Nom du GPU | GV10B | NV28 |
---|---|---|
Architecture | Volta | Kelvin |
Fonderie | TSMC | TSMC |
Taille du processus | 12 nm | 150 nm |
Transistors | 9,000 million | 36 million |
Taille des matrices | 350 mm² | 101 mm² |
GPU variant | — | NV28GL A2 |
Date de publication | May 14th, 2020 | — |
---|---|---|
Génération | Tegra | — |
Production | End-of-life | — |
Prix de lancement | 399 USD | — |
Bus interface | IGP | — |
Base clock | 854 MHz | — |
---|---|---|
Boost clock | 1377 MHz | — |
Mémoire clock | System Shared | 325 MHz 650 Mbps effective |
GPU clock | — | 300 MHz |
Taille de la mémoire | System Shared | 128 MB |
---|---|---|
Type de mémoire | System Shared | DDR |
Bus mémoire | System Shared | 128 bit |
Bande passante | System Dependent | 10.40 GB/s |
Shading units | 384 | — |
---|---|---|
TMUs | 24 | 8 |
ROPs | 16 | 4 |
Comptage SM | 6 | — |
Tensor cores | 48 | — |
Cache L1 | 128 KB (per SM) | — |
Cache L2 | 256 KB | — |
Pixel shaders | — | 4 |
Vertex shaders | — | 2 |
Taux de pixel | 22.03 GPixel/s | 1.200 GPixel/s |
---|---|---|
Taux de texture | 33.05 GTexel/s | 2.400 GTexel/s |
FP16 (half) performance | 2.115 TFLOPS (2:1) | — |
FP32 (float) performance | 1,058 GFLOPS | — |
FP64 (double) performance | 528.8 GFLOPS (1:2) | — |
Vertex rate | — | 150.0 MVertices/s |
Taille Slot | IGP | Single-slot |
---|---|---|
Longueur | 70 mm 2.8 inches | — |
Taille | 45 mm 1.8 inches | — |
TDP | 15 W | unknown |
Sorties | No outputs | 2x DVI1x S-Video |
Suggestions pour le PSU | — | 200 W |
Connecteurs de puissance | — | None |
DirectX | 12 (12_1) | 8.1 |
---|---|---|
OpenGL | 4.6 | 1.5 |
OpenCL | 1.2 | — |
Vulkan | 1.2 | — |
CUDA | 7.2 | — |
Shader model | 6.4 | — |
Pixel shader | — | 1.3 |
Vertex shader | — | 1.1 |
Date de publication | — | Nov 12th, 2002 |
---|---|---|
Génération | — | Quadro4 XGL |
Production | — | End-of-life |
Bus interface | — | AGP 8x |
Critiques | — | 26 in our database |