GPU model | TU117 | GA102 |
---|---|---|
Architecture | Turing | Ampere |
Foundry | TSMC | Samsung |
Manufacturing process | 12 nm | 8 nm |
Number of transistors | 4,700 million | 28,300 million |
Die size | 200 mm² | 628 mm² |
GPU variant | — | GA102-200-KD-A1 |
Release date | Aug 25th, 2020 | — |
---|---|---|
Generation | GeForce MX (4xx) | — |
Production | Active | — |
Interface | PCIe 4.0 x4 | — |
Base clock | 1035 MHz | 1440 MHz |
---|---|---|
Boost clock | 1275 MHz | 1710 MHz |
Memory clock | 2000 MHz 8 Gbps effective | 1188 MHz 19 Gbps effective |
Memory size | 2 GB | 10 GB |
---|---|---|
Memory type | GDDR5 | GDDR6X |
Memory bus width | 64 bit | 320 bit |
Bandwidth | 64.00 GB/s | 760.3 GB/s |
Shading units | 896 | 8704 |
---|---|---|
TMUs | 56 | 272 |
ROPs | 32 | 96 |
SM count | 14 | 68 |
Cache L1 | 64 KB (per SM) | 128 KB (per SM) |
Cache L2 | 512 KB | 5 MB |
Tensor cores | — | 272 |
RT cores | — | 68 |
Pixel rate | 40.80 GPixel/s | 164.2 GPixel/s |
---|---|---|
Texture rate | 71.40 GTexel/s | 465.1 GTexel/s |
FP16 (half) performance | 4.570 TFLOPS (2:1) | 29.77 TFLOPS (1:1) |
FP32 (float) performance | 2.285 TFLOPS | 29.77 TFLOPS |
FP64 (double) performance | 71.40 GFLOPS (1:32) | 465.1 GFLOPS (1:64) |
TDP | 31 W | 320 W |
---|---|---|
Outputs | No outputs | 1x HDMI3x DisplayPort |
Power connectors | None | 1x 12-pin |
Slot width | — | Dual-slot |
Length | — | 285 mm 11.2 inches |
Suggested PSU | — | 700 W |
Board number | — | PG132 SKU 30 |
Width | — | 112 mm 4.4 inches |
Height | — | 40 mm 1.6 inches |
DirectX | 12 (12_1) | 12 Ultimate (12_2) |
---|---|---|
OpenGL | 4.6 | 4.6 |
OpenCL | 3.0 | 3.0 |
Vulkan | 1.2 | 1.2 |
CUDA | 7.5 | 8.6 |
Shader model | 6.6 | 6.6 |
Release date | — | Sep 1st, 2020 |
---|---|---|
Generation | — | GeForce 30 |
Production | — | Active |
Launch price | — | 699 USD |
Interface | — | PCIe 4.0 x16 |
gpu.details.availability | — | Sep 17th, 2020 |
Predecessor | — | GeForce 20 |
Current price | — | Amazon / Newegg |
Reviews | — | 99 in our database |