AMD announced the general availability of the AMD Instinct MI200 accelerator family and ROCm 5 software. Built on the AMD CDNA 2 architecture and including the MI210, MI250, and MI250X introduced in November last year, the AMD Instinct MI200 series accelerators address the growing demand for compute-accelerated data center workloads by delivering premium technology to a broad HPC and AI customer base. and shortens exploration time. Brad McCredie, Corporate Vice President of GPU and Accelerated Processing at AMD
, said in a statement, “With twice the current platforms compared to our previous generation accelerators, increased customer acceptance in HPC and AI applications, and new support from commercial ISVs for critical workloads. We continue to drive adoption of AMD Instinct MI200 accelerators and the ROCm 5 software ecosystem. Now with the launch of the AMD Instinct MI210 accelerator in the MI200 family, our customers need state-of-the-art accelerated processing for large-scale HPC and AI workloads or exascale-class technology in a commercial format. They can choose the accelerator that works best for their workload.”
Powering the future of HPC and artificial intelligence
AMD Instinct MI200 series accelerators are designed to support discovery in exascale systems and enable researchers, scientists and engineers to tackle our most pressing challenges, from climate change to vaccine research. AMD Instinct MI210 accelerators offer premium technologies exclusively for customers who need HPC and AI performance in PCIe® format. Powered by the AMD CDNA™ 2 architecture, AMD Instinct MI210 accelerators also extend AMD performance leadership in dual-precision (FP64) computing on PCIe form factor cards. It also provides a robust solution for accelerated deep learning training, offering a wide range of mixed precision capabilities based on AMD Matrix Core Technology.
Complements the ecosystem with ROCm
The AMD ROCm platform, an open software platform that allows researchers, scientists and engineers to leverage the power of AMD Instinct accelerators to drive scientific discoveries, is a multi-dimensional platform that supports the best HPC and AI applications. built on the foundation of numerous apps and libraries.
Partners and expanding customer ecosystem
Software with the addition of commercial ISVs including AMD, Ansys®, Cascade Technologies, and TempoQuest as more purpose-built applications are optimized to work with ROCm and AMD Instinct accelerators continues to grow its ecosystem.
AMD also has different partners to address next-generation computing challenges such as ASUS, Dell Technologies, Gigabyte, HPE, Lenovo, Supermicro and System Integrators such as Colfax, Exxact, KOI Computers, Nor-Tech, Penguin, and Smetric. enables them to provide solutions. Supercomputing customers are already taking advantage of these new benefits, including the Frontier installation at Oak Ridge National Laboratory, KTH/Dardel, CSC/LUMI and Cines/Adastra.
Easy access for customers and partners
AMD Accelerator Cloud provides customers with an environment to remotely access and evaluate AMD Instinct accelerators and AMD ROCm software. Whether porting legacy code, benchmarking an application, or testing multi-GPU or multi-node scaling, AMD Accelerator Cloud provides potential customers and partners with fast and easy access to the latest GPUs and software. AMD Accelerator Cloud is also used to support a variety of events such as hackathons and ROCm training sessions available to both existing and prospective customers, allowing developers to hone their skills and learn how to get the most out of AMD Instinct accelerators.
40293
