ATI Radeon HD 3870 vs NVIDIA Tesla K40m
Processeur graphique
| Nom du GPU | RV670 | GK110B |
|---|---|---|
| GPU variant | RV670 XT (215-0708005) | — |
| Architecture | TeraScale | Kepler |
| Fonderie | TSMC | TSMC |
| Taille du processus | 55 nm | 28 nm |
| Transistors | 666 million | 7,080 million |
| Taille des matrices | 192 mm² | 561 mm² |
Carte graphique
| Date de publication | Nov 19th, 2007 | Nov 22nd, 2013 |
|---|---|---|
| Génération | Radeon R600 | Tesla |
| Production | End-of-life | End-of-life |
| Prix de lancement | 269 USD | 7,699 USD |
| Bus interface | PCIe 2.0 x16 | PCIe 3.0 x16 |
| Critiques | 80 in our database | — |
| Prédécesseur | Radeon R500 PCIe | — |
| Successeur | Radeon R700 | — |
Vitesse clock
| GPU clock | 777 MHz | — |
|---|---|---|
| Mémoire clock | 1126 MHz 2.3 Gbps effective | 1502 MHz 6 Gbps effective |
| Base clock | — | 745 MHz |
| Boost clock | — | 876 MHz |
Mémoire
| Taille de la mémoire | 512 MB | 12 GB |
|---|---|---|
| Type de mémoire | GDDR4 | GDDR5 |
| Bus mémoire | 256 bit | 384 bit |
| Bande passante | 72.06 GB/s | 288.4 GB/s |
Render Config
| Shading units | 320 | 2880 |
|---|---|---|
| TMUs | 16 | 240 |
| ROPs | 16 | 48 |
| Compute units | 4 | — |
| Cache L2 | 256 KB | 1536 KB |
| Comptage SMX | — | 15 |
| Cache L1 | — | 16 KB (per SMX) |
Performance théorique
| Taux de pixel | 12.43 GPixel/s | 52.56 GPixel/s |
|---|---|---|
| Taux de texture | 12.43 GTexel/s | 210.2 GTexel/s |
| FP32 (float) performance | 497.3 GFLOPS | 5.046 TFLOPS |
| FP64 (double) performance | 99.46 GFLOPS (1:5) | 1.682 TFLOPS (1:3) |
Board Design
| Taille Slot | Dual-slot | Dual-slot |
|---|---|---|
| Longueur | 241 mm 9.5 inches | 267 mm 10.5 inches |
| TDP | 106 W | 245 W |
| Suggestions pour le PSU | 300 W | 550 W |
| Sorties | 2x DVI1x S-Video | No outputs |
| Connecteurs de puissance | 1x 6-pin | — |
| Board number | B339 | — |
Caractéristiques graphiques
| DirectX | 10.1 (10_1) | 12 (11_1) |
|---|---|---|
| OpenGL | 3.3 (full) 4.0 (partial) | 4.6 |
| OpenCL | — | 3.0 |
| Vulkan | — | 1.1 |
| Shader model | 4.1 | 5.1 |
| CUDA | — | 3.5 |
