GPUDatacenterReleased Nov 6, 2018

AMD Radeon Instinct MI60

32 GB HBM2 · 4096-bit · 4,096 cores

The Radeon Instinct MI60 was AMD's flagship data center accelerator of 2018 and, together with the MI50, among the first GPUs built on a 7 nm process. AMD presented it as the first PCI Express 4.0 capable GPU, and a pair of Infinity Fabric Links let accelerators talk to each other directly inside a server rather than going through the host bus. AMD aimed it at machine learning, high performance computing and cloud rendering through the open-source ROCm software stack. The MI50 and MI60 were the last accelerators to carry the Radeon Instinct name before the line moved to the CDNA architecture as AMD Instinct.

General

Manufacturer: AMD
Model: Radeon Instinct MI60
Segment: Datacenter
GPU chip: Vega 20
Architecture: GCN 5 (Vega)
Release date: Nov 6, 2018
Price: Search on Amazon

Memory

Memory size: 32 GB
Memory type: HBM2
Bus width: 4096-bit
Bandwidth: 1,024 GB/s
Memory clock: 1,000 MHz

Clocks & silicon

GPU clock: 1,500 MHz
Boost clock: 1,800 MHz
Shader units: 4,096 Stream Processors (64 CU)
Tensor / AI cores: —
RT cores: —
Transistors: 13.2 billion
Technology (nm): 7 nm
TSMC N7
TDP / TBP: 300 W

Interface & APIs

Bus interface: PCIe 4.0 x16
SLI / multi-GPU: AMD Infinity Fabric Link
DirectX: —
OpenGL: 4.6
Vulkan: 1.0
Other APIs: OpenCL 2.0, ROCm (HIP)

AI support

Precisions: FP8FP16BF16INT8INT4TF32FP64
AI accelerator: None (no matrix units); Rapid Packed Math with INT8/INT4 dot-product instructions
Peak AI TOPS: 59 TOPS
Notes: Peak vector throughput is 29.5 TFLOPS FP16, 14.7 TFLOPS FP32 and 7.4 TFLOPS FP64 at a 1:2 rate. INT8 and INT4 dot-product instructions accelerate inference. The 32 GB HBM2 frame buffer is ECC protected.

All AMD GPUs →

Datacenter gpus from AMD announced around the same time as the AMD Radeon Instinct MI60. Add any of them to your compare list to see the specs side by side.

gpu