Radeon Vega 7nm

The World’s First “Vega 7nm” GPU

Radeon MI60 versus MI25

Improved Performance Per Watt

The Industry’s first 7nm GPU design dramatically improves performance per watt over previous generation products1

autonomous driving car

Optimized Deep Learning Operations

For Training and Inference

Optimized Deep Learning Operations with mixed FP16, FP32 & INT8 capabilities brings customers supercharged compute capabilities to meet today’s demanding system requirements of handling large data efficiently for training complex neural networks used in deep learning and running inference against them.

Ten Tech aerodynamics test

World’s Leading Double Precision PCIe® Accelerator

Ideal for HPC Workloads

Ultra-fast double precision performance with up to 7.4 TFLOPS FP64 performance on the AMD Radeon Instinct™ MI60 Compute GPU, allows 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.

Radeon Instinct MI50

AMD Infinity Fabric™ Link

Advanced Peer-to-Peer GPU Communications

Two Infinity Fabric™ Links per GPU for high speed Direct-Connect GPU clusters delivering up to 200 GB/s GPU peer-to-peer bandwidth.           

Technologies for Driving the Next Era of Compute

PCIe® Gen 4 Capable

Ultra-fast HBM2 Memory

1TB/s memory bandwidth

Full Chip ECC and Enterprise-class RAS

Advanced System GPU Monitoring

MxGPU Virtualized Compute

Large Bar Support

ROCm Open Ecosystem

Radeon Instinct MI60

Radeon Instinct™ MI60 Accelerator 

Unleash Discovery on the World’s Fastest Double Precision PCIe® Accelerator1

Radeon Instinct™ MI50 Accelerator 

Unleash Deep Learning Discovery 

Footnotes
  1. Measurements conducted by AMD Performance Labs as of Oct 16, 2018 on the Radeon Instinct™ MI50 and MI60 “Vega20” architecture based accelerators.  Results are estimates only and may vary. Performance may vary based on use of latest drivers. PC/system manufacturers may vary configurations yielding different results. The results calculated for Radeon Instinct MI60 resulted in 29.5 TFLOPS peak half precision (FP16), 14.75 TFLOPS peak single precision (FP32) and 7.3 TFLOPS peak double precision (FP64) floating-point performance. The results calculated for Radeon Instinct MI50 resulted in 24.96 TFLOPS peak half precision (FP16), 12.48 TFLOPS peak single precision (FP32) and 6.24 TFLOPS peak double precision (FP64) floating-point performance. AMD TFLOPS calculations conducted with the following equation for Radeon Instinct MI50 | MI60 GPU: FLOPS calculations are performed by taking the engine clock from the highest DPM state and multiplying it by xx CUs per GPU. Then, multiplying that number by xx stream processors, which exist in each CU. Then, that number is multiplied by 2 FLOPS per clock for FP32. To calculate TFLOPS for FP16, 4 FLOPS per clock were used.  The FP64 TFLOPS rate is calculated using 1/2 rate. The results calculated for Radeon Instinct MI25 resulted in 24.6 TFLOPS peak half precision (FP16), 12.3 TFLOPS peak single precision (FP32) and 768 GFLOPS peak double precision (FP64) floating-point performance. AMD TFLOPS calculations conducted with the following equation for Radeon Instinct MI25 GPU: FLOPS calculations are performed by taking the engine clock from the highest DPM state and multiplying it by xx CUs per GPU. Then, multiplying that number by xx stream processors, which exist in each CU. Then, that number is multiplied by 2 FLOPS per clock for FP32. To calculate TFLOPS for FP16, 4 FLOPS per clock were used.  The FP64 TFLOPS rate is calculated using 1/16th rate. Industry supporting documents / web pages: http://www.tsmc.com/english/dedicatedFoundry/technology/7nm.htm https://www.globalfoundries.com/sites/default/files/product-briefs/product-brief-7lp-7nm-finfet-technology.pdf AMD has not independently tested or verified external/third party results/data and bears no responsibility for any errors or omissions therein.