AMD Radeon 550 vs NVIDIA GRID A100A
Processeur graphique
| Nom du GPU | Lexa | GA100 |
|---|---|---|
| GPU variant | Lexa PRO (215-0904018) | — |
| Architecture | GCN 4.0 | Ampere |
| Fonderie | GlobalFoundries | TSMC |
| Taille du processus | 14 nm | 7 nm |
| Transistors | 2,200 million | 54,200 million |
| Taille des matrices | 103 mm² | 826 mm² |
Carte graphique
| Date de publication | Apr 20th, 2017 | May 14th, 2020 |
|---|---|---|
| Génération | Polaris | GRID |
| Production | End-of-life | Active |
| Prix de lancement | 79 USD | — |
| Bus interface | PCIe 3.0 x8 | PCIe 4.0 x16 |
| Prédécesseur | Arctic Islands | — |
| Successeur | Vega | — |
Vitesse clock
| Base clock | 1100 MHz | 900 MHz |
|---|---|---|
| Boost clock | 1183 MHz | 1005 MHz |
| Mémoire clock | 1750 MHz 7 Gbps effective | 1215 MHz 2.4 Gbps effective |
Mémoire
| Taille de la mémoire | 2 GB | 48 GB |
|---|---|---|
| Type de mémoire | GDDR5 | HBM2E |
| Bus mémoire | 64 bit | 6144 bit |
| Bande passante | 56.00 GB/s | 1,866 GB/s |
Render Config
| Shading units | 512 | 6912 |
|---|---|---|
| TMUs | 32 | 432 |
| ROPs | 16 | 192 |
| Compute units | 8 | — |
| Cache L1 | 16 KB (per CU) | 192 KB (per SM) |
| Cache L2 | 256 KB | 48 MB |
| Comptage SM | — | 108 |
| Tensor cores | — | 432 |
Performance théorique
| Taux de pixel | 18.93 GPixel/s | 193.0 GPixel/s |
|---|---|---|
| Taux de texture | 37.86 GTexel/s | 434.2 GTexel/s |
| FP16 (half) performance | 1,211 GFLOPS (1:1) | 55.57 TFLOPS (4:1) |
| FP32 (float) performance | 1,211 GFLOPS | 13.89 TFLOPS |
| FP64 (double) performance | 75.71 GFLOPS (1:16) | 6.947 TFLOPS (1:2) |
Board Design
| Taille Slot | Dual-slot | IGP |
|---|---|---|
| Longueur | 145 mm 5.7 inches | — |
| TDP | 50 W | 400 W |
| Suggestions pour le PSU | 250 W | 800 W |
| Sorties | 1x DVI1x HDMI1x DisplayPort | No outputs |
| Connecteurs de puissance | None | None |
| Board number | D090-01 | — |
Caractéristiques graphiques
| DirectX | 12 (12_0) | — |
|---|---|---|
| OpenGL | 4.6 | — |
| OpenCL | 2.1 | 3.0 |
| Vulkan | 1.2 | — |
| Shader model | 6.4 | — |
| CUDA | — | 8.0 |
