NVIDIA A100 PCIe vs NVIDIA A10G
Processeur graphique
| Nom du GPU | GA100 | GA102 |
|---|---|---|
| Architecture | Ampere | Ampere |
| Fonderie | TSMC | Samsung |
| Taille du processus | 7 nm | 8 nm |
| Transistors | 54,200 million | 28,300 million |
| Taille des matrices | 826 mm² | 628 mm² |
| GPU variant | — | GA102-890-A1 |
Carte graphique
| Date de publication | Jun 22nd, 2020 | Apr 12th, 2021 |
|---|---|---|
| Génération | Tesla | Tesla |
| Production | Active | Active |
| Bus interface | PCIe 4.0 x16 | PCIe 4.0 x16 |
| gpu.details.availability | — | 2021 |
Vitesse clock
| GPU clock | 1410 MHz | — |
|---|---|---|
| Mémoire clock | 1215 MHz 2.4 Gbps effective | 1563 MHz 12.5 Gbps effective |
| Base clock | 765 MHz | 1320 MHz |
| Boost clock | 1410 MHz | 1710 MHz |
Mémoire
| Taille de la mémoire | 40 GB | 12 GB |
|---|---|---|
| Type de mémoire | HBM2E | GDDR6 |
| Bus mémoire | 5120 bit | 384 bit |
| Bande passante | 1,555 GB/s | 600.2 GB/s |
Render Config
| Shading units | 6912 | 9216 |
|---|---|---|
| TMUs | 432 | 288 |
| ROPs | 160 | 96 |
| Comptage SM | 108 | 72 |
| Tensor cores | 432 | 288 |
| Cache L1 | 192 KB (per SM) | 128 KB (per SM) |
| Cache L2 | 40 MB | 6 MB |
| RT cores | — | 72 |
Performance théorique
| Taux de pixel | 225.6 GPixel/s | 164.2 GPixel/s |
|---|---|---|
| Taux de texture | 609.1 GTexel/s | 492.5 GTexel/s |
| FP16 (half) performance | 77.97 TFLOPS (4:1) | 31.52 TFLOPS (1:1) |
| FP32 (float) performance | 19.49 TFLOPS | 31.52 TFLOPS |
| FP64 (double) performance | 9.746 TFLOPS (1:2) | 985.0 GFLOPS (1:32) |
Board Design
| Taille Slot | Dual-slot | Single-slot |
|---|---|---|
| Longueur | 267 mm 10.5 inches | 267 mm 10.5 inches |
| TDP | 250 W | 150 W |
| Suggestions pour le PSU | 600 W | 450 W |
| Sorties | No outputs | No outputs |
| Connecteurs de puissance | 8-pin EPS | 8-pin EPS |
| Board number | P1001 SKU 200 | PG133 SKU 215 |
| Taille | — | 112 mm 4.4 inches |
Caractéristiques graphiques
| DirectX | — | 12 Ultimate (12_2) |
|---|---|---|
| OpenGL | — | 4.6 |
| OpenCL | 3.0 | 3.0 |
| Vulkan | — | 1.2 |
| CUDA | 8.0 | 8.6 |
| Shader model | — | 6.6 |
