GPU model | GA102 | GV100 |
---|---|---|
GPU variant | GA102-890-A1 | — |
Architecture | Ampere | Volta |
Foundry | Samsung | TSMC |
Manufacturing process | 8 nm | 12 nm |
Number of transistors | 28,300 million | 21,100 million |
Die size | 628 mm² | 815 mm² |
Release date | Apr 12th, 2021 | Mar 27th, 2018 |
---|---|---|
gpu.details.availability | 2021 | — |
Generation | Tesla | Tesla |
Production | Active | End-of-life |
Interface | PCIe 4.0 x16 | PCIe 3.0 x16 |
Base clock | 1320 MHz | 1230 MHz |
---|---|---|
Boost clock | 1710 MHz | 1380 MHz |
Memory clock | 1563 MHz 12.5 Gbps effective | 876 MHz 1752 Mbps effective |
Memory size | 12 GB | 32 GB |
---|---|---|
Memory type | GDDR6 | HBM2 |
Memory bus width | 384 bit | 4096 bit |
Bandwidth | 600.2 GB/s | 897.0 GB/s |
Shading units | 9216 | 5120 |
---|---|---|
TMUs | 288 | 320 |
ROPs | 96 | 128 |
SM count | 72 | 80 |
Tensor cores | 288 | 640 |
RT cores | 72 | — |
Cache L1 | 128 KB (per SM) | 128 KB (per SM) |
Cache L2 | 6 MB | 6 MB |
Pixel rate | 164.2 GPixel/s | 176.6 GPixel/s |
---|---|---|
Texture rate | 492.5 GTexel/s | 441.6 GTexel/s |
FP16 (half) performance | 31.52 TFLOPS (1:1) | 28.26 TFLOPS (2:1) |
FP32 (float) performance | 31.52 TFLOPS | 14.13 TFLOPS |
FP64 (double) performance | 985.0 GFLOPS (1:32) | 7.066 TFLOPS (1:2) |
Slot width | Single-slot | Dual-slot |
---|---|---|
Length | 267 mm 10.5 inches | — |
Width | 112 mm 4.4 inches | — |
TDP | 150 W | 250 W |
Suggested PSU | 450 W | 600 W |
Outputs | No outputs | No outputs |
Power connectors | 8-pin EPS | 2x 8-pin |
Board number | PG133 SKU 215 | PG500 SKU 200 |
DirectX | 12 Ultimate (12_2) | 12 (12_1) |
---|---|---|
OpenGL | 4.6 | 4.6 |
OpenCL | 3.0 | 3.0 |
Vulkan | 1.2 | 1.2 |
CUDA | 8.6 | 7.0 |
Shader model | 6.6 | 6.6 |