Nom du GPU | Aldebaran | Vega 10 |
---|---|---|
GPU variant | Aldebaran | Vega 10 XT GL (215-0894124) |
Architecture | CDNA 2.0 | GCN 5.0 |
Fonderie | TSMC | GlobalFoundries |
Taille du processus | 6 nm | 14 nm |
Transistors | 58,200 million | 12,500 million |
Taille des matrices | unknown | 495 mm² |
Date de publication | Dec 2021 | Aug 8th, 2017 |
---|---|---|
Génération | Radeon Instinct | Radeon Pro |
Production | Active | Active |
Bus interface | PCIe 4.0 x16 | PCIe 3.0 x16 |
Prix de lancement | — | 6,999 USD |
Base clock | 1000 MHz | 1440 MHz |
---|---|---|
Boost clock | 1700 MHz | 1500 MHz |
Mémoire clock | 1600 MHz 3.2 Gbps effective | 945 MHz 1890 Mbps effective |
Taille de la mémoire | 64 GB | 16 GB |
---|---|---|
Type de mémoire | HBM2e | HBM2 |
Bus mémoire | 4096 bit | 2048 bit |
Bande passante | 1,638 GB/s | 483.8 GB/s |
Shading units | 6656 | 4096 |
---|---|---|
TMUs | 416 | 256 |
ROPs | 0 | 64 |
Compute units | 104 | 64 |
gpu.details.mcm | 1 | — |
Cache L1 | 16 KB (per CU) | 16 KB (per CU) |
Cache L2 | 16 MB | 4 MB |
Taux de pixel | 0 MPixel/s | 96.00 GPixel/s |
---|---|---|
Taux de texture | 707.2 GTexel/s | 384.0 GTexel/s |
FP16 (half) performance | 181.0 TFLOPS (8:1) | 24.58 TFLOPS (2:1) |
FP32 (float) performance | 22.63 TFLOPS | 12.29 TFLOPS |
FP64 (double) performance | 22.63 TFLOPS (1:1) | 768.0 GFLOPS (1:16) |
Taille Slot | Dual-slot | Dual-slot |
---|---|---|
Longueur | 267 mm 10.5 inches | 267 mm 10.5 inches |
Taille | 111 mm 4.4 inches | 111 mm 4.4 inches |
TDP | 300 W | 260 W |
Suggestions pour le PSU | 700 W | 600 W |
Sorties | No outputs | 6x mini-DisplayPort |
Connecteurs de puissance | 2x 8-pin | 1x 6-pin + 1x 8-pin |
DirectX | — | 12 (12_1) |
---|---|---|
OpenGL | — | 4.6 |
OpenCL | 3.0 | 2.1 |
Vulkan | — | 1.2 |
Shader model | — | 6.4 |