AMD (Advanced Micro Devices), a U.S.-based manufacturer of semiconductors, microprocessors, and graphics cards, recently introduced the Instinct MI200 6nm GPUs, exascale-class GPU accelerators. The AMD Instinct MI200 series accelerators include the AMD Instinct™ MI250X high-performance computing (HPC) and artificial intelligence (AI) accelerator.
Built on the AMD CDNA™ 2 architecture, the AMD Instinct MI200 GPUs deliver industry-leading application performance for a wide range of HPC workloads. According to AMD, the AMD Instinct MI250X accelerator delivers up to 4.9 times the performance of competing accelerators for double-precision (FP64) HPC applications and exceeds 380 teraflops of maximum theoretical half-precision (FP16) for AI workloads to enable disruptive approaches by further accelerating data-driven research.
According to Forrest Norrod, senior vice president and general manager, Data Center and Embedded Solutions Business Group, AMD:
“AMD Instinct MI200 accelerators deliver breakthrough performance in HPC and AI, helping scientists make generational leaps in research that can dramatically shorten the time from initial hypothesis to discovery. With key innovations in architecture, packaging and system design, the AMD Instinct MI200 series accelerators are the most advanced data center GPUs ever created, delivering exceptional performance to supercomputers and data centers to solve the world’s most complex problems.”
Frontier Supercomputer
AMD collaborated with the U.S. Department of Energy, Oak Ridge National Laboratory and HPE to design the Frontier supercomputer, which is expected to deliver more than 1.5 exaflops of peak computing power. Powered by optimized third-generation AMD EPYC™ processors and AMD Instinct MI250X accelerators, Frontier is expected to push the boundaries of scientific discovery by dramatically improving the performance of large-scale AI, analysis and simulation. It will enable scientists to perform more computations, identify new data patterns, and develop innovative data analysis methods that drive scientific discovery.
Thomas Zacharia, Director, Oak Ridge National Laboratory said:
“The Frontier supercomputer is the culmination of a strong collaboration between AMD, HPE and the U.S. Department of Energy to deliver an exascale-capable system that pushes the boundaries of scientific discovery by dramatically improving the performance of large-scale artificial intelligence, analysis and simulation.”
The AMD Instinct MI200 series accelerators, coupled with third-generation AMD EPYC processors and the ROCm™ 5.0 open software platform, are designed to propel new discoveries for the exascale era and address the most pressing challenges, from climate change to vaccine research.
Key capabilities and features of AMD Instinct MI200 accelerators include:
- AMD CDNA™ 2 Architecture – Second generation matrix cores accelerating FP64 and FP32 matrix operations, delivering up to 4 times the maximum theoretical performance of FP64 over previous generation AMD GPUs. 1,3,4
- Industry-leadingpackaging technology – Industry’s first multi-dies GPU design with 2.5D Elevated Fanout Bridge (EFB) technology offering 1.8 times more cores and 2.7 times more memory bandwidth compared to the previous generation AMD GPU, delivering the industry’s best aggregate peak theoretical memory bandwidth at 3.2 terabytes per second. 4,5,6
- AMD Infinity Fabric™ third-generation technology: up to 8 Infinity Fabric links connect the AMD Instinct MI200 with third-generation EPYC processors and other GPUs in the node to enable unified CPU/GPU memory coherency and maximize system throughput, enabling easier ramping of CPU code to harness the power of accelerators.
AMD Infinity Hub offers a range of containers supporting the Radeon Instinct™ MI50, AMD Instinct™ MI100 or AMD Instinct MI200 accelerators, including several applications such as Chroma, CP2k, LAMMPS, NAMD, OpenMM and more, as well as the popular ML frameworks TensorFlow and PyTorch. New containers are continually being added to the hub.
The AMD Instinct MI250X and AMD Instinct MI250 are available in the open-hard computing accelerator module or OCP Accelerator Module Form Factor (OAM). The AMD Instinct MI210 will be available as a PCIe® card in OEM servers.
The AMD MI250X accelerator is currently available from HPE in the HPE Cray EX supercomputer, and additional AMD Instinct MI200 accelerators are expected in systems from leading OEM and ODM partners in the enterprise markets in Q1 2022, including ASUS, ATOS, Dell Technologies, Gigabyte, Hewlett Packard Enterprise (HPE), Lenovo, Penguin Computing and Supermicro.
Translated from AMD présente ses nouveaux GPU Instinct MI 200 6 nm pour accélérer davantage la recherche axée sur la donnée
