Datacenter designers deploying Machine Learning and AI systems today require systems capable of running more complex workloads that are massively parallel in nature, while continuing to improve system efficiencies. Improvements in the capabilities of accelerators over the last decade, along with the advancement in software, are providing designers with the option to build more efficient heterogeneous computing systems to help them meet today’s challenges.

The Radeon Instinct™ MI8 Server accelerator is a highly efficient, cost-effective inference and HPC solution delivering 8.2 TFLOPS of peak performance with 4GB of ultrafast HBM1 memory, making it the perfect solution for running Deep Learning inference applications, where lots of new smaller data set inputs are being run at half or single precision against trained neural networks to discover new knowledge.1 The MI8 accelerator is also the perfect open solution for general purpose HPC systems deployed in Financial, Energy, Life Science, Automotive, Academic (Research & Teaching), Government Labs and other

Specification

GPU Architecture:                         AMD “Fiji”

Stream Processors:                      4,096

Performance:

Half-Precision (FP16)                    8.2 TFLOPS

Single-Precision (FP32)                8.2 TFLOPS

Double-Precision (FP64)              512 GFLOPS

GPU Memory:                               04GB HBM1

Memory Bandwidth:                    Up to 512 GB/s

Bus Interface:                                PCIe® Gen 3 x16

MxGPU Capability:                        Yes

Board Form Factor:                       Full-Height, Dual-Slot

Length:                                            6”

Thermal Solution:                          Passively Cooled

Standard Max Power:                   175W TDP

OS Support:                                    Linux® 64-bit

ROCm Software Platform:            Yes

Programing Environment:

ISO C++, OpenCL™, CUDA (via AMD’s HIP conversion tool) and Python5 (via Anaconda’s NUMBA)