The MI100 is the first GPU accelerator with new AMD CDNA architecture.
This week Advanced Micro Devices announced the new AMD Instinct MI100 accelerator. The company claims that the MI100 world’s fastest HPC GPU and the first x86 server GPU to surpass the 10 teraflops (FP64) performance barrier.
The MI100 is supported by new accelerated compute platforms from AMD partners Dell, GIGABYTE, HPE, and Supermicro. This GPU, combined with AMD EPYC CPUs and the ROCm 4.0 open software platform, is designed to “propel new discoveries ahead of the exascale era,” according to AMD.
Delivering peak performance for AI
Built on the new AMD CDNA architecture, the AMD Instinct MI100 GPU enables a new class of accelerated systems for HPC and AI when paired with second generation AMD EPYC processors.
The MI100 offers up to 11.5 TFLOPS of peak FP64 performance for HPC. It also delivers up to 46.1 TFLOPS peak FP32 Matrix performance for AI and machine learning workloads.
With new AMD Matrix Core technology, the MI100 delivers a nearly 7x boost in FP16 theoretical peak floating point performance. This is ideal for AI training workloads compared to AMD’s prior generation accelerators.
Designed for scientific research
Supercomputers are most often used to run scientific workloads such as weather prediction models and physics simulations. Traditionally, they were constructed mainly with central processing units (CPUs).
However, researchers today are making more use fo machine learning (ML) as part of their scientific computing. For this reason, new supercomputers are often commissioned with large numbers of GPUs as well as CPUs. This trend is creating a booming demand for accelerators such as the MI100.
“Today AMD takes a major step forward in the journey toward exascale computing as we unveil the AMD Instinct MI100 – the world’s fastest HPC GPU,” said Brad McCredie, corporate vice president, Data Center GPU and Accelerated Processing, AMD.
“Squarely targeted toward the workloads that matter in scientific computing, our latest accelerator, when combined with the AMD ROCm open software platform, is designed to provide scientists and researchers a superior foundation for their work in HPC.”
The AMD Instinct MI100 accelerators are expected by end of the year. They will be available in systems from major OEM and ODM partners in the enterprise markets.