GPU Name | Haswell GT3 | TU102 |
---|---|---|
Architektur | Generation 7.5 | Turing |
Hersteller | Intel | TSMC |
Prozessgröße | 22 nm | 12 nm |
Transistoren | 1,300 million | 18,600 million |
Chipgröße | 181 mm² | 754 mm² |
GPU Variante | — | TU102-875-A1 |
Veröffentlichungsdatum | May 27th, 2013 | — |
---|---|---|
Generation | HD Graphics-M (Haswell) | — |
Produktion | End-of-life | — |
Businterface | Ring Bus | — |
Reviews | 1 in our database | — |
Basis-Taktfrequenz | 200 MHz | 1065 MHz |
---|---|---|
Boost-Takt | 1100 MHz | 1395 MHz |
Speicher-Taktung | System Shared | 1575 MHz 12.6 Gbps effective |
Speichergröße | System Shared | 16 GB |
---|---|---|
Speichertyp | System Shared | GDDR6 |
Speicherbus | System Shared | 384 bit |
Bandbreite | System Dependent | 604.8 GB/s |
Shading-Einheiten | 320 | 4608 |
---|---|---|
TMUs | 40 | 288 |
ROPs | 4 | 96 |
Execution Units | 40 | — |
Anzahl SM | — | 72 |
Tensor Cores | — | 576 |
RT-Kerne | — | 72 |
Cache L1 | — | 64 KB (per SM) |
Cache L2 | — | 6 MB |
Pixelrate | 4.400 GPixel/s | 133.9 GPixel/s |
---|---|---|
Texturrate | 44.00 GTexel/s | 401.8 GTexel/s |
FP32 (float) Leistung | 704.0 GFLOPS | 12.86 TFLOPS |
FP64 (double) Leistung | 176.0 GFLOPS (1:4) | 401.8 GFLOPS (1:32) |
FP16 (half) Leistung | — | 25.71 TFLOPS (2:1) |
Steckplatzbreite | IGP | Dual-slot |
---|---|---|
TDP | 30 W | 260 W |
Ausgabe | No outputs | No outputs |
Länge | — | 267 mm 10.5 inches |
Vorgeschlagenes Netzteil | — | 600 W |
Stromverbindung | — | 1x 6-pin + 1x 8-pin |
Board-Nummer | — | PG150 SKU 215 |
DirectX | 12 (11_1) | 12 Ultimate (12_2) |
---|---|---|
OpenGL | 4.3 | 4.6 |
OpenCL | 1.2 | 3.0 |
Vulkan | 1.0 | 1.2 |
Shader-Modell | 5.1 | 6.6 |
CUDA | — | 7.5 |
Veröffentlichungsdatum | — | Unknown |
---|---|---|
Generation | — | GRID |
Produktion | — | Active |
Businterface | — | PCIe 3.0 x16 |