GPU model | GP102 | GV100 |
---|---|---|
Architecture | Pascal | Volta |
Foundry | TSMC | TSMC |
Manufacturing process | 16 nm | 12 nm |
Number of transistors | 11,800 million | 21,100 million |
Die size | 471 mm² | 815 mm² |
Release date | Sep 13th, 2016 | Jun 21st, 2017 |
---|---|---|
Generation | Tesla | Tesla |
Production | End-of-life | End-of-life |
Launch price | 5,699 USD | — |
Interface | PCIe 3.0 x16 | PCIe 3.0 x16 |
Base clock | 1303 MHz | 1245 MHz |
---|---|---|
Boost clock | 1531 MHz | 1380 MHz |
Memory clock | 1808 MHz 14.5 Gbps effective | 876 MHz 1752 Mbps effective |
Memory size | 24 GB | 16 GB |
---|---|---|
Memory type | GDDR5X | HBM2 |
Memory bus width | 384 bit | 4096 bit |
Bandwidth | 694.3 GB/s | 897.0 GB/s |
Shading units | 3840 | 5120 |
---|---|---|
TMUs | 240 | 320 |
ROPs | 96 | 128 |
SM count | 30 | 80 |
Cache L1 | 48 KB (per SM) | 128 KB (per SM) |
Cache L2 | 3 MB | 6 MB |
Tensor cores | — | 640 |
Pixel rate | 147.0 GPixel/s | 176.6 GPixel/s |
---|---|---|
Texture rate | 367.4 GTexel/s | 441.6 GTexel/s |
FP16 (half) performance | 183.7 GFLOPS (1:64) | 28.26 TFLOPS (2:1) |
FP32 (float) performance | 11.76 TFLOPS | 14.13 TFLOPS |
FP64 (double) performance | 367.4 GFLOPS (1:32) | 7.066 TFLOPS (1:2) |
Slot width | Dual-slot | Dual-slot |
---|---|---|
Length | 267 mm 10.5 inches | — |
Width | 111 mm 4.4 inches | — |
TDP | 250 W | 300 W |
Suggested PSU | 600 W | 700 W |
Outputs | No outputs | No outputs |
Power connectors | 1x 6-pin + 1x 8-pin | 2x 8-pin |
Board number | PG610 SKU 200 | PG500 SKU 200 |
DirectX | 12 (12_1) | 12 (12_1) |
---|---|---|
OpenGL | 4.6 | 4.6 |
OpenCL | 3.0 | 3.0 |
Vulkan | 1.2 | 1.2 |
CUDA | 6.1 | 7.0 |
Shader model | 6.4 | 6.6 |