AMD Locks Horns With NVIDIA, Unveils AMD MI100 GPU Accelerator

AMD gives tough competition to its rival NVIDIA in the GPU space with the release of Instinct MI 100 accelerator. The company claims that the new GPU is the world’s fastest HPC GPU.

The growing interest in AI, machine learning, IoT, and big data has surged the demand for high-performance computing (HPC). HPC systems can process, store, and analyze large volumes of data at high speed and have gauged the attention of different verticals, including government agencies, defense, healthcare, financial services, and energy. Since HPC can perform complex computational problems that lead to groundbreaking scientific discoveries, its market size is forecastedOpens a new window to grow from $32.11 billion in 2017 to $45 billion in 2022. 

This multi-billion dollar market has the interest of HPC vendors such as IBM, HPE, Intel, NVIDIA, and AMD. While NVIDIA is leading the pack due to the high demand for HPC and AI applications, Advanced Micro Devices (AMD) has gradually become a strong competitor of NVIDIA. In June 2020, the TOP500 supercomputers list revealedOpens a new window that NVIDIA and AMD processors had powered the world’s fastest supercomputers. 

As both the companies continue to advance in the HPC market, AMD made an entry in the data center GPUs space with the release of the new AMD Instinct MI100 accelerator for HPC and AI projects at the SC20 supercomputing virtual conferenceOpens a new window . The company claims that the MI100 accelerator is the world’s fastest HPC GPU accelerator for scientific workloads and the first to surpass the ten teraflops of double-precision (FP64) peak performance barrier.

Brad McCredie, corporate vice president, data center GPU and accelerated processing, AMD, saidOpens a new window , “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.”

The MI100 GPU is a direct competitor of NVIDIA’s A100 datacenter GPUs, which recently got upgraded to the A100 80GB version to power supercomputing performance in AI and ML research.

Also Read: 6 Data Center Trends You Should Know

The Technology Behind AMD Instinct MI100 Accelerator

The new GPU is built on the company’s new compute DNA (CDNA) architecture and is combined with AMD’s 2nd generation EPYC processors to help researchers drive accelerated computing into the exascale computing era. Combined with 2nd generation EPYC processors, the new GPU offers up to 11.5 teraflops of peak FP64 performance for HPC and up to 46.1 teraflops peak FP32 matrix performance for AI and machine learning workloads.

McCredie explainedOpens a new window , “Powered by the first AMD CDNA architecture, and the perfect complement to 2nd Gen AMD EPYC processors, the AMD Instinct MI100 accelerators deliver groundbreaking GPU compute performance.”

Since the new GPU is specifically designed for scientific research, it is supported by AMD ROCm — an open-source exascale-class platform that enables developers and researchers to develop high-performance applications easily. This open software platform will prevent vendor lock-in and equip developers with the necessary tools, compilers, APIs, and libraries to accelerate code development and solve complex computational challenges.

McCredie further addedOpens a new window , “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 new MI100 accelerator will be shipped by the end of 2020 from major partners, including Dell, Gigabyte, HPE, and Supermicro.

AMD is working hard to conquer the HPC market currently ruled by Intel and NVIDIA. The Xilinx $35 billion acquisition deal was AMD’s strategic step to lead the data center market. Now with the MI100 accelerator, the company plans to give an edge to NVIDIA.

Karl Freund, a senior analyst at Moor Insights and Strategy, saidOpens a new window , “Should NVIDIA be worried about AMD’s new entry into the data center? In AI, no. In HPC, yes. AMD’s new GPU is an excellent stepping stone to its exascale platform. I expect AMD will be selected by price-sensitive supercomputer installations, which may not have a tremendous need for bleeding-edge AI on the same platform in the short term. However, many or even most HPC users see AI as an integral part of the workflow these days.”

AMD seems on a mission to challenge its rivals in the CPU and GPU space. AMD’s EPYC CPUs have outweighed Intel’s Xeon CPUs and taken a sizable market share from the company. Now, with the MI100 GPU accelerators, AMD plans to overshadow the industry heavyweight NVIDIA. However, AMD will have to ensure that its new GPU accelerator performs remarkably well, as claimed by the company, to shake NVIDIA’s strong foothold in the HPC market.

Is AMD Instinct MI100 GPU accelerator a strong competitor to NVIDIA GPU? Comment below or let us know on LinkedInOpens a new window , TwitterOpens a new window , or FacebookOpens a new window . We’d love to hear from you!