Radeon Instinct MI60

Unleash Deep Learning Discovery

The Radeon Instinct™ MI50 compute card is designed to deliver high levels of performance for deep learning, high performance computing (HPC), cloud computing, and rendering systems. This new accelerator is designed with optimized deep learning operations, exceptional double precision performance, and hyper-fast HBM2 memory delivering 1 TB/s memory bandwidth speeds.

Scale your datacenter server designs with AMD’s Infinity Fabric™ Link technology that can be used to directly connect up to 2 GPU clusters of 4 GPUs in a single server at up to 6x the speed of PCIe® 3.01.

Quickly achieve reliable and accurate results in large-scale system deployments with the new Radeon Instinct™ MI60 and MI50 which are equipped with full-chip ECC2 and RAS capabilities3.

Combine this finely balanced and ultra-scalable solution with our ROCm open ecosystem that includes Radeon Instinct optimized MIOpen libraries supporting frameworks like TensorFlow and Caffe 2, and you have a solution ready for the next era of compute and machine intelligence.

Key Features

Based on “Vega 7nm” Technology with 60 supercharged Compute Units (3840 Stream Processors)

Up to 53.6 TOPS INT8 Performance for Inference Workloads

Up to 26.8 TFLOPS FP16 and 13.4 TFLOPS FP32 Performance for Training Workloads

Up to 6.7 TFLOPS Double Precision for HPC

16 GB Ultra-fast HBM2 ECC Memory with up to 1 TB/s Memory Bandwidth

The World’s First PCIe® Gen 4 x16 Capable GPU4

AMD Infinity Fabric™ Link – up to 200 GB/s peer-to-peer GPU communication speeds

ROCm Open Ecosystem

many computer screens

Optimized for Deep Learning

The Radeon Instinct™ MI50 server accelerator designed on the world’s first 7nm FinFET technology process brings customers a full-feature set based on the industry newest technologies. The MI50 is AMD’s workhorse accelerator offering that is ideal for large scale deep learning. Delivering 26.8 TFLOPS of native half-precision (FP16) or 13.4 TFLOPS single-precision (FP32) peak floating point performance and INT8 support and combined with 16GB of high-bandwidth HBM2 ECC memory2, the Radeon Instinct™ MI50 brings customers the compute and memory performance needed for enterprise-class, mid-range compute capable of training complex neural networks for a variety of demanding machine deep learning applications in a cost-effective design.

server room technology

Accuracy and Speed Now Go Hand-in-Hand

For High Performance Compute (HPC) workloads, the Radeon Instinct™ MI50 accelerator delivers incredible double precision speeds of up to 6.7 TFLOPS, allowing scientists and researches across the globe to more efficiently process HPC parallel codes across several industries including life sciences, energy, finance, automotive and aerospace, academics, government, defense and more.

AMD’s next-generation HPC solutions are designed to deliver optimal compute density and performance per node with the efficiency required to handle today’s massively parallel data-intensive codes; as well as, to provide a powerful, flexible solution for general purpose HPC deployments. The ROCm software platform brings a scalable HPC-class solution that provides fully open-source Linux drivers, HCC compilers, tools and libraries to give scientists and researchers system control down to the metal.

ROCm logo

ROCm, a New Era in Open GPU Computing

Platform for GPU-enabled HPC and deep learning.

Radeon Vega 7nm

“Vega 7nm” Graphics Technology

Improved Performance per Watt, Optimized Deep Learning Operations and much more!

Instinct™ MI50

GPU Specifications
GPU Architecture
Vega20
Lithography
TSMC 7nm FinFET
Stream Processors
3840
Compute Units
60
Peak Engine Clock
1746 MHz
Peak Half Precision (FP16) Performance
26.8 TFLOPs
Peak Single Precision (FP32) Performance
13.4 TFLOPs
Peak Double Precision (FP64) Performance
6.7 TFLOPs
Peak INT8 Performance
53.6 TFLOPs
OS Support
Linux x86_64
GPU Memory
Memory Size
16 GB
Memory Type (GPU)
HBM2
Memory Interface
4096-bit
Memory Clock
1 GHz
Memory Bandwidth
1024 GB/s
Memory ECC Support
Yes (Full-Chip)
Board Specifications
Form Factor
PCIe Add-in Card
Bus Type
PCIe 4.0 x16
PCIe 3.0 x16
TDP
300W
Cooling
Passive
Board Width
Double Slot
Board Length
10.5" (267 mm)
Board Height
Full Height
External Power Connectors
1x PCIe 6-pin, 1x PCIe 8-pin
Additional Features
RAS Support
Yes
Software API Support
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
11/18/2018