NVIDIA Jetson AGX Xavier GPU vs NVIDIA Jetson Nano GPU
Processeur graphique
| Nom du GPU | GV10B | GM20B |
|---|---|---|
| Architecture | Volta | Maxwell 2.0 |
| Fonderie | TSMC | TSMC |
| Taille du processus | 12 nm | 20 nm |
| Transistors | 9,000 million | 2,000 million |
| Taille des matrices | 350 mm² | 118 mm² |
| GPU variant | — | TM660M-A2 |
Carte graphique intégré
| Date de publication | Oct 2018 | Mar 2019 |
|---|---|---|
| Génération | Tegra | Tegra |
| Production | End-of-life | End-of-life |
| Prix de lancement | 899 USD | — |
| Bus interface | IGP | IGP |
Vitesse clock
| Base clock | 854 MHz | 640 MHz |
|---|---|---|
| Boost clock | 1377 MHz | 921 MHz |
| Mémoire clock | System Shared | System Shared |
Mémoire
| Taille de la mémoire | System Shared | System Shared |
|---|---|---|
| Type de mémoire | System Shared | System Shared |
| Bus mémoire | System Shared | System Shared |
| Bande passante | System Dependent | System Dependent |
Render Config
| Shading units | 512 | 128 |
|---|---|---|
| TMUs | 32 | 16 |
| ROPs | 16 | 16 |
| Comptage SM | 8 | — |
| Tensor cores | 64 | — |
| Cache L1 | 128 KB (per SM) | 48 KB (per SMM) |
| Cache L2 | 512 KB | 256 KB |
| Comptage SMM | — | 1 |
Performance théorique
| Taux de pixel | 22.03 GPixel/s | 14.74 GPixel/s |
|---|---|---|
| Taux de texture | 44.06 GTexel/s | 14.74 GTexel/s |
| FP16 (half) performance | 2.820 TFLOPS (2:1) | 471.6 GFLOPS (2:1) |
| FP32 (float) performance | 1,410 GFLOPS | 235.8 GFLOPS |
| FP64 (double) performance | 705.0 GFLOPS (1:2) | 7.368 GFLOPS (1:32) |
Board Design
| Taille Slot | IGP | IGP |
|---|---|---|
| Longueur | 100 mm 3.9 inches | 70 mm 2.8 inches |
| Taille | 87 mm 3.4 inches | 45 mm 1.8 inches |
| TDP | 30 W | 10 W |
| Sorties | No outputs | No outputs |
Caractéristiques graphiques
| DirectX | 12 (12_1) | 12 (12_1) |
|---|---|---|
| OpenGL | 4.6 | 4.6 |
| OpenCL | 1.2 | 1.2 |
| Vulkan | 1.2 | 1.1 |
| CUDA | 7.2 | 5.3 |
| Shader model | 6.4 | 6.4 |
