Nom du GPU | Gemini Lake GT1 | GA100 |
---|---|---|
Architecture | Generation 9.5 | Ampere |
Fonderie | Intel | TSMC |
Taille du processus | 14 nm | 7 nm |
Transistors | unknown | 54,200 million |
Taille des matrices | unknown | 826 mm² |
Date de publication | Dec 11th, 2017 | — |
---|---|---|
Génération | HD Graphics-T (Goldmont Plus) | — |
Production | Active | — |
Bus interface | Ring Bus | — |
Critiques | 1 in our database | — |
Base clock | 200 MHz | 1065 MHz |
---|---|---|
Boost clock | 650 MHz | 1410 MHz |
Mémoire clock | System Shared | 1593 MHz 3.2 Gbps effective |
Taille de la mémoire | System Shared | 80 GB |
---|---|---|
Type de mémoire | System Shared | HBM2e |
Bus mémoire | System Shared | 5120 bit |
Bande passante | System Dependent | 2,039 GB/s |
Shading units | 96 | 6912 |
---|---|---|
TMUs | 12 | 432 |
ROPs | 2 | 160 |
Unités d'exécution | 12 | — |
Comptage SM | — | 108 |
Tensor cores | — | 432 |
Cache L1 | — | 192 KB (per SM) |
Cache L2 | — | 80 MB |
Taux de pixel | 1.300 GPixel/s | 225.6 GPixel/s |
---|---|---|
Taux de texture | 7.800 GTexel/s | 609.1 GTexel/s |
FP16 (half) performance | 249.6 GFLOPS (2:1) | 77.97 TFLOPS (4:1) |
FP32 (float) performance | 124.8 GFLOPS | 19.49 TFLOPS |
FP64 (double) performance | 31.20 GFLOPS (1:4) | 9.746 TFLOPS (1:2) |
Taille Slot | IGP | Dual-slot |
---|---|---|
TDP | 5 W | 250 W |
Sorties | No outputs | No outputs |
Longueur | — | 267 mm 10.5 inches |
Suggestions pour le PSU | — | 600 W |
Connecteurs de puissance | — | 8-pin EPS |
Board number | — | P1001 SKU 200 |
DirectX | 12 (12_1) | — |
---|---|---|
OpenGL | 4.6 | — |
OpenCL | 3.0 | 3.0 |
Vulkan | 1.2 | — |
Shader model | 6.4 | — |
CUDA | — | 8.0 |
Date de publication | — | Jun 28th, 2021 |
---|---|---|
Génération | — | Tesla |
Production | — | Active |
Bus interface | — | PCIe 4.0 x16 |