NVIDIA GRID K200 vs NVIDIA Quadro P620
Processeur graphique
| Nom du GPU | GK104 | GP107 |
|---|---|---|
| Architecture | Kepler | Pascal |
| Fonderie | TSMC | Samsung |
| Taille du processus | 28 nm | 14 nm |
| Transistors | 3,540 million | 3,300 million |
| Taille des matrices | 294 mm² | 132 mm² |
Carte graphique
| Date de publication | Jun 28th, 2013 | Feb 1st, 2018 |
|---|---|---|
| Génération | GRID | Quadro |
| Production | End-of-life | End-of-life |
| Prix de lancement | 235 USD | — |
| Bus interface | PCIe 3.0 x16 | PCIe 3.0 x16 |
Vitesse clock
| GPU clock | 745 MHz | — |
|---|---|---|
| Mémoire clock | 1250 MHz 5 Gbps effective | 1252 MHz 5 Gbps effective |
| Base clock | — | 1266 MHz |
| Boost clock | — | 1354 MHz |
Mémoire
| Taille de la mémoire | 256 MB | 2 GB |
|---|---|---|
| Type de mémoire | GDDR5 | GDDR5 |
| Bus mémoire | 256 bit | 128 bit |
| Bande passante | 160.0 GB/s | 80.13 GB/s |
Render Config
| Shading units | 1536 | 512 |
|---|---|---|
| TMUs | 128 | 32 |
| ROPs | 32 | 16 |
| Comptage SMX | 8 | — |
| Cache L1 | 16 KB (per SMX) | 48 KB (per SM) |
| Cache L2 | 512 KB | 1024 KB |
| Comptage SM | — | 4 |
Performance théorique
| Taux de pixel | 23.84 GPixel/s | 21.66 GPixel/s |
|---|---|---|
| Taux de texture | 95.36 GTexel/s | 43.33 GTexel/s |
| FP32 (float) performance | 2.289 TFLOPS | 1,386 GFLOPS |
| FP64 (double) performance | 95.36 GFLOPS (1:24) | 43.33 GFLOPS (1:32) |
| FP16 (half) performance | — | 21.66 GFLOPS (1:64) |
Board Design
| Taille Slot | IGP | Single-slot |
|---|---|---|
| TDP | 225 W | 40 W |
| Suggestions pour le PSU | 550 W | 200 W |
| Sorties | No outputs | 4x mini-DisplayPort |
| Longueur | — | 145 mm 5.7 inches |
| Taille | — | 69 mm 2.7 inches |
| Connecteurs de puissance | — | None |
Caractéristiques graphiques
| DirectX | 12 (11_0) | 12 (12_1) |
|---|---|---|
| OpenGL | 4.6 | 4.6 |
| OpenCL | 3.0 | 3.0 |
| Vulkan | 1.1 | 1.2 |
| CUDA | 3.0 | 6.1 |
| Shader model | 5.1 | 6.4 |
