Nom du GPU | Navi 23 | GP102 |
---|---|---|
GPU variant | Navi 23 XL (215-130000016) | GP102-875-A1 |
Architecture | RDNA 2.0 | Pascal |
Fonderie | TSMC | TSMC |
Taille du processus | 7 nm | 16 nm |
Transistors | 11,060 million | 11,800 million |
Taille des matrices | 237 mm² | 471 mm² |
Date de publication | Oct 13th, 2021 | Oct 1st, 2016 |
---|---|---|
Génération | Navi II | Quadro |
Prédécesseur | Navi | — |
Production | Active | End-of-life |
Prix de lancement | 329 USD | 5,999 USD |
Bus interface | PCIe 4.0 x8 | PCIe 3.0 x16 |
Successeur | Navi III | — |
Base clock | 1626 MHz | 1506 MHz |
---|---|---|
Boost clock | 2491 MHz | 1645 MHz |
Mémoire clock | 1750 MHz 14 Gbps effective | 1127 MHz 9 Gbps effective |
Game clock | 2044 MHz | — |
Taille de la mémoire | 8 GB | 24 GB |
---|---|---|
Type de mémoire | GDDR6 | GDDR5X |
Bus mémoire | 128 bit | 384 bit |
Bande passante | 224.0 GB/s | 432.8 GB/s |
Shading units | 1792 | 3840 |
---|---|---|
TMUs | 112 | 240 |
ROPs | 64 | 96 |
Compute units | 28 | — |
RT cores | 28 | — |
gpu.details.l0-cache | 32 KB per WGP | — |
Cache L1 | 128 KB per Array | 48 KB (per SM) |
Cache L2 | 2 MB | 3 MB |
gpu.details.l3-cache | 32 MB | — |
Comptage SM | — | 30 |
Taux de pixel | 159.4 GPixel/s | 157.9 GPixel/s |
---|---|---|
Taux de texture | 279.0 GTexel/s | 394.8 GTexel/s |
FP16 (half) performance | 17.86 TFLOPS (2:1) | 197.4 GFLOPS (1:64) |
FP32 (float) performance | 8.928 TFLOPS | 12.63 TFLOPS |
FP64 (double) performance | 558.0 GFLOPS (1:16) | 394.8 GFLOPS (1:32) |
Taille Slot | Dual-slot | Dual-slot |
---|---|---|
TDP | 132 W | 250 W |
Suggestions pour le PSU | 300 W | 600 W |
Sorties | 1x HDMI3x DisplayPort | 1x DVI4x DisplayPort |
Connecteurs de puissance | 1x 8-pin | 1x 6-pin + 1x 8-pin |
Longueur | 190 mm 7.5 inches | 267 mm 10.5 inches |
Taille | 110 mm 4.3 inches | 111 mm 4.4 inches |
Taille | 40 mm 1.6 inches | — |
Board number | — | PG611 SKU 500 |
DirectX | 12.0 Ultimate (12_2) | 12 (12_1) |
---|---|---|
OpenGL | 4.6 | 4.6 |
OpenCL | 2.1 | 3.0 |
Vulkan | 1.2 | 1.2 |
Shader model | 6.5 | 6.4 |
CUDA | — | 6.1 |