GPU model | TU102 | GP102 |
---|---|---|
GPU variant | TU102-875-A1 | — |
Architecture | Turing | Pascal |
Foundry | TSMC | TSMC |
Manufacturing process | 12 nm | 16 nm |
Number of transistors | 18,600 million | 11,800 million |
Die size | 754 mm² | 471 mm² |
Release date | Unknown | Sep 13th, 2016 |
---|---|---|
Generation | GRID | Tesla |
Production | Active | End-of-life |
Interface | PCIe 3.0 x16 | PCIe 3.0 x16 |
Launch price | — | 5,699 USD |
Base clock | 1065 MHz | 1303 MHz |
---|---|---|
Boost clock | 1395 MHz | 1531 MHz |
Memory clock | 1750 MHz 14 Gbps effective | 1808 MHz 14.5 Gbps effective |
Memory size | 8 GB | 24 GB |
---|---|---|
Memory type | GDDR6 | GDDR5X |
Memory bus width | 384 bit | 384 bit |
Bandwidth | 672.0 GB/s | 694.3 GB/s |
Shading units | 4608 | 3840 |
---|---|---|
TMUs | 288 | 240 |
ROPs | 96 | 96 |
SM count | 72 | 30 |
Tensor cores | 576 | — |
RT cores | 72 | — |
Cache L1 | 64 KB (per SM) | 48 KB (per SM) |
Cache L2 | 6 MB | 3 MB |
Pixel rate | 133.9 GPixel/s | 147.0 GPixel/s |
---|---|---|
Texture rate | 401.8 GTexel/s | 367.4 GTexel/s |
FP16 (half) performance | 25.71 TFLOPS (2:1) | 183.7 GFLOPS (1:64) |
FP32 (float) performance | 12.86 TFLOPS | 11.76 TFLOPS |
FP64 (double) performance | 401.8 GFLOPS (1:32) | 367.4 GFLOPS (1:32) |
Slot width | Dual-slot | Dual-slot |
---|---|---|
Length | 267 mm 10.5 inches | 267 mm 10.5 inches |
TDP | 260 W | 250 W |
Suggested PSU | 600 W | 600 W |
Outputs | No outputs | No outputs |
Power connectors | 1x 6-pin + 1x 8-pin | 1x 6-pin + 1x 8-pin |
Board number | PG150 SKU 220 | PG610 SKU 200 |
Width | — | 111 mm 4.4 inches |
DirectX | 12 Ultimate (12_2) | 12 (12_1) |
---|---|---|
OpenGL | 4.6 | 4.6 |
OpenCL | 3.0 | 3.0 |
Vulkan | 1.2 | 1.2 |
CUDA | 7.5 | 6.1 |
Shader model | 6.6 | 6.4 |