GPU model | GA100 | Wayne |
---|---|---|
Architecture | Ampere | VLIW Vec4 |
Foundry | TSMC | TSMC |
Manufacturing process | 7 nm | 28 nm |
Number of transistors | 54,200 million | 11 million |
Die size | 826 mm² | 85 mm² |
GPU variant | — | Tegra 4i |
Release date | Jun 22nd, 2020 | — |
---|---|---|
Generation | Tesla | — |
Production | Active | — |
Interface | PCIe 4.0 x16 | — |
GPU clock | 1410 MHz | 660 MHz |
---|---|---|
Memory clock | 1215 MHz 2.4 Gbps effective | System Shared |
Base clock | 765 MHz | — |
Boost clock | 1410 MHz | — |
Memory size | 40 GB | System Shared |
---|---|---|
Memory type | HBM2E | System Shared |
Memory bus width | 5120 bit | System Shared |
Bandwidth | 1,555 GB/s | System Dependent |
Shading units | 6912 | — |
---|---|---|
TMUs | 432 | 2 |
ROPs | 160 | 2 |
SM count | 108 | — |
Tensor cores | 432 | — |
Cache L1 | 192 KB (per SM) | — |
Cache L2 | 40 MB | — |
Pixel shaders | — | 48 |
Vertex shaders | — | 12 |
Pixel rate | 225.6 GPixel/s | 1.320 GPixel/s |
---|---|---|
Texture rate | 609.1 GTexel/s | 1.320 GTexel/s |
FP16 (half) performance | 77.97 TFLOPS (4:1) | — |
FP32 (float) performance | 19.49 TFLOPS | — |
FP64 (double) performance | 9.746 TFLOPS (1:2) | — |
Vertex rate | — | 1.980 GVertices/s |
Slot width | Dual-slot | IGP |
---|---|---|
Length | 267 mm 10.5 inches | — |
TDP | 250 W | 20 W |
Suggested PSU | 600 W | — |
Outputs | No outputs | No outputs |
Power connectors | 8-pin EPS | — |
Board number | P1001 SKU 200 | — |
DirectX | — | — |
---|---|---|
OpenGL | — | ES 3.0 |
OpenCL | 3.0 | — |
Vulkan | — | — |
CUDA | 8.0 | — |
Shader model | — | — |
Pixel shader | — | — |
Vertex shader | — | — |
Release date | — | Feb 19th, 2013 |
---|---|---|
Generation | — | Tegra |
Production | — | End-of-life |
Interface | — | IGP |