Nom du GPU | GP108 | TU106 |
---|---|---|
GPU variant | GP108-300-A1 | TU106-125-A1 |
Architecture | Pascal | Turing |
Fonderie | Samsung | TSMC |
Taille du processus | 14 nm | 12 nm |
Transistors | 1,800 million | 10,800 million |
Taille des matrices | 74 mm² | 445 mm² |
Date de publication | May 17th, 2017 | Jun 18th, 2020 |
---|---|---|
Génération | GeForce 10 | GeForce 16 |
Prédécesseur | GeForce 900 | GeForce 10 |
Successeur | GeForce 20 | GeForce 20 |
Production | End-of-life | Active |
Prix de lancement | 79 USD | — |
Bus interface | PCIe 3.0 x4 | PCIe 3.0 x16 |
Critiques | 6 in our database | — |
Base clock | 1228 MHz | 1410 MHz |
---|---|---|
Boost clock | 1468 MHz | 1590 MHz |
Mémoire clock | 1502 MHz 6 Gbps effective | 1500 MHz 12 Gbps effective |
Taille de la mémoire | 2 GB | 4 GB |
---|---|---|
Type de mémoire | GDDR5 | GDDR6 |
Bus mémoire | 64 bit | 128 bit |
Bande passante | 48.06 GB/s | 192.0 GB/s |
Shading units | 384 | 896 |
---|---|---|
TMUs | 24 | 56 |
ROPs | 16 | 32 |
Comptage SM | 3 | 14 |
Cache L1 | 48 KB (per SM) | 64 KB (per SM) |
Cache L2 | 512 KB | 1024 KB |
Tensor cores | — | 112 |
RT cores | — | 14 |
Taux de pixel | 23.49 GPixel/s | 50.88 GPixel/s |
---|---|---|
Taux de texture | 35.23 GTexel/s | 89.04 GTexel/s |
FP16 (half) performance | 17.62 GFLOPS (1:64) | 5.699 TFLOPS (2:1) |
FP32 (float) performance | 1,127 GFLOPS | 2.849 TFLOPS |
FP64 (double) performance | 35.23 GFLOPS (1:32) | 89.04 GFLOPS (1:32) |
Taille Slot | Single-slot | Dual-slot |
---|---|---|
Longueur | 145 mm 5.7 inches | 229 mm 9 inches |
Taille | 69 mm 2.7 inches | 111 mm 4.4 inches |
Taille | 18 mm 0.7 inches | 35 mm 1.4 inches |
TDP | 30 W | 90 W |
Suggestions pour le PSU | 200 W | 250 W |
Sorties | 1x DVI1x HDMI | 1x DVI1x HDMI1x DisplayPort |
Connecteurs de puissance | None | 1x 6-pin |
Board number | PG110 SKU 0 | — |
DirectX | 12 (12_1) | 12 Ultimate (12_2) |
---|---|---|
OpenGL | 4.6 | 4.6 |
OpenCL | 3.0 | 3.0 |
Vulkan | 1.2 | 1.2 |
CUDA | 6.1 | 7.5 |
Shader model | 6.4 | 6.6 |