GPU model | GF108 | GV100 |
---|---|---|
Architecture | Fermi | Volta |
Foundry | TSMC | TSMC |
Manufacturing process | 40 nm | 12 nm |
Number of transistors | 585 million | 21,100 million |
Die size | 116 mm² | 815 mm² |
Release date | Dec 1st, 2011 | — |
---|---|---|
Generation | GeForce 600M | — |
Production | End-of-life | — |
Interface | PCIe 2.0 x16 | — |
Predecessor | GeForce 500M | — |
Successor | GeForce 700M | — |
GPU clock | 672 MHz | — |
---|---|---|
Shader clock | 1344 MHz | — |
Memory clock | 800 MHz 1600 Mbps effective | 876 MHz 1752 Mbps effective |
Base clock | — | 1297 MHz |
Boost clock | — | 1530 MHz |
Memory size | 1024 MB | 32 GB |
---|---|---|
Memory type | DDR3 | HBM2 |
Memory bus width | 64 bit | 4096 bit |
Bandwidth | 12.80 GB/s | 897.0 GB/s |
Shading units | 48 | 5120 |
---|---|---|
TMUs | 8 | 320 |
ROPs | 2 | 128 |
SM count | 1 | 80 |
Cache L1 | 64 KB (per SM) | 128 KB (per SM) |
Cache L2 | 128 KB | 6 MB |
Tensor cores | — | 640 |
Pixel rate | 1.344 GPixel/s | 195.8 GPixel/s |
---|---|---|
Texture rate | 5.376 GTexel/s | 489.6 GTexel/s |
FP32 (float) performance | 129.0 GFLOPS | 15.67 TFLOPS |
FP64 (double) performance | 10.75 GFLOPS (1:12) | 7.834 TFLOPS (1:2) |
FP16 (half) performance | — | 31.33 TFLOPS (2:1) |
TDP | 12 W | 250 W |
---|---|---|
Outputs | No outputs | No outputs |
Slot width | — | Dual-slot |
Suggested PSU | — | 600 W |
Power connectors | — | None |
DirectX | 12 (11_0) | 12 (12_1) |
---|---|---|
OpenGL | 4.6 | 4.6 |
OpenCL | 1.1 | 3.0 |
Vulkan | — | 1.2 |
CUDA | 2.1 | 7.0 |
Shader model | 5.1 | 6.6 |
Release date | — | Mar 27th, 2018 |
---|---|---|
Generation | — | Tesla |
Production | — | End-of-life |
Interface | — | PCIe 3.0 x16 |