AMD FireStream 9250 vs NVIDIA Switch GPU 20nm
Grafikprozessor
| GPU Name | RV770 | GM20B |
|---|---|---|
| GPU Variante | RV770 PRO | ODNX02-A2 |
| Architektur | TeraScale | Maxwell 2.0 |
| Hersteller | TSMC | TSMC |
| Prozessgröße | 55 nm | 20 nm |
| Transistoren | 956 million | 2,000 million |
| Chipgröße | 256 mm² | 118 mm² |
Grafikkarte
| Veröffentlichungsdatum | Jun 16th, 2008 | Mar 17th, 2017 |
|---|---|---|
| Generation | FireStream | Console GPU |
| Produktion | End-of-life | End-of-life |
| Businterface | PCIe 2.0 x16 | IGP |
| Originalpreis | — | 299 USD |
Takt-Geschwindigkeiten
| GPU-Taktung | 625 MHz | — |
|---|---|---|
| Speicher-Taktung | 993 MHz 1986 Mbps effective | 1600 MHz 3.2 Gbps effective |
| Basis-Taktfrequenz | — | 384 MHz |
| Boost-Takt | — | 768 MHz |
Speicher
| Speichergröße | 1024 MB | 4 GB |
|---|---|---|
| Speichertyp | GDDR3 | DDR4 |
| Speicherbus | 256 bit | 64 bit |
| Bandbreite | 63.55 GB/s | 25.60 GB/s |
Render-Konfiguration
| Shading-Einheiten | 800 | 256 |
|---|---|---|
| TMUs | 40 | 16 |
| ROPs | 16 | 16 |
| Compute Units | 10 | — |
| Cache L1 | 16 KB (per CU) | — |
| Cache L2 | 256 KB | — |
| Anzahl SMM | — | 2 |
Theoretische Leistung
| Pixelrate | 10.00 GPixel/s | 12.29 GPixel/s |
|---|---|---|
| Texturrate | 25.00 GTexel/s | 12.29 GTexel/s |
| FP32 (float) Leistung | 1,000 GFLOPS | 393.2 GFLOPS |
| FP64 (double) Leistung | 200.0 GFLOPS (1:5) | 12.29 GFLOPS (1:32) |
| FP16 (half) Leistung | — | 786.4 GFLOPS (2:1) |
Boarddesign
| Steckplatzbreite | Single-slot | IGP |
|---|---|---|
| Länge | 234 mm 9.2 inches | 239 mm 9.4 inches |
| Breite | 111 mm 4.4 inches | 101 mm 4 inches |
| TDP | 150 W | 15 W |
| Vorgeschlagenes Netzteil | 450 W | 350 W |
| Ausgabe | 1x DVI | No outputs |
| Stromverbindung | 1x 6-pin | None |
| Höhe | — | 28 mm 1.1 inches |
| Gewicht | — | 0.4 kg (0.88 lbs) |
| Board-Nummer | — | HAC-001 |
Grafik-Features
| DirectX | 10.1 (10_1) | 12 (12_1) |
|---|---|---|
| OpenGL | 3.3 | 4.6 |
| OpenCL | 1.1 | 1.2 |
| Vulkan | — | 1.1 |
| Shader-Modell | 4.1 | 6.4 |
| CUDA | — | 5.3 |
