LAS VEGAS, October 26, 2022 – Tachyum™ today unveiled a high-performance, low-latency, low-cost and low-power, highly scalable exascale flattened networking solution that provides a superior alternative to the more expensive, proprietary and limited scalability InfiniBand communications standard.
Ethernet has a long history of providing networking solutions that have constantly evolved to address customer requirements and have consistently dominated the market. Data center networking topologies have likewise evolved from the traditional Clos architecture with multiple levels of hierarchy to various levels of flattened network topologies that reduce latency while saving cost and power.
With the world moving away from multi-layered networks, companies like Meta are transitioning to flattened networks with fewer layers that enable them to reap massive benefits, such as optimizing data center costs and efficiency. Tachyum has developed 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 to help it achieve its goal of doing more with less.
Among the performance milestones reached through Tachyum’s flattened networking capabilities is achieving network bandwidth with RoCE of 186 Gbps between servers. Additionally, bandwidth increases to 224 Gbps per lane will be available in 2025, and 1024 lanes in 2U switching with OSFP-XD in the near future.
“Again, Tachyum shows its industry leadership with novel flattened networking, which provides a roadmap for many to follow in the future,” said Dr. Radoslav Danilak, founder and CEO of Tachyum. “Prodigy’s groundbreaking architecture will have the best possible networking topology for Prodigy clusters, including upcoming supercomputer deployments. This technology should be a requirement for Slovakian supercomputers. Anything else is grossly unjustifiably inferior.”
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.
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.
A video demonstration of Tachyum’s HPC networking for supercomputers can be found below.
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/.