GPU model | GA100 | NV17 |
---|---|---|
Architecture | Ampere | Celsius |
Foundry | TSMC | TSMC |
Manufacturing process | 7 nm | 150 nm |
Number of transistors | 54,200 million | 29 million |
Die size | 826 mm² | 65 mm² |
GPU variant | — | MX440 |
Release date | Jun 22nd, 2020 | Feb 6th, 2002 |
---|---|---|
Generation | Tesla | GeForce 4 MX |
Production | Active | End-of-life |
Interface | PCIe 4.0 x16 | AGP 4x |
Predecessor | — | GeForce 3 |
Successor | — | GeForce 4 Ti |
GPU clock | 1410 MHz | 275 MHz |
---|---|---|
Memory clock | 1215 MHz 2.4 Gbps effective | 200 MHz 400 Mbps effective |
Base clock | 765 MHz | — |
Boost clock | 1410 MHz | — |
Memory size | 40 GB | 64 MB |
---|---|---|
Memory type | HBM2E | DDR |
Memory bus width | 5120 bit | 128 bit |
Bandwidth | 1,555 GB/s | 6.400 GB/s |
Shading units | 6912 | — |
---|---|---|
TMUs | 432 | 4 |
ROPs | 160 | 2 |
SM count | 108 | — |
Tensor cores | 432 | — |
Cache L1 | 192 KB (per SM) | — |
Cache L2 | 40 MB | — |
Pixel shaders | — | 2 |
Vertex shaders | — | 0 |
Pixel rate | 225.6 GPixel/s | 550.0 MPixel/s |
---|---|---|
Texture rate | 609.1 GTexel/s | 1.100 GTexel/s |
FP16 (half) performance | 77.97 TFLOPS (4:1) | — |
FP32 (float) performance | 19.49 TFLOPS | — |
FP64 (double) performance | 9.746 TFLOPS (1:2) | — |
Slot width | Dual-slot | Single-slot |
---|---|---|
Length | 267 mm 10.5 inches | — |
TDP | 250 W | unknown |
Suggested PSU | 600 W | 200 W |
Outputs | No outputs | 1x DVI1x VGA |
Power connectors | 8-pin EPS | None |
Board number | P1001 SKU 200 | — |
DirectX | — | 7.0 |
---|---|---|
OpenGL | — | 1.5 |
OpenCL | 3.0 | — |
Vulkan | — | — |
CUDA | 8.0 | — |
Shader model | — | — |
Pixel shader | — | 1.1 |
Vertex shader | — | 1.1 |