GPU model | GK20A | GV100 |
---|---|---|
Architecture | Kepler 2.0 | Volta |
Foundry | TSMC | TSMC |
Manufacturing process | 28 nm | 12 nm |
Number of transistors | unknown | 21,100 million |
Die size | unknown | 815 mm² |
Release date | Oct 15th, 2014 | — |
---|---|---|
Generation | Tegra | — |
Production | End-of-life | — |
Interface | IGP | — |
Base clock | 756 MHz | 1230 MHz |
---|---|---|
Boost clock | 951 MHz | 1380 MHz |
Memory clock | System Shared | 876 MHz 1752 Mbps effective |
Memory size | System Shared | 32 GB |
---|---|---|
Memory type | System Shared | HBM2 |
Memory bus width | System Shared | 4096 bit |
Bandwidth | System Dependent | 897.0 GB/s |
Shading units | 192 | 5120 |
---|---|---|
TMUs | 8 | 320 |
ROPs | 4 | 128 |
SMX count | 1 | — |
Cache L1 | 16 KB (per SMX) | 128 KB (per SM) |
Cache L2 | 128 KB | 6 MB |
SM count | — | 80 |
Tensor cores | — | 640 |
Pixel rate | 3.804 GPixel/s | 176.6 GPixel/s |
---|---|---|
Texture rate | 7.608 GTexel/s | 441.6 GTexel/s |
FP32 (float) performance | 365.2 GFLOPS | 14.13 TFLOPS |
FP64 (double) performance | 11.41 GFLOPS (1:32) | 7.066 TFLOPS (1:2) |
FP16 (half) performance | — | 28.26 TFLOPS (2:1) |
Slot width | IGP | Dual-slot |
---|---|---|
TDP | 8 W | 250 W |
Outputs | No outputs | No outputs |
Suggested PSU | — | 600 W |
Power connectors | — | 2x 8-pin |
Board number | — | PG500 SKU 200 |
DirectX | 12 (11_0) | 12 (12_1) |
---|---|---|
OpenGL | ES 3.1 | 4.6 |
OpenCL | — | 3.0 |
Vulkan | 1.1 | 1.2 |
CUDA | 3.2 | 7.0 |
Shader model | 5.1 | 6.6 |
Release date | — | Mar 27th, 2018 |
---|---|---|
Generation | — | Tesla |
Production | — | End-of-life |
Interface | — | PCIe 3.0 x16 |