Radeon Instinct™ MI6 Accelerator

Versatile training and inference accelerator for machine intelligence and deep learning.

Based on the “Polaris” GRAPHICS ARCHITECTURE built to tackle training and inference workloads

36 COMPUTE UNITS available to run many smaller batches of data simultaneously against trained deep learning neural networks

Up to 5.7 TFLOPS of peak FP32 and FP16 compute performance to speed up compute intensive machine intelligence

38 GFLOPS/watt peak FP16/FP32 performance. Offering a great low-cost, highly-efficient server solution for inference and edge-training

16GB of GDDR5 MEMORY, up to 224GB/s memory bandwidth

Passively cooled, 150W TDP board power – single slot board designed to fit in most standard server designs

MxGPU for Virtualized Compute Workloads – drive greater utilization and capacity in the data center

ROCm Software Platform provides open source Hyperscale and HPC-class solutions

ROCm logo

ROCm, a New Era in Open GPU Computing

Platform for GPU-enabled HPC and ultrascale computing.

The Ultimate Solution for Your Computing Needs

machine intelligence digital brain abstract image

Inference and Edge Training for Deep Learning

  • ROCm software platform provides open source Hyperscale platform
  • Open source Linux® drivers, HCC compiler, tools and libraries for full control from the metal forward
  • Optimized MIOpen deep learning framework libraries
  • Large BAR support for mGPU peer to peer
  • MxGPU SR-IOV hardware virtualization for optimized system utilizations
  • Open industry standard support of multiple architectures and industry standard interconnect technologies
HPC

HPC General Purpose and Development

  • ROCm software platform provides open source HPC-class platform
  • Open source Linux® drivers, HCC compiler, tools and libraries for full control from the metal forward
  • MxGPU SR-IOV hardware virtualization for optimized system utilizations

Instinct™ MI6

GPU Specifications
GPU Architecture
Polaris
Lithography
14nm FinFET
Stream Processors
2304
Peak Half Precision (FP16) Performance
5.73 TFLOPs
Peak Single Precision (FP32) Performance
5.73 TFLOPs
Peak Double Precision (FP64) Performance
358 GFLOPs
GPU Memory
Memory Size
16 GB
Memory Type (GPU)
GDDR5
Memory Bandwidth
224 GB/s
Memory ECC Support
No
Board Specifications
Form Factor
PCIe Add-in Card
Bus Type
PCIe 3.0 x16
TDP
150W
Cooling
Passive
Board Width
Single Slot
Board Length
9.5" (241 mm)
Board Height
Full Height
External Power Connectors
1x PCIe 6-pin
Additional Features
Supported Technologies
OpenCL 2.0
Solutions
HPC and Machine Intelligence
Software API Support
DirectX
12.0 (feature level 12_0)
OpenGL
4.6
OpenCL
2.0
Vulkan
1.0
Product Basics
Product Family
Radeon Instinct™
Product Line
Radeon Instinct™ MI Series
Platform
Server
Launch Date
June 2017