GPU model | Ellesmere | TU104 |
---|---|---|
GPU variant | Ellesmere XT | TU104-895-A1 |
Architecture | GCN 4.0 | Turing |
Foundry | GlobalFoundries | TSMC |
Manufacturing process | 14 nm | 12 nm |
Number of transistors | 5,700 million | 13,600 million |
Die size | 232 mm² | 545 mm² |
Release date | Dec 12th, 2016 | Sep 13th, 2018 |
---|---|---|
Generation | Radeon Instinct | Tesla |
Production | Active | Active |
Interface | PCIe 3.0 x16 | PCIe 3.0 x16 |
Base clock | 1120 MHz | 585 MHz |
---|---|---|
Boost clock | 1233 MHz | 1590 MHz |
Memory clock | 1750 MHz 7 Gbps effective | 1250 MHz 10 Gbps effective |
Memory size | 8 GB | 16 GB |
---|---|---|
Memory type | GDDR5 | GDDR6 |
Memory bus width | 256 bit | 256 bit |
Bandwidth | 224.0 GB/s | 320.0 GB/s |
Shading units | 2304 | 2560 |
---|---|---|
TMUs | 144 | 160 |
ROPs | 32 | 64 |
Compute units | 36 | — |
Cache L1 | 16 KB (per CU) | 64 KB (per SM) |
Cache L2 | 2 MB | 4 MB |
SM count | — | 40 |
Tensor cores | — | 320 |
RT cores | — | 40 |
Pixel rate | 39.46 GPixel/s | 101.8 GPixel/s |
---|---|---|
Texture rate | 177.6 GTexel/s | 254.4 GTexel/s |
FP16 (half) performance | 5.682 TFLOPS (1:1) | 65.13 TFLOPS (8:1) |
FP32 (float) performance | 5.682 TFLOPS | 8.141 TFLOPS |
FP64 (double) performance | 355.1 GFLOPS (1:16) | 254.4 GFLOPS (1:32) |
Slot width | Single-slot | Single-slot |
---|---|---|
Length | 241 mm 9.5 inches | 168 mm 6.6 inches |
Width | 111 mm 4.4 inches | — |
TDP | 150 W | 70 W |
Suggested PSU | 450 W | 250 W |
Outputs | No outputs | No outputs |
Power connectors | 1x 6-pin | None |
Board number | D122 | PG183 SKU 200 |
DirectX | 12 (12_0) | 12 Ultimate (12_2) |
---|---|---|
OpenGL | 4.6 | 4.6 |
OpenCL | 2.1 | 3.0 |
Vulkan | 1.2 | 1.2 |
Shader model | 6.4 | 6.6 |
CUDA | — | 7.5 |