GPU model | Mars | GF108 |
---|---|---|
GPU variant | Mars XT | — |
Architecture | GCN 1.0 | Fermi |
Foundry | TSMC | TSMC |
Manufacturing process | 28 nm | 40 nm |
Number of transistors | 950 million | 585 million |
Die size | 77 mm² | 116 mm² |
Release date | Mar 1st, 2013 | May 15th, 2012 |
---|---|---|
Generation | Solar System (HD 8600M) | GeForce 600A |
Production | End-of-life | End-of-life |
Interface | PCIe 3.0 x8 | PCIe 2.0 x16 |
Predecessor | London | — |
Successor | Crystal System | GeForce 700A |
Base clock | 775 MHz | — |
---|---|---|
Boost clock | 825 MHz | — |
Memory clock | 900 MHz 1800 Mbps effective | 900 MHz 1800 Mbps effective |
GPU clock | — | 660 MHz |
Shader clock | — | 1320 MHz |
Memory size | 1024 MB | 1024 MB |
---|---|---|
Memory type | DDR3 | DDR3 |
Memory bus width | 64 bit | 128 bit |
Bandwidth | 14.40 GB/s | 28.80 GB/s |
Shading units | 384 | 48 |
---|---|---|
TMUs | 24 | 8 |
ROPs | 8 | 4 |
Compute units | 6 | — |
Cache L1 | 16 KB (per CU) | 64 KB (per SM) |
Cache L2 | 128 KB | 256 KB |
SM count | — | 1 |
Pixel rate | 6.600 GPixel/s | 1.320 GPixel/s |
---|---|---|
Texture rate | 19.80 GTexel/s | 5.280 GTexel/s |
FP32 (float) performance | 633.6 GFLOPS | 126.7 GFLOPS |
FP64 (double) performance | — | 10.56 GFLOPS (1:12) |
TDP | unknown | 49 W |
---|---|---|
Outputs | No outputs | No outputs |
Slot width | — | Single-slot |
DirectX | 12 (11_1) | 12 (11_0) |
---|---|---|
OpenGL | 4.6 | 4.6 |
OpenCL | 1.2 | 1.1 |
Vulkan | 1.2 | — |
Shader model | 5.1 | 5.1 |
CUDA | — | 2.1 |