ATI Radeon HD 4850 vs NVIDIA GeForce GT 320 OEM
Processeur graphique
| Nom du GPU | RV770 | GT215 |
|---|---|---|
| GPU variant | RV770 PRO (215-0669065) | GT215-250-A2 |
| Architecture | TeraScale | Tesla 2.0 |
| Fonderie | TSMC | TSMC |
| Taille du processus | 55 nm | 40 nm |
| Transistors | 956 million | 727 million |
| Taille des matrices | 256 mm² | 144 mm² |
Carte graphique
| Date de publication | Jun 25th, 2008 | Feb 2nd, 2010 |
|---|---|---|
| Génération | Radeon R700 | GeForce 300 |
| Production | End-of-life | End-of-life |
| Prix de lancement | 199 USD | — |
| Bus interface | PCIe 2.0 x16 | PCIe 2.0 x16 |
| Critiques | 127 in our database | — |
| Prédécesseur | Radeon R600 | GeForce 200 |
| Successeur | Evergreen | GeForce 400 |
Vitesse clock
| GPU clock | 625 MHz | 540 MHz |
|---|---|---|
| Mémoire clock | 993 MHz 1986 Mbps effective | 790 MHz 1580 Mbps effective |
| Shader clock | — | 1302 MHz |
Mémoire
| Taille de la mémoire | 512 MB | 1024 MB |
|---|---|---|
| Type de mémoire | GDDR3 | GDDR3 |
| Bus mémoire | 256 bit | 128 bit |
| Bande passante | 63.55 GB/s | 25.28 GB/s |
Render Config
| Shading units | 800 | 72 |
|---|---|---|
| TMUs | 40 | 24 |
| ROPs | 16 | 8 |
| Compute units | 10 | — |
| Cache L1 | 16 KB (per CU) | — |
| Cache L2 | 256 KB | 64 KB |
| Comptage SM | — | 9 |
Performance théorique
| Taux de pixel | 10.00 GPixel/s | 4.320 GPixel/s |
|---|---|---|
| Taux de texture | 25.00 GTexel/s | 12.96 GTexel/s |
| FP32 (float) performance | 1,000 GFLOPS | 187.5 GFLOPS |
| FP64 (double) performance | 200.0 GFLOPS (1:5) | — |
Board Design
| Taille Slot | Single-slot | Single-slot |
|---|---|---|
| Longueur | 246 mm 9.7 inches | 175 mm 6.9 inches |
| TDP | 110 W | 43 W |
| Suggestions pour le PSU | 300 W | 200 W |
| Sorties | 2x DVI1x S-Video | 1x DVI1x HDMI1x VGA |
| Connecteurs de puissance | 1x 6-pin | None |
| Board number | B501-31 | P684 |
Caractéristiques graphiques
| DirectX | 10.1 (10_1) | 11.1 (10_1) |
|---|---|---|
| OpenGL | 3.3 | 3.3 |
| OpenCL | 1.1 | 1.1 |
| Vulkan | — | — |
| Shader model | 4.1 | 4.1 |
| CUDA | — | 1.2 |
