Naši softwaroví inženýři pracují na tom, aby odemkli plný potenciál Univerzálního Procesoru Tachyum Prodigy® od prvního dne s bohatým ekosystémem aplikací, systémového softwaru, rámců a knihoven které nativně podporují instrukční sadu Prodigy.
Softwarový ekosystém Prodigy také neustále roste a v plánech je neustálé přidávání nového softwaru.
Prodigy navíc nabízí možnost spustit programy sestavené pro x86, Arm a RISC-V, čímž pro zákazníky partnery zpřístupní rychlé a snadné testování a evaluaci bez nutnosti dalšího nastavování na jejich stávající architektuře před spuštěním jejich aplikací nativně na Prodigy.
This CPU emulator can be used as either a standalone functional emulator of Prodigy ISA or as part of cross-mode GDB allowing to debug Prodigy user mode application on any other host platform.
QEMU is a generic and open-source machine emulator and virtualizer. It supports both User-mode emulation and Full-system emulation.
Tachyum port can be used for two purposes - to emulate Prodigy ISA on any other supported architecture or to run pre-existing applications on Prodigy Linux/BSD system.
When used as a user-mode binary emulator, QEMU enables dynamic binary translation from x86-64, ARM v8 and RISC-V.
GDB, the GNU Project debugger, allows you to see what is going on “inside” another program while it executes or what another program was doing at the moment it crashed.
With GDB, a Prodigy application can be debugged in various scenarios: as a part of cross-toolchain using either functional simulator or QEMU as execution engine on non-Prodigy host; connecting to GDB server running on native or full-system-mode-emulation Linux/BSD kernel, from any supported host; or as a native Linux/BSD debugger.
“Open On-Chip Debugger” allows to connect to a target using low-level JTAG interface to help facilitate system bring-up and initial stages of very low-level debugging.
Kgdb is intended to be used as a source level debugger for the Linux kernel. It is used along with gdb to debug a Linux kernel.
JTAG is more than debugging and programming. Processors use JTAG to provide access to their debug/emulation functions and all FPGAs and CPLDs use JTAG to provide access to their programming functions.
Compilers and Libraries
The GNU Compiler Collection includes front ends for C, C++, Objective-C, Fortran, Ada, Go, and D, as well as libraries for these languages (libstdc++,…).
Existing since 1987, this compiler is a de-facto standard for almost every major CPU platform.
The GNU C Library project provides the core libraries for the GNU system and GNU/Linux systems, as well as many other systems that use Linux as the kernel. These libraries provide critical APIs including ISO C11, POSIX.1-2008, BSD, OS-specific APIs and more. These APIs include such foundational facilities as open, read, write, malloc, printf, getaddrinfo, dlopen, pthread_create, crypt, login, exit and more.
The GNU C Library is designed to be a backwards compatible, portable, and high-performance ISO C library. It aims to ollow all relevant standards including ISO C11, POSIX.1-2008, and IEEE 754-2008.
Musl is a C standard library implementation for Linux. Some of musl’s major advantages over glibc and uClibc/uClibc-ng are its size, correctness, static linking support, and clean code.
The Go language has always been defined by a spec, not an implementation. The Go team has written two different compilers that implement that spec:
gcis the original compiler, and the go tool uses it by default,
gccgois a different implementation with a different focus.
Both of these compilers are natively supported by the Tachyum Prodigy® ISA.
The Clang project provides a language front-end and tooling infrastructure for languages in the C language family (C, C++, Objective C/C++, OpenCL, CUDA, and RenderScript) for the LLVM project.
Clang is considered to be a production quality C, Objective-C, C++ and Objective-C++ compiler. As example, Clang is used in production to build performance-critical software like Chrome or Firefox.
Boot Loaders and Monitors
UEFI stands for “Unified Extensible Firmware Interface.” The UEFI Specification defines a new model for the interface between personal-computer operating systems and platform firmware. The interface consists of data tables that contain platform-related information, plus boot and runtime service calls that are available to the operating system and its loader. Together, these provide a standard environment for booting an operating system and running pre-boot applications.
The OpenBMC project is a Linux Foundation collaborative open-source project whose goal is to produce an open-source implementation of the Baseboard Management Controllers (BMC) Firmware Stack. OpenBMC is a Linux distribution for BMCs meant to work across heterogeneous systems that include enterprise, high-performance computing (HPC), telecommunications, and cloud-scale data centers.
GNU GRUB is a Multiboot boot loader. It was derived from GRUB, the GRand Unified Bootloader, which was originally designed and implemented by Erich Stefan Boleyn.
systemd-boot is a UEFI boot manager which executes configured EFI images. The default entry is selected by a configured pattern (glob) or an on-screen menu.
OS Kernels and Distro Tools
The Linux kernel is a free and open-source, monolithic, Unix-like operating system kernel. It is deployed on a wide variety of computing systems, from personal computers, mobile devices, mainframes, and supercomputers to embedded devices, such as routers, wireless access points, private branch exchanges, set-top boxes, FTA receivers, smart TVs, personal video recorders, and NAS appliances.
Its availability, continuous development, and ongoing support have spawned a plethora of operating system distributions, commonly also called Linux.
SELinux is a security enhancement to Linux which allows users and administrators more control over access control.
Access can be constrained on such variables as which users and applications can access which resources. These resources may take the form of files. Standard Linux access controls, such as file modes (-rwxr-xr-x) are modifiable by the user and the applications which the user runs. Conversely, SELinux access controls are determined by a policy loaded on the system which may not be changed by careless users or misbehaving applications.
FreeBSD is an operating system used to power modern servers, desktops, and embedded platforms. A large community has continually developed it for almost thirty years. Its advanced networking, security, and storage features have made FreeBSD the platform of choice for many of the busiest web sites and most pervasive embedded networking and storage devices.
Virtualization & Interoperability
KVM (for Kernel-based Virtual Machine) is a full virtualization solution for Linux on x86 hardware containing virtualization extensions. It consists of a loadable kernel module, kvm.ko, that provides the core virtualization infrastructure and a processor specific module, kvm-intel.ko or kvm-amd.ko.
The Xen Project is focused on advancing virtualization in a number of different commercial and open source applications, including server virtualization, Infrastructure as a Service (IaaS), desktop virtualization, security applications, embedded and hardware appliances, and automotive/aviation.
Docker is a set of platform as a service products that use OS-level virtualization to deliver software in packages called containers. Containers are isolated from one another and bundle their own software, libraries and configuration files; they can communicate with each other through well-defined channels.
Kubernetes is an open-source system for automating deployment, scaling, and management of containerized applications.
It groups containers that make up an application into logical units for easy management and discovery. Kubernetes builds upon 15 years of experience of running production workloads at Google, combined with best-of-breed ideas and practices from the community.
Samba is the standard Windows interoperability suite of programs for Linux and Unix.
Since 1992, Samba has provided secure, stable and fast file and print services for all clients using the SMB/CIFS protocol, such as all versions of DOS and Windows, OS/2, Linux and many others.
Samba is an important component to seamlessly integrate Linux/Unix Servers and Desktops into Active Directory environments. AD Domain Controller is also supported on Tachyum Prodigy® architecture.
Ansible is a radically simple IT automation system. It handles configuration management, application deployment, cloud provisioning, ad-hoc task execution, network automation, and multi-node orchestration. Ansible makes complex changes like zero-downtime rolling updates with load balancers easy.
C is a general-purpose, procedural computer programming language supporting structured programming, lexical variable scope, and recursion, with a static type system.
C++ is a general-purpose programming language created by Bjarne Stroustrup as an extension of the C programming language, or “C with Classes”.
Erlang is a programming language used to build massively scalable soft real-time systems with requirements on high availability. Some of its uses are in telecoms, banking, e-commerce, computer telephony and instant messaging. Erlang’s runtime system has built-in support for concurrency, distribution and fault tolerance.
Fortran is a general-purpose, compiled imperative programming language that is especially suited to numeric computation and scientific computing.
Go is a statically typed, compiled programming language designed at Google. Go is syntactically similar to C, but with memory safety, garbage collection, structural typing, and CSP-style concurrency.
JVM (Java Virtual Machine) is an abstract machine. It is a specification that provides runtime environment in which java bytecode can be executed.
Lua is a powerful, efficient, lightweight, embeddable scripting language. It supports procedural programming, object-oriented programming, functional programming, data-driven programming, and data description.
OpenJDK is a free and open-source implementation of the Java Platform, Standard Edition.
Perl is a family of two high-level, general-purpose, interpreted, dynamic programming languages. It runs on over 100 platforms from portables to mainframes and is suitable for both rapid prototyping and large-scale development projects.
PHP is a popular open source general-purpose scripting language that is especially suited to web development and can be embedded into HTML.
Python is an interpreted, high-level and general-purpose programming language. It is used successfully in thousands of real-world business applications around the world, including many large and mission critical systems.
R is a language and environment for statistical computing and graphics. It is a GNU project which is similar to the S language and environment.
R provides a wide variety of statistical (linear and nonlinear modelling, classical statistical tests, time-series analysis, classification, clustering, …) and graphical techniques, and is highly extensible.
R can be used, for instance, in cooperation with Gnuplot, which is also supported.
Ruby is an interpreted, high-level and general-purpose programming language with a focus on simplicity and productivity.
Tcl is a high-level, general-purpose, interpreted, dynamic programming language. It is suitable for a very wide range of uses, including web and desktop applications, networking, administration, testing and many more.
Performance Testing and Optimization
Benchmark suites developed by The Standard Performance Evaluation Corporation (SPEC).
Test suite consists of 12 benchmark programs, designed to test exclusively the integer performance of the system.
Test suite consists of benchmark programs designed to test exclusively the integer performance of the system.
Test suite contains 17 benchmark programs, designed to evaluate the floating-point operations performance of the system.
Test is organized in 2 suites: SPECrate 2017 Floating Point and SPECspeed 2017 Floating Point containing in total 23 benchmark programs, designed to evaluate the floating-point operations performance of the system.
The Linpack Benchmark is a measure of a computer’s floating-point rate of execution. It is determined by running a computer program that solves a dense system of linear equations. The Linpack Benchmark is something that grew out of the Linpack software project. It was originally intended to give users of the package a feeling for how long it would take to solve certain matrix problems.
HPL is a High-Performance Linpack benchmark implementation. The code solves a uniformely random system of linear equations and reports time and floating-point execution rate using a standard formula for operation count.
Web and E-mail
Apache is the most widely used open-source cross-platform web server software. The Apache HTTP Server Project is part of the Apache Software Foundation.
Postfix is a free and open-source mail transfer agent that routes and delivers electronic mail.
Dovecot is an open-source IMAP and POP3 server for Unix-like operating systems, written primarily with security in mind.
MariaDB Server is one of the most popular open source relational databases. It’s made by the original developers of MySQL. MariaDB supports a lot of different storage engines.
PostgreSQL is a powerful, open source object-relational database system with over 30 years of active development that has earned it a strong reputation for reliability, feature robustness, and performance.
MongoDB is a general purpose, document-based, distributed database built for modern application developers and for the cloud era. MongoDB is a NoSQL database program, which stores data in JSON-like documents with dynamic schema.
SQLite is the most used database engine in the world. SQLite is a C-language library that implements a small, fast, self-contained, high reliability, full-featured SQL database engine.
RocksDB is a storage engine with key/value interface, where keys and values are arbitrary byte streams. It is a C++ library. It was developed at Facebook based on LevelDB and provides backwards-compatible support for LevelDB APIs.
Editors and VCS
Git is an open source distributed version control system designed to handle everything from small to very large projects with speed and efficiency.
Subversion (abbreviate Svn after its command name svn) is a software versioning and revision control system distributed as open source under the Apache License.
Vim is a highly configurable text editor built to make creating and changing any kind of text very efficient. It is included as “vi” with most UNIX systems and with Apple OS X.
Emacs is the advanced, extensible, customizable, self-documenting editor.
Sed is a stream editor, which is used to perform basic text transformation on an input stream.
Gawk is the GNU implementation of Awk, a specialized programming language for the easy manipulation of formatted text, such as tables of data.
Gnuplot is a portable command-line driven graphing utility for Linux, OS/2, MS Windows, OSX, VMS, and many other platforms.
It was originally created to allow scientists and students to visualize mathematical functions and data interactively, but has grown to support many non-interactive uses such as web scripting.
Gnuplot works well with the R programming language, which is also supported.
Grep is a command-line utility for searching plain-text data sets for lines that match a regular expression.
gzip is a single-file/stream lossless data compression utility, where the resulting compressed file generally has the suffix .gz.
GNU tar is an archiving program designed to store multiple files in a single file (an archive), and to manipulate such archives.
Zsh is a shell designed for interactive use, although it is also a powerful scripting language. Many of the useful features of bash, ksh, and tcsh were incorporated into zsh; many original features were added.
Netscape Portable Runtime (NSPR) provides a platform-neutral API for system level and libc-like functions. The API is used in the Mozilla clients, many of Red Hat’s and Oracle’s server applications, and other software offerings.
Ceph is an open-source software storage platform, implements object storage on a single distributed computer cluster, and provides 3-in-1 interfaces for object-, block- and file-level storage.
The Apache™ Hadoop® project develops open-source software for reliable, scalable, distributed computing.
The Apache Hadoop software library is a framework that allows for the distributed processing of large data sets across clusters of computers using simple programming models. It is designed to scale up from single servers to thousands of machines, each offering local computation and storage.
Apache Spark™ is a multi-language engine for executing data engineering, data science, and machine learning on single-node machines or clusters.
Operating System Libraries
Buildroot is a simple, efficient and easy-to-use tool to generate embedded Linux systems through cross-compilation.
Can handle everything: Cross-compilation toolchain, root filesystem generation, kernel image compilation and bootloader compilation.
Thanks to its kernel-like menuconfig, gconfig and xconfig configuration interfaces, building a basic system with Buildroot is easy and typically takes 15-30 minutes.
The Yocto Project (YP) is an open-source collaboration project that helps developers create custom Linux-based systems regardless of the hardware architecture.
The project provides a flexible set of tools and a space where embedded developers worldwide can share technologies, software stacks, configurations, and best practices that can be used to create tailored Linux images for embedded and IOT devices, or anywhere a customized Linux OS is needed.
With tens of thousands of users, RabbitMQ is one of the most popular open source message brokers.
RabbitMQ is lightweight and easy to deploy on premises and in the cloud. It supports multiple messaging protocols. RabbitMQ can be deployed in distributed and federated configurations to meet high-scale, high-availability requirements.
gRPC is a modern open source high performance Remote Procedure Call (RPC) framework that can run in any environment. It can efficiently connect services in and across data centers with pluggable support for load balancing, tracing, health checking and authentication. It is also applicable in last mile of distributed computing to connect devices, mobile applications and browsers to backend services.
A specification for a set of compiler directives, library routines, and environment variables that can be used to specify high-level parallelism in Fortran and C/C++ programs. OpenMP is managed by the OpenMP Architecture Review Board (OpenMP ARB).
The Open MPI Project is an open source Message Passing Interface implementation that is developed and maintained by a consortium of academic, research, and industry partners. Open MPI is therefore able to combine the expertise, technologies, and resources from all across the High Performance Computing community in order to build the best MPI library available.
Dense and Sparse BLAS - optimized using Tachyum vector and matrix instructions in standard and low precision modes.
Dense and Sparse LAPACK - LU, QR, Cholesky, Eigenvector/Eigenvalue, SVD decompositions.
LAPACK is written in Fortran 90 and provides routines for solving systems of simultaneous linear equations, least-squares solutions of linear systems of equations, eigenvalue problems, and singular value problems.
Iterative methods, conjugate gradient and polynomial filtered Lanczos optimized using Tachyum BLAS and GEMM.
Optimized using Tachyum vector and matrix.
AI Frameworks Custom Extensions
The BLAS, Basic Linear Algebra Subprograms are routines that provide standard building blocks for performing basic vector and matrix operations.
Static and dynamic loss scaling.
Magnitude based block pruning
Activation & Loss Function – optimized utilizing Tachyum vector instructions in standard and low precision modes.
Dense GEMM (General Matrix Multiply) library implemented utilizing Tachyum matrix instructions in standard and low precision modes, stochastic rounding, single and multithreaded.
Custom Sparse GEMM library implemented utilizing Tachyum vector and matrix instructions.
Convolutional and Dense operators implemented utilizing Tachyum matrix instructions in standard and low precision modes, including depthwise separable and pointwise convolutions.
Circulant and Butterfly Convolutional and Dense operators implemented utilizing custom FFT (fast Fourier transform) for matrix multiplication.
Activation & Loss functions optimized utilizing Tachyum vector instructions in standard and low precision modes.
Convolutional and Dense operators implemented utilizing Tachyum custom Eigen+GEMM libraries and Tachyum matrix instructions in standard and low precision modes.
INT8 quantized version of TensorFlow optimized utilizing Tachyum vector and matrix instructions in standard and low precision modes.
Implemented in standard and low precision modes.
- Resnet: Residual Networks.
- Vision Transformer – ViT, DeiT
- MobileNet: Model designed to be used in mobile applications.
- ShuffleNet: An extremely efficient convolutional neural network for mobile devices.
Implemented in standard and low precision modes.
- YOLO: A clever convolutional neural network (CNN) for doing object detection in real-time.
- MaskRCNN: A deep neural network aimed to solve instance segmentation problem in machine learning or computer vision.
Implemented in standard and low precision modes with block structured sparsity.
- BERT: Bidirectional Encoder Representations from Transformers is a Transformer-based machine learning technique for NLP pre-training developed by Google.
- Sparse Transformer: Transformer based architecture which utilises sparse factorizations of the attention matrix.
Neural ODE: Neural Ordinary Differential Equations.
Graph Neural ODE: A counterpart to GNNs (graph neural networks) where the input-output relationship is determined by a continuum of GNN layers, blending discrete topological structures and differential equations.
Neural PDE: Neural partial differential equation.
HPC/AI for Life Sciences and Physical Sciences
DeepMD-kit is a package written in Python/C++, designed to minimize the effort required to build deep learning based model of interatomic potential energy and force field and to perform molecular dynamics (MD). This brings new hopes to addressing the accuracy-versus-efficiency dilemma in molecular simulations.
Quantum ESPRESSO is an integrated suite of Open-Source computer codes for electronic-structure calculations and materials modeling at the nanoscale. It is based on density-functional theory, plane waves, and pseudopotentials.
LAMMPS is a classical molecular dynamics code with a focus on materials modeling. It’s an acronym for Large-scale Atomic/Molecular Massively Parallel Simulator. It can be used to model atoms or, more generically, as a parallel particle simulator at the atomic, meso, or continuum scale.
Tachyum is assembling a unified software distribution, which will include a set of native software packages required for common tasks. This will allow our customers and partners to explore the Prodigy software ecosystem early, assist us in delivering ported software that has been tested in real-world scenarios and help us verify that our ported software is capable of handling actual customer use cases.
The unified software distribution package will also include binaries required for integration with x86-based applications, which will give customers the ability to test their own code and applications in the Prodigy ecosystem. It will also feature a variety of demo applications and user software, along with tools for testing and debugging.
Having all this functionality bundled in a single, unified, and well-tested distribution package will provide customers and partners with an easy, straightforward platform for early Prodigy testing ahead of silicon availability while reducing support requirements. Additionally, it will give the users a way to effortlessly restore the original image in case of a fatal error occurring during their testing. The package will be available for our partners to download in late 2023.