GPU model | Vega 20 | GA100 |
---|---|---|
GPU variant | Vega 20 GLXT | — |
Architecture | GCN 5.1 | Ampere |
Foundry | TSMC | TSMC |
Manufacturing process | 7 nm | 7 nm |
Number of transistors | 13,230 million | 54,200 million |
Die size | 331 mm² | 826 mm² |
Release date | Nov 18th, 2018 | Apr 12th, 2021 |
---|---|---|
Generation | Radeon Instinct | Tesla |
Production | Active | Active |
Interface | PCIe 4.0 x16 | PCIe 4.0 x16 |
gpu.details.availability | — | 2021 |
Base clock | 1200 MHz | 930 MHz |
---|---|---|
Boost clock | 1746 MHz | 1440 MHz |
Memory clock | 1000 MHz 2 Gbps effective | 1215 MHz 2.4 Gbps effective |
Memory size | 16 GB | 24 GB |
---|---|---|
Memory type | HBM2 | HBM2 |
Memory bus width | 4096 bit | 3072 bit |
Bandwidth | 1,024 GB/s | 933.1 GB/s |
Shading units | 3840 | 3584 |
---|---|---|
TMUs | 240 | 224 |
ROPs | 64 | 96 |
Compute units | 60 | — |
Cache L1 | 16 KB (per CU) | 192 KB (per SM) |
Cache L2 | 4 MB | 24 MB |
SM count | — | 56 |
Tensor cores | — | 224 |
Pixel rate | 111.7 GPixel/s | 138.2 GPixel/s |
---|---|---|
Texture rate | 419.0 GTexel/s | 322.6 GTexel/s |
FP16 (half) performance | 26.82 TFLOPS (2:1) | 10.32 TFLOPS (1:1) |
FP32 (float) performance | 13.41 TFLOPS | 10.32 TFLOPS |
FP64 (double) performance | 6.705 TFLOPS (1:2) | 5.161 TFLOPS (1:2) |
Slot width | Dual-slot | Dual-slot |
---|---|---|
Length | 267 mm 10.5 inches | 267 mm 10.5 inches |
TDP | 300 W | 165 W |
Suggested PSU | 700 W | 450 W |
Outputs | 1x mini-DisplayPort | No outputs |
Power connectors | 1x 6-pin + 1x 8-pin | 8-pin EPS |
Width | — | 112 mm 4.4 inches |
Board number | — | P1001 SKU 205 |
DirectX | 12 (12_1) | — |
---|---|---|
OpenGL | 4.6 | — |
OpenCL | 2.1 | 3.0 |
Vulkan | 1.2 | — |
Shader model | 6.4 | — |
CUDA | — | 8.0 |