GPU model | GA100 | GA102 |
---|---|---|
Architecture | Ampere | Ampere |
Foundry | TSMC | Samsung |
Manufacturing process | 7 nm | 8 nm |
Number of transistors | 54,200 million | 28,300 million |
Die size | 826 mm² | 628 mm² |
Release date | Nov 16th, 2020 | Apr 12th, 2021 |
---|---|---|
Generation | Tesla | Quadro |
Production | Active | Active |
Interface | PCIe 4.0 x16 | PCIe 4.0 x16 |
Base clock | 1275 MHz | 1170 MHz |
---|---|---|
Boost clock | 1410 MHz | 1695 MHz |
Memory clock | 1593 MHz 3.2 Gbps effective | 2000 MHz 16 Gbps effective |
Memory size | 80 GB | 24 GB |
---|---|---|
Memory type | HBM2e | GDDR6 |
Memory bus width | 5120 bit | 384 bit |
Bandwidth | 2,039 GB/s | 768.0 GB/s |
Shading units | 6912 | 8192 |
---|---|---|
TMUs | 432 | 256 |
ROPs | 160 | 96 |
SM count | 108 | 64 |
Tensor cores | 432 | 256 |
Cache L1 | 192 KB (per SM) | 128 KB (per SM) |
Cache L2 | 40 MB | 6 MB |
RT cores | — | 64 |
Pixel rate | 225.6 GPixel/s | 162.7 GPixel/s |
---|---|---|
Texture rate | 609.1 GTexel/s | 433.9 GTexel/s |
FP16 (half) performance | 77.97 TFLOPS (4:1) | 27.77 TFLOPS (1:1) |
FP32 (float) performance | 19.49 TFLOPS | 27.77 TFLOPS |
FP64 (double) performance | 9.746 TFLOPS (1:2) | 867.8 GFLOPS (1:32) |
Slot width | IGP | Dual-slot |
---|---|---|
TDP | 400 W | 230 W |
Suggested PSU | 800 W | 550 W |
Outputs | No outputs | 4x DisplayPort |
Power connectors | None | 1x 8-pin |
Length | — | 267 mm 10.5 inches |
Width | — | 112 mm 4.4 inches |
DirectX | — | 12 Ultimate (12_2) |
---|---|---|
OpenGL | — | 4.6 |
OpenCL | 3.0 | 3.0 |
Vulkan | — | 1.2 |
CUDA | 8.0 | 8.6 |
Shader model | — | 6.6 |