighter fleet across the world will require massive processing power to fuse sensor data into actionable information. The critical lynchpin for air-to-air and air-to-ground combat effectiveness is superior onboard computational capability with state-of-the-art processors to fuel rapidly evolving conventional and AI algorithms.

The new white paper explains Prodigy’s performance benefits to onboard computers and the algorithms they run, and its cost advantages in a challenging economic environment. As a universal processor, Prodigy enables both high-speed flight control, and AI-driven sensor fusion and decision-cueing, based on battlespace information optimally presented to the pilot, with one platform.

“While the U.S. Department of Defense funds new airframes, onboard avionics with advanced computational capabilities are considerably cheaper to implement and are a more substantial, immediate benefit to mission performance,” said Dr. Radoslav Danilak, founder and CEO of Tachyum. “This white paper is a call to arms for how defense and aerospace platform manufacturers across the world, including EU, can provide cost-effective, high-impact sensor and onboard computer upgrades that will restore military air dominance.”

The Defense Advanced Research Projects Agency (DARPA) is currently running the Air Combat Evolution (ACE) program, one of more than 600 Department of Defense projects aimed at incorporating AI into combat systems including future autonomous/pilotless fighters. DARPA expects Explainable AI—in which AI decision-making is provable, deterministic, and can be reconstructed—to be essential to future national security projects. Prodigy processors offer particularly high performance in Explainable AI algorithms, which are more CPU-intensive than other AI/ML algorithms, and not well suited for existing AI systems.

With Airbus, Dassault and other key partners having announced agreement on their next-generation European fighter program, Tachyum’s Prodigy processor is optimally positioned to add value to this critical plan. Prodigy will significantly enhance the EU Fighter’s operational performance with its breakthrough computational capabilities, including the best AI performance (training and inference) on the planet. Tachyum will leverage its European presence and reputation to insure, that the EU maintains a decisive air dominance advantage, during this increasingly dangerous geopolitical period in Europe.

Prodigy, a 128-core processor running at up to 5.7GHz, unifies the functions of a CPU, GPGPU and TPU in a single, homogeneous processor architecture, and delivers massive performance for far less cost than competing technologies. Prodigy was designed to handle matrix and vector processing from the ground up, rather than as an afterthought. Because of its utility for both high-performance AI and standard applications, Prodigy- eliminates the need for expensive dedicated AI computing architectures. Tachyum’s Prodigy can deliver up to 6x the performance of the highest-speed x86 processors for AI applications, 4x for cloud workloads, and up to 3x the highest-performing GPU for HPC applications.

Follow Tachyum

About Tachyum

Tachyum Successfully Runs UEFI on Prodigy FPGA

Tue, 31 Jan 2023 00:00:00 +0000

LAS VEGAS, January 31, 2023 – Tachyum™ today announced that it has successfully demonstrated Unified Extensible Firmware Interface (UEFI) running on the Prodigy Field-Programmable Gate Array (FPGA) hardware emulation system, representing a critical piece of the company’s test strategy and ensuring that Tachyum’s Prodigy hardware is maturing on pace for production later this year.

Replacing legacy BIOS, UEFI is a next-generation interface between operating systems and platform firmware. UEFI was developed to allow support for new technologies during the boot process before the OS loads. Supported bootloaders for Prodigy include Systemd-Boot and GRUB2.

UEFI requires non-cacheable access that has been debugged. As part of its recent testing, Tachyum successfully completed the boot flow from UEFI firmware to Linux OS. By starting execution of the Prodigy processor in non-cacheable mode, UEFI running on the actual Prodigy design proves readiness of components integrated into the software stack.

Together with Baseboard Management Controller (BMC), Tachyum now has core pieces for motherboard and system manufacturers at the ready. Customers will have the opportunity to access Prodigy’s source code so that they can create their own UEFI. This advancement in status indicates that Tachyum’s Prodigy hardware is rapidly maturing for release.

“Every effort is being made to ensure that Prodigy will run optimally upon its release later this year, taking each step of the testing process one by one to validate its readiness,” said Dr. Radoslav Danilak, founder and CEO of Tachyum. “By achieving the completion of our UEFI development and making it readily available to developers, we are advancing our expectations that Prodigy will deliver the performance, cost and power consumption benefits needed to tackle the most-demanding data center operations of hyperscale, high-performance computing and artificial intelligence workloads.”

Prodigy, the world’s first Universal Processor, delivers a revolutionary new architecture that unifies the functionality of CPU, GPGPU, and TPU into a single monolithic device, and dynamically reallocates server resources to maximize utilization without expensive and power-hungry accelerators. Tachyum has provided ideal server, cabinet, networking, and storage designs to help meet the high demands of cloud and HPC/AI workloads with Prodigy.

A video demonstrating UEFI with an option for an Extensible Firmware Interface (EFI)-stub extension is available below. With this option, it’s possible to build the Linux kernel as a PE/COFF or portable executable common object file format image and run it directly from the UEFI shell as a standard EFI executable binary.

Follow Tachyum

About Tachyum

Tachyum Releases Native Storage Solutions for Prodigy Based Supercomputers and Data Centers

Tue, 24 Jan 2023 00:00:00 +0000

LAS VEGAS, January 24, 2023 – Tachyum™ today released a white paper “Tachyum Storage Strategy and Direction for Prodigy Data Center Deployments” detailing a storage strategy to maximize system performance and efficiency for Prodigy® data center scalable deployments while minimizing cost and management overhead.

The white paper showcases Tachyum’s system-level storage expertise with a storage networking architecture that offers flexibility and scalability in capacity and performance across any data center. The white paper includes options with both HDD and SSD-based solutions to address a broad range of customer demands.

“Tachyum’s Prodigy Universal Processor provides a revolutionary high-performance processor architecture that delivers unprecedented performance and TCO savings, and using the recommended storage equipment and technologies will maximize cluster performance for both HPC/AI and cloud deployments,” said Dr. Radoslav Danilak, founder and CEO of Tachyum. “Tachyum’s world-class systems and solutions engineering teams have guaranteed optimal Prodigy performance and cost savings by developing server, cabinet, networking, and storage designs that are tailored to Prodigy, ensuring that Prodigy’s groundbreaking new CPU architecture will have the best possible system solutions for Prodigy clusters, including upcoming supercomputer deployments. In addition, by offering cutting-edge solutions for both HDDs and SSDs, Tachyum’s solutions are future-proofed from the market dynamics in the storage industry as SSDs and HDDs converge to price parity.”

The suggested storage subsystems for Prodigy adopt a storage cluster approach, using state-of-the-art storage technologies:

HDD modular clusters, each comprised of a 2U server and a 5U JBOD for a total of 7U, with 7 of the 7U clusters fitting into a 52U rack storing 11.8 PB of capacity using 20TB drives
SSD modular clusters, comprised of 1U or 2U storage servers with SATA or NVMe drives; 24 2U servers can fit into a 52U rack storing 17.97PB of capacity using 15.36TB SATA SSDs

Tachyum has included recommendations for data center footprints of 4000 and 6000 square feet that include storage racks capable of supporting any of the above technologies, and can be configured based on the organization’s requirements.

Prodigy compute nodes run Ceph natively on Prodigy, and leverage open-source Ceph’s scalability, fault tolerance, ability to natively support block, file, and object storage. In addition to Ceph, Prodigy supports a wide range of popular file systems running natively as a client including NFS, CIFS/Samba, S3, ZFS, GFS2, and PNFS to address a wide array of customer and market requirements.

Those interested in learning more about Tachyum’s storage strategy for Prodigy data center deployments can download Storage Strategy and Direction for Prodigy Data Center Deployments.

A video demonstration of Ceph client running on Prodigy is available below.

Follow Tachyum

About Tachyum

Settlement Reached in Lawsuit Brought by Tachyum Against Cadence

Tue, 17 Jan 2023 00:00:00 +0000

LAS VEGAS, January 17, 2023 – Tachyum™ today announced that it has reached an amicable settlement with Cadence Design Systems, Inc. in a lawsuit brought by Tachyum on June 8, 2022, in Santa Clara County Superior Court in California. The matter is now considered ended, as all issues have been resolved. Details of the settlement will remain undisclosed.

With the case now settled, Tachyum will continue its work towards reaching production-ready status of its Prodigy universal processor unabated. The company recently achieved its latest milestone of running LINPACK benchmarks using Prodigy’s Floating-Point Unit (FPU) on a Field Programmable Gate Array (FPGA). The next milestone to be achieved is running vector operations, including mask operations and operations of unaligned vectors. The vectorization in the compiler reaching the production stage and vectorizing compilers and vectorized libraries will be fully available before chip shipments this year.

Prodigy delivers unprecedented data center performance, power, and economics, reducing CAPEX and OPEX significantly. Because of its utility for both high-performance and line-of-business applications, Prodigy-powered data center servers can seamlessly and dynamically switch between workloads, eliminating the need for expensive dedicated AI hardware and dramatically increasing server utilization. Tachyum’s Prodigy integrates 128 high-performance custom-designed 64-bit compute cores, to deliver up to 4x the performance of the highest-performing x86 processors for cloud workloads, up to 3x that of the highest performing GPU for HPC, and 6x for AI applications.

Those interested in learning more about the unique capabilities of Prodigy can download the full Prodigy Architectural Overview White Paper PDF. The paper unveils architectural characteristics of the system-on-chip (SoC), including chip layout, and the architecture of the world’s first universal processor. More details about Prodigy AI features can be found in Tachyum Prodigy on the Leading Edge of AI Industry Trends White Paper PDF.

Tachyum has developed networking infrastructure to support HPC/AI deployments for the Prodigy Universal Processor. Those interested in learning more about Tachyum’s flat networking for supercomputers can download a white paper detailing the trends in high performance networking technologies.

Follow Tachyum

About Tachyum

Tachyum Closes DDR5 Timing at over 6400MT/s Providing Massive Bandwidth for Prodigy Chip

Tue, 10 Jan 2023 00:00:00 +0000

LAS VEGAS, January 10, 2023 – Tachyum™ today announced that the IP components – DDR5 RAM controller and high-performance, low-power DSP-based PHY – incorporated into the Prodigy Universal Processor have allowed it to achieve speeds of 6400 MT/s at nominal voltage for Prodigy chip which provides headroom for expected speeds of up to, or even over, 7200 MT/s.

The critical components supplied by a global leader in high-speed DDR DRAM controllers and DDR DRAM PHYs for the world’s technology infrastructure, have enabled Tachyum engineers to integrate its IP into Prodigy less than 7 months after entering into a technology partnership. The quality of this IP and the support provided by working closely with Tachyum engineers have allowed the company to close DDR5 timings in record time.

The Samsung 1-beta both 16Gb and 32Gb monolithic DDR5 DRAM, Micron 1-beta DDR5, and Hynix 1-beta DDR5 will be mass produced and available later this year, enabling the mainstream cost of DDR5 and at the same time, reaching 7200 MT/s performance.

While consumer grade DDR5 UDIMM and SODIMM are shipping in small quantities, server grade DDR5 RDIMM is not shipping yet. The industry expects DDR5 RDIMM mass production and mass availability before the end of this year. This will be just in time for Tachyum Prodigy to enter the market as one of the first server processors shipping with DDR5 RDIMM this year. Tachyum has already received DDR5 RDIMM samples from all key vendors enabling Tachyum to validate DDR5 before tape-out this year.

The performance of the DRAM subsystem is essential to support the extremely high-performance needs of Prodigy as it takes on challenging applications in hyperscale data centers, private cloud and AI/HPC. Together, with Tachyum bandwidth amplification and super-sparsity technology, the IP components will enable Prodigy to reach processing speeds of HBM-enabled GPGPUs to satisfy the most-demanding computing applications. Tachyum anticipates that after industry-wide delays, server-quality DDR5 RDIMM will be ready on 2H 2023 for incorporation into Prodigy at its launch.

“As we near tape-out and volume production of Prodigy, getting all the components implemented and tested is a critical step to a successful launch,” said Dr. Radoslav Danilak, founder and CEO of Tachyum. “Working with the team at our DDR5 supplier to integrate their IP in record time has allowed us to continue meeting our timeline while ensuring the speeds and feeds are optimized for customers upon rollout. Because of the performance needs of Prodigy, this is not a simple plug-and-play operation but one that requires high-quality components and world-class support to accomplish. We are excited by this latest milestone and look forward to completing testing soon.”

Prodigy delivers unprecedented data center performance, power, and economics, reducing CAPEX and OPEX significantly. Because of its utility for both high-performance and line-of-business applications, Prodigy-powered data center servers can seamlessly and dynamically switch between workloads, eliminating the need for expensive dedicated AI hardware and dramatically increasing server utilization. Tachyum’s Prodigy integrates 128 high-performance custom-designed 64-bit compute cores, to deliver up to 4x the performance of the highest-performing x86 processors for cloud workloads, up to 3x that of the highest performing GPU for HPC, and 6x for AI applications.

Follow Tachyum

About Tachyum

Tachyum Celebrates 2022 and Announces 2023 Series C and Volume Production Plans

Tue, 03 Jan 2023 00:00:00 +0000

LAS VEGAS, January 3, 2023 – Tachyum™ ended 2022 with accomplishments including the worldwide debut of Prodigy®, the world’s first universal processor for high-performance computing and more than a dozen commercialization partnerships, effectively moving the startup to a leadership position in semiconductors.

2022 marked the introduction of Tachyum’s Prodigy to the commercial market. Prodigy exceeded its performance targets and is significantly faster than any processors currently available in hyperscale, HPC and AI markets. With its higher performance and performance per-dollar and per-watt, Tachyum’s Prodigy processor will enable the world’s fastest AI supercomputer, currently in planning stages.

Tachyum signed 14 significant MOUs with prestigious universities, research institutes, and innovative companies like the Faculty of Information Technology at Czech Technical University in Prague, Kempelen Institute of Intelligent Technologies, M Computers, Picacity, LuxProvide S.A. (Meluxina supercomputer), MatLogica, and Cologne Chip. Other agreements are in progress.

Technical Achievements

The launch of Prodigy followed the successful preproduction and Quality Assurance (QA) phases for hardware and software testing on FPGA emulation boards. During 2022, Tachyum successfully demonstrated Prodigy’s integration with major platforms to address multiple customer needs. These included FreeBSD, Security-Enhanced Linux (SELinux), KVM (Kernel-based Virtual Machine) hypervisor virtualization, and native Docker under the Go programming language (Golang).

Software ecosystem enhancements also included improvements to Prodigy’s Unified Extensible Firmware Interface (UEFI) specification-based BIOS (Basic Input Output System) replacement firmware, incorporating the latest versions of the QEMU emulator and GNU Compiler Collection (GCC). These improvements allow quick and seamless integration of data center technologies into Tachyum-based environments.

Tachyum completed the final piece of its core software stack with a Baseboard Management Controller (BMC) running on a Prodigy emulation system. This enables Tachyum to provide OEM/ODM and system integrators with complete software and firmware stack, and serves as a key component of the upcoming Tachyum Prodigy 4 socket reference design.

In its hardware accomplishments, Tachyum built its IEEE-compliant floating-point unit (FPU) from the ground up — one of the most advanced in the world, with the highest clock speeds — and progressed to running applications in Linux interactive mode on Prodigy FPGA hardware with SMP (Symmetric Multi-Processing) Linux and the FPU. This proved the stability of the system and allowed Tachyum to move forward with additional testing. It completed LINPACK benchmarks using Prodigy’s FPU on a FPGA. LINPACK measures a system’s floating-point computing power by solving a dense system of linear equations to determine performance. It is a widely used benchmark for supercomputers.

The company published three technical white papers that unveiled never-before-disclosed architectural designs of the system-on-chip (SOC) and AI training techniques, revealing how Prodigy addresses trends in AI, enables deep learning workloads that are more environmentally responsible with lower energy consumption and reduced carbon emissions. One paper defined a groundbreaking high-performance, low-latency, low-cost, low-power, highly scalable exascale-flattened networking solution that provides a superior alternative to the more expensive, proprietary and limited scalability InfiniBand communications standard.

Around the world

Tachyum was a highlight of exhibits at Expo 2020 Dubai with the world premiere of the Prodigy Universal Processor for supercomputers, and presented Prodigy at LEAP22 in Riyadh, Saudi Arabia. Tachyum was named one of the Most Innovative AI Solutions Providers to watch by Enterprise World. Company executives were among the featured presenters at ISC High Performance 2022 and Supercomputing 2022 events.

“Our technical teams have accomplished a great deal this year, and the results speak to it: Prodigy is faster than any other processor for the hyperscale, HPC and AI markets,” said Dr. Radoslav Danilak, founder and CEO of Tachyum. “As we move to the commercialization stage and start our C round, 2023 will see a new era in data center computing, an end to extreme power consumption, resource underutilization, and the merciful death of the processor performance plateau that has stymied modern applications and stifled innovation.”

Looking forward

With its Series C funding, expected to close in 2023, Tachyum will finance the volume production of Prodigy Universal Processor Chip and be positioned for sustained profitability, as well as increase headcount.

2023 will see the company move to tape-out, silicon samples, production, and shipments. After running LINPACK benchmarks using Prodigy’s FPU on a FPGA there are only four more steps to go before the final netlist of Prodigy: running UEFI and boot loaders loading Linux on the FPGA, completing vector-based LINPACK testing with I/O, followed by I/O with virtualization, RAS (Reliability, Availability and Serviceability).

Prodigy delivers unprecedented data center performance, power, and economics, reducing CAPEX and OPEX significantly. Because of its utility for both high-performance and line-of-business applications, Prodigy-powered data center servers can seamlessly and dynamically switch between workloads, eliminating the need for expensive dedicated AI hardware and dramatically increasing server utilization. Tachyum’s Prodigy integrates 128 high-performance custom-designed 64-bit compute cores, to deliver up to 4x the performance of the highest-performing x86 processors for cloud workloads, up to 3x that of the highest performing GPU for HPC, and 6x for AI applications.

Follow Tachyum