GPU model | GP104 | GA102 |
---|---|---|
GPU variant | GP104-850-A1 | — |
Architecture | Pascal | Ampere |
Foundry | TSMC | Samsung |
Manufacturing process | 16 nm | 8 nm |
Number of transistors | 7,200 million | 28,300 million |
Die size | 314 mm² | 628 mm² |
Release date | Feb 6th, 2017 | Nov 23rd, 2021 |
---|---|---|
Generation | Quadro | Quadro |
Production | End-of-life | Active |
Launch price | 815 USD | — |
Interface | PCIe 3.0 x16 | PCIe 4.0 x16 |
Base clock | 1202 MHz | 1050 MHz |
---|---|---|
Boost clock | 1480 MHz | 1650 MHz |
Memory clock | 1901 MHz 7.6 Gbps effective | 2000 MHz 16 Gbps effective |
Memory size | 8 GB | 20 GB |
---|---|---|
Memory type | GDDR5 | GDDR6 |
Memory bus width | 256 bit | 320 bit |
Bandwidth | 243.3 GB/s | 640.0 GB/s |
Shading units | 1792 | 7168 |
---|---|---|
TMUs | 112 | 224 |
ROPs | 64 | 96 |
SM count | 14 | 56 |
Cache L1 | 48 KB (per SM) | 128 KB (per SM) |
Cache L2 | 2 MB | 6 MB |
Tensor cores | — | 224 |
RT cores | — | 56 |
Pixel rate | 94.72 GPixel/s | 158.4 GPixel/s |
---|---|---|
Texture rate | 165.8 GTexel/s | 369.6 GTexel/s |
FP16 (half) performance | 82.88 GFLOPS (1:64) | 23.65 TFLOPS (1:1) |
FP32 (float) performance | 5.304 TFLOPS | 23.65 TFLOPS |
FP64 (double) performance | 165.8 GFLOPS (1:32) | 739.2 GFLOPS (1:32) |
Slot width | Single-slot | Dual-slot |
---|---|---|
Length | 241 mm 9.5 inches | 267 mm 10.5 inches |
Width | 111 mm 4.4 inches | 112 mm 4.4 inches |
TDP | 105 W | 200 W |
Suggested PSU | 300 W | 550 W |
Outputs | 4x DisplayPort | 4x DisplayPort |
Power connectors | 1x 6-pin | 1x 8-pin |
Board number | PG410 SKU 501 | — |
DirectX | 12 (12_1) | 12 Ultimate (12_2) |
---|---|---|
OpenGL | 4.6 | 4.6 |
OpenCL | 3.0 | 3.0 |
Vulkan | 1.2 | 1.2 |
CUDA | 6.1 | 8.6 |
Shader model | 6.4 | 6.6 |