Nom du GPU | Aldebaran | TU104 |
---|---|---|
GPU variant | Aldebaran | — |
Architecture | CDNA 2.0 | Turing |
Fonderie | TSMC | TSMC |
Taille du processus | 6 nm | 12 nm |
Transistors | 58,200 million | 13,600 million |
Taille des matrices | unknown | 545 mm² |
Date de publication | Dec 2021 | — |
---|---|---|
Génération | Radeon Instinct | — |
Production | Active | — |
Bus interface | PCIe 4.0 x16 | — |
Base clock | 1000 MHz | 780 MHz |
---|---|---|
Boost clock | 1700 MHz | 1380 MHz |
Mémoire clock | 1600 MHz 3.2 Gbps effective | 1625 MHz 13 Gbps effective |
Taille de la mémoire | 64 GB | 8 GB |
---|---|---|
Type de mémoire | HBM2e | GDDR6 |
Bus mémoire | 4096 bit | 256 bit |
Bande passante | 1,638 GB/s | 416.0 GB/s |
Shading units | 6656 | 2560 |
---|---|---|
TMUs | 416 | 160 |
ROPs | 0 | 64 |
Compute units | 104 | — |
gpu.details.mcm | 1 | — |
Cache L1 | 16 KB (per CU) | 64 KB (per SM) |
Cache L2 | 16 MB | 4 MB |
Comptage SM | — | 40 |
Tensor cores | — | 320 |
RT cores | — | 40 |
Taux de pixel | 0 MPixel/s | 88.32 GPixel/s |
---|---|---|
Taux de texture | 707.2 GTexel/s | 220.8 GTexel/s |
FP16 (half) performance | 181.0 TFLOPS (8:1) | 14.13 TFLOPS (2:1) |
FP32 (float) performance | 22.63 TFLOPS | 7.066 TFLOPS |
FP64 (double) performance | 22.63 TFLOPS (1:1) | 220.8 GFLOPS (1:32) |
Taille Slot | Dual-slot | IGP |
---|---|---|
Longueur | 267 mm 10.5 inches | — |
Taille | 111 mm 4.4 inches | — |
TDP | 300 W | 80 W |
Suggestions pour le PSU | 700 W | — |
Sorties | No outputs | No outputs |
Connecteurs de puissance | 2x 8-pin | None |
DirectX | — | 12 Ultimate (12_2) |
---|---|---|
OpenGL | — | 4.6 |
OpenCL | 3.0 | 3.0 |
Vulkan | — | 1.2 |
Shader model | — | 6.6 |
CUDA | — | 7.5 |
Date de publication | — | May 27th, 2019 |
---|---|---|
Génération | — | Quadro Mobile (Tx000) |
Production | — | Active |
Bus interface | — | PCIe 3.0 x16 |