Nom du GPU | Scrapper | GA100 |
---|---|---|
Architecture | TeraScale 3 | Ampere |
Fonderie | TSMC | TSMC |
Taille du processus | 32 nm | 7 nm |
Transistors | 1,303 million | 54,200 million |
Taille des matrices | 246 mm² | 826 mm² |
Date de publication | May 23rd, 2013 | — |
---|---|---|
Génération | Richland (HD 8000) | — |
Production | End-of-life | — |
Bus interface | IGP | — |
Prédécesseur | Trinity | — |
Successeur | Kabini | — |
Base clock | 720 MHz | 900 MHz |
---|---|---|
Boost clock | 844 MHz | 1005 MHz |
Mémoire clock | System Shared | 1215 MHz 2.4 Gbps effective |
Taille de la mémoire | System Shared | 48 GB |
---|---|---|
Type de mémoire | System Shared | HBM2E |
Bus mémoire | System Shared | 6144 bit |
Bande passante | System Dependent | 1,866 GB/s |
Shading units | 256 | 6912 |
---|---|---|
TMUs | 16 | 432 |
ROPs | 8 | 192 |
Compute units | 4 | — |
Comptage SM | — | 108 |
Tensor cores | — | 432 |
Cache L1 | — | 192 KB (per SM) |
Cache L2 | — | 48 MB |
Taux de pixel | 6.752 GPixel/s | 193.0 GPixel/s |
---|---|---|
Taux de texture | 13.50 GTexel/s | 434.2 GTexel/s |
FP32 (float) performance | 432.1 GFLOPS | 13.89 TFLOPS |
FP16 (half) performance | — | 55.57 TFLOPS (4:1) |
FP64 (double) performance | — | 6.947 TFLOPS (1:2) |
Taille Slot | IGP | IGP |
---|---|---|
TDP | 65 W | 400 W |
Sorties | No outputs | No outputs |
Suggestions pour le PSU | — | 800 W |
Connecteurs de puissance | — | None |
DirectX | 11.2 (11_0) | — |
---|---|---|
OpenGL | 4.4 | — |
OpenCL | 1.2 | 3.0 |
Vulkan | — | — |
Shader model | 5.0 | — |
CUDA | — | 8.0 |
Date de publication | — | May 14th, 2020 |
---|---|---|
Génération | — | GRID |
Production | — | Active |
Bus interface | — | PCIe 4.0 x16 |