NVIDIA A100 SXM4 40 GB vs NVIDIA A10G
Grafikprozessor
| GPU Name | GA100 | GA102 |
|---|---|---|
| Architektur | Ampere | Ampere |
| Hersteller | TSMC | Samsung |
| Prozessgröße | 7 nm | 8 nm |
| Transistoren | 54,200 million | 28,300 million |
| Chipgröße | 826 mm² | 628 mm² |
| GPU Variante | — | GA102-890-A1 |
Grafikkarte
| Veröffentlichungsdatum | May 14th, 2020 | Apr 12th, 2021 |
|---|---|---|
| Generation | Tesla | Tesla |
| Produktion | Active | Active |
| Businterface | PCIe 4.0 x16 | PCIe 4.0 x16 |
| gpu.details.availability | — | 2021 |
Takt-Geschwindigkeiten
| GPU-Taktung | 1410 MHz | — |
|---|---|---|
| Speicher-Taktung | 1215 MHz 2.4 Gbps effective | 1563 MHz 12.5 Gbps effective |
| Basis-Taktfrequenz | 1095 MHz | 1320 MHz |
| Boost-Takt | 1410 MHz | 1710 MHz |
Speicher
| Speichergröße | 40 GB | 12 GB |
|---|---|---|
| Speichertyp | HBM2E | GDDR6 |
| Speicherbus | 5120 bit | 384 bit |
| Bandbreite | 1,555 GB/s | 600.2 GB/s |
Render-Konfiguration
| Shading-Einheiten | 6912 | 9216 |
|---|---|---|
| TMUs | 432 | 288 |
| ROPs | 160 | 96 |
| Anzahl SM | 108 | 72 |
| Tensor Cores | 432 | 288 |
| Cache L1 | 192 KB (per SM) | 128 KB (per SM) |
| Cache L2 | 40 MB | 6 MB |
| RT-Kerne | — | 72 |
Theoretische Leistung
| Pixelrate | 225.6 GPixel/s | 164.2 GPixel/s |
|---|---|---|
| Texturrate | 609.1 GTexel/s | 492.5 GTexel/s |
| FP16 (half) Leistung | 77.97 TFLOPS (4:1) | 31.52 TFLOPS (1:1) |
| FP32 (float) Leistung | 19.49 TFLOPS | 31.52 TFLOPS |
| FP64 (double) Leistung | 9.746 TFLOPS (1:2) | 985.0 GFLOPS (1:32) |
Boarddesign
| Steckplatzbreite | IGP | Single-slot |
|---|---|---|
| TDP | 400 W | 150 W |
| Vorgeschlagenes Netzteil | 800 W | 450 W |
| Ausgabe | No outputs | No outputs |
| Stromverbindung | None | 8-pin EPS |
| Länge | — | 267 mm 10.5 inches |
| Breite | — | 112 mm 4.4 inches |
| Board-Nummer | — | PG133 SKU 215 |
Grafik-Features
| DirectX | — | 12 Ultimate (12_2) |
|---|---|---|
| OpenGL | — | 4.6 |
| OpenCL | 3.0 | 3.0 |
| Vulkan | — | 1.2 |
| CUDA | 8.0 | 8.6 |
| Shader-Modell | — | 6.6 |
