GPU model | GV100 | GP102 |
---|---|---|
GPU variant | GV100-400-A1 | GP102-450-A1 |
Architecture | Volta | Pascal |
Foundry | TSMC | TSMC |
Manufacturing process | 12 nm | 16 nm |
Number of transistors | 21,100 million | 11,800 million |
Die size | 815 mm² | 471 mm² |
Release date | Dec 7th, 2017 | Apr 6th, 2017 |
---|---|---|
Generation | GeForce 10 | GeForce 10 |
Predecessor | GeForce 900 | GeForce 900 |
Successor | GeForce 20 | GeForce 20 |
Production | End-of-life | Active |
Launch price | 2,999 USD | 1,199 USD |
Interface | PCIe 3.0 x16 | PCIe 3.0 x16 |
Reviews | 4 in our database | 9 in our database |
Base clock | 1200 MHz | 1405 MHz |
---|---|---|
Boost clock | 1455 MHz | 1582 MHz |
Memory clock | 848 MHz 1696 Mbps effective | 1426 MHz 11408 MHz effective |
Memory size | 12 GB | 12 GB |
---|---|---|
Memory type | HBM2 | GDDR5X |
Memory bus width | 3072 bit | 384 bit |
Bandwidth | 651.3 GB/s | 547.6 GB/s |
Shading units | 5120 | 3840 |
---|---|---|
TMUs | 320 | 240 |
ROPs | 96 | 96 |
SM count | 80 | 30 |
Tensor cores | 640 | — |
Cache L1 | 96 KB (per SM) | 48 KB (per SM) |
Cache L2 | 4.5 MB | 3 MB |
Pixel rate | 139.7 GPixel/s | 151.9 GPixel/s |
---|---|---|
Texture rate | 465.6 GTexel/s | 379.7 GTexel/s |
FP16 (half) performance | 29.80 TFLOPS (2:1) | 189.8 GFLOPS (1:64) |
FP32 (float) performance | 14.90 TFLOPS | 12.15 TFLOPS |
FP64 (double) performance | 7.450 TFLOPS (1:2) | 379.7 GFLOPS (1:32) |
Slot width | Dual-slot | Dual-slot |
---|---|---|
Length | 267 mm 10.5 inches | 267 mm 10.5 inches |
Width | 112 mm 4.4 inches | 112 mm 4.4 inches |
Height | 40 mm 1.6 inches | 40 mm 1.6 inches |
TDP | 250 W | 250 W |
Suggested PSU | 600 W | 600 W |
Outputs | 1x HDMI3x DisplayPort | 1x HDMI3x DisplayPort |
Power connectors | 1x 6-pin + 1x 8-pin | 1x 6-pin + 1x 8-pin |
Board number | PG500 SKU 0 | PG611 SKU 30 |
DirectX | 12 (12_1) | 12 (12_1) |
---|---|---|
OpenGL | 4.6 | 4.6 |
OpenCL | 3.0 | 1.2 |
Vulkan | 1.2 | 1.2.131 |
CUDA | 7.0 | 6.1 |
Shader model | 6.6 | 6.4 |