Nom du GPU | GA104 | GA102 |
---|---|---|
GPU variant | GA104-401-A1 | GA102-300-A1 |
Architecture | Ampere | Ampere |
Fonderie | Samsung | TSMC |
Taille du processus | 8 nm | 7 nm |
Transistors | 17,400 million | 40,000 million |
Taille des matrices | 392 mm² | 627 mm² |
Date de publication | Jan 2022 | 2020 |
---|---|---|
gpu.details.availability | Jan 2022 | — |
Génération | GeForce 30 | GeForce 30 |
Prédécesseur | GeForce 20 | — |
Production | Unreleased | Unreleased |
Prix de lancement | 699 USD | 1,399 USD |
Bus interface | PCIe 4.0 x16 | PCIe 4.0 x16 |
Critiques | 43 in our database | — |
Base clock | 1575 MHz | 1410 MHz |
---|---|---|
Boost clock | 1770 MHz | 1740 MHz |
Mémoire clock | 1188 MHz 19 Gbps effective | 2375 MHz 19000 MHz effective |
Taille de la mémoire | 16 GB | 12 GB |
---|---|---|
Type de mémoire | GDDR6X | GDDR6X |
Bus mémoire | 256 bit | 384 bit |
Bande passante | 608.3 GB/s | 912.0 GB/s |
Shading units | 6144 | 5120 |
---|---|---|
TMUs | 192 | 320 |
ROPs | 96 | 96 |
Comptage SM | 48 | 80 |
Tensor cores | 192 | 640 |
RT cores | 48 | 160 |
Cache L1 | 128 KB (per SM) | 192 KB (per SM) |
Cache L2 | 4 MB | 6 MB |
Taux de pixel | 169.9 GPixel/s | 167.0 GPixel/s |
---|---|---|
Taux de texture | 339.8 GTexel/s | 556.8 GTexel/s |
FP16 (half) performance | 21.75 TFLOPS (1:1) | 35.64 TFLOPS (2:1) |
FP32 (float) performance | 21.75 TFLOPS | 17.82 TFLOPS |
FP64 (double) performance | 339.8 GFLOPS (1:64) | 556.8 GFLOPS (1:32) |
Taille Slot | Dual-slot | Dual-slot |
---|---|---|
Longueur | 267 mm 10.5 inches | 267 mm 10.5 inches |
Taille | 112 mm 4.4 inches | — |
TDP | 290 W | 320 W |
Suggestions pour le PSU | 600 W | 700 W |
Sorties | 1x HDMI3x DisplayPort | 1x HDMI3x DisplayPort1x USB Type-C |
Connecteurs de puissance | 1x 12-pin | 1x 12-pin |
Board number | — | PG132 SKU 20 |
DirectX | 12 Ultimate (12_2) | 12 Ultimate (12_2) |
---|---|---|
OpenGL | 4.6 | 4.6 |
OpenCL | 3.0 | 2.0 |
Vulkan | 1.2 | 1.2.140 |
CUDA | 8.6 | 8.5 |
Shader model | 6.6 | 6.5 |