Introducing AMD Instinct MI100, the world's fastest HPC GPU

In a press release, AMD today announced the new AMD Instinct MI100 accelerator, the world's fastest HPC GPU and the first x86 server GPU to break through the 10 teraflops (FP64) barrier. Supported by the new accelerated computing platforms from Dell, Gigabyte, HPE and Supermicro, MI100, combined with AMD EPYC CPUs and the ROCm 4.0 open software platform, is designed to drive new breakthroughs ahead of the exascale era.

Based on the new AMD CDNA architecture, the AMD Instinct MI100 GPU delivers up to 11.5 TFLOPS of peak FP64 performance for HPC and FP32 peak performance of up to 46.1 TFLOPS for artificial intelligence and machine learning workloads. With the new AMD Matrix Core technology, MI100 also delivers a nearly seven-fold increase in theoretical FP16 peak floating-point performance for AI training workloads compared to AMD's previous generation accelerators.

Read Also: Here are the first photos of the mighty ASRock Radeon RX 6800 XT Taichi

Key features and</a> features of the AMD Instinct MI100 throttle include:

New AMD CDNA architecture: designed to prepare AMD GPUs for the exascale era by offering very high performance and energy efficiency. Leading FP64 and FP32 Performance for HPC Workloads: Delivers peak FP64 performance of 11.5TFLOPS and peak FP32 performance of 23.1TFLOPS, enabling scientists and researchers around the world to accelerate discoveries in areas such as science, energy , finance, universities, government, defense and more. New Matrix Core technology for HPC and AI: Enhanced performance for a full range of single and mixed precision matrix operations, such as FP32, FP16, bFloat16, Int8 and Int4, designed to increase convergence of HPC and AI. 2nd Gen AMD Infinity Fabric Technology: Instinct MI100 provides approximately twice the peer-to-peer (P2P) peak I / O bandwidth on PCIe 4.0 with up to 340GB / s of aggregate bandwidth per card with three AMD Infinity Fabric links. In a server, MI100 GPUs can be configured with up to two fully connected quad GPU hives, each providing up to 552GB / s of P2P I / O bandwidth for fast data sharing. Ultra-fast HBM2 memory: Features 32GB high-bandwidth HBM2 memory with a 1.2GHz clock rate and offers 1.23TB / s bandwidth to support large data sets and help eliminate bottlenecks in moving data in and out of memory. Support for the latest PCIe Gen 4.0 interface: Designed with support for the latest PCIe Gen 4.0 technology that provides up to 64GB / s of theoretical bandwidth to transport data from the CPU to the GPU. If you are looking for a good motherboard for your new AMD Ryzen-based build, on Amazon you will find the excellent ASUS Prime X570-PRO at a great price!