GPU model | G92 | TU102 |
---|---|---|
GPU variant | G92-420-A2 | TU102-300A-K1-A1 |
Architecture | Tesla | Turing |
Foundry | TSMC | TSMC |
Manufacturing process | 65 nm | 12 nm |
Number of transistors | 754 million | 18,600 million |
Die size | 324 mm² | 754 mm² |
Release date | Mar 28th, 2008 | Sep 20th, 2018 |
---|---|---|
Generation | GeForce 9 | GeForce 20 |
Production | End-of-life | Active |
Launch price | 299 USD | 999 USD |
Interface | PCIe 2.0 x16 | PCIe 3.0 x16 |
Reviews | 84 in our database | 88 in our database |
Predecessor | GeForce 8 | GeForce 10 |
Successor | GeForce 200 | GeForce 30 |
Current price | — | Amazon / Newegg |
GPU clock | 675 MHz | — |
---|---|---|
Shader clock | 1688 MHz | — |
Memory clock | 1100 MHz 2.2 Gbps effective | 1750 MHz 14 Gbps effective |
Base clock | — | 1350 MHz |
Boost clock | — | 1545 MHz |
Memory size | 512 MB | 11 GB |
---|---|---|
Memory type | GDDR3 | GDDR6 |
Memory bus width | 256 bit | 352 bit |
Bandwidth | 70.40 GB/s | 616.0 GB/s |
Shading units | 128 | 4352 |
---|---|---|
TMUs | 64 | 272 |
ROPs | 16 | 88 |
SM count | 16 | 68 |
Cache L2 | 64 KB | 5.5 MB |
Tensor cores | — | 544 |
RT cores | — | 68 |
Cache L1 | — | 64 KB (per SM) |
Pixel rate | 10.80 GPixel/s | 136.0 GPixel/s |
---|---|---|
Texture rate | 43.20 GTexel/s | 420.2 GTexel/s |
FP32 (float) performance | 432.1 GFLOPS | 13.45 TFLOPS |
FP16 (half) performance | — | 26.90 TFLOPS (2:1) |
FP64 (double) performance | — | 420.2 GFLOPS (1:32) |
Slot width | Dual-slot | Dual-slot |
---|---|---|
Length | 267 mm 10.5 inches | 267 mm 10.5 inches |
Width | 111 mm 4.4 inches | 116 mm 4.6 inches |
Height | 38 mm 1.5 inches | 35 mm 1.4 inches |
TDP | 140 W | 250 W |
Suggested PSU | 300 W | 600 W |
Outputs | 2x DVI1x S-Video | 1x HDMI3x DisplayPort1x USB Type-C |
Power connectors | 2x 6-pin | 2x 8-pin |
Board number | P392 | PG150 SKU 32 |
DirectX | 11.1 (10_0) | 12 Ultimate (12_2) |
---|---|---|
OpenGL | 3.3 | 4.6 |
OpenCL | 1.1 | 3.0 |
Vulkan | — | 1.2 |
CUDA | 1.1 | 7.5 |
Shader model | 4.0 | 6.6 |