AMD Radeon RX 560X vs NVIDIA Tesla S870
Graphics Processor
| GPU model | Polaris 21 | G80 |
|---|---|---|
| GPU variant | Polaris 21 XT (215-0908004) | — |
| Architecture | GCN 4.0 | Tesla |
| Foundry | GlobalFoundries | TSMC |
| Manufacturing process | 14 nm | 90 nm |
| Number of transistors | 3,000 million | 681 million |
| Die size | 123 mm² | 484 mm² |
Graphics Card
| Release date | Apr 11th, 2018 | May 2nd, 2007 |
|---|---|---|
| Generation | Polaris | Tesla |
| Production | Active | End-of-life |
| Interface | PCIe 3.0 x8 | PCIe 1.0 x16 |
| Predecessor | Polaris | — |
| Successor | Vega | — |
| Launch price | — | 11,999 USD |
Clocks
| Base clock | 1175 MHz | — |
|---|---|---|
| Boost clock | 1275 MHz | — |
| Memory clock | 1750 MHz 7 Gbps effective | 800 MHz 1600 Mbps effective |
| GPU clock | — | 600 MHz |
| Shader clock | — | 1350 MHz |
Memory Configuration
| Memory size | 4 GB | 1536 MB |
|---|---|---|
| Memory type | GDDR5 | GDDR3 |
| Memory bus width | 128 bit | 384 bit |
| Bandwidth | 112.0 GB/s | 76.80 GB/s |
Render Configuration
| Shading units | 1024 | 128 |
|---|---|---|
| TMUs | 64 | 32 |
| ROPs | 16 | 24 |
| Compute units | 16 | — |
| Cache L1 | 16 KB (per CU) | — |
| Cache L2 | 1024 KB | 96 KB |
| SM count | — | 16 |
Performance
| Pixel rate | 20.40 GPixel/s | 14.40 GPixel/s |
|---|---|---|
| Texture rate | 81.60 GTexel/s | 38.40 GTexel/s |
| FP16 (half) performance | 2.611 TFLOPS (1:1) | — |
| FP32 (float) performance | 2.611 TFLOPS | 345.6 GFLOPS |
| FP64 (double) performance | 163.2 GFLOPS (1:16) | — |
Dimensions & Outputs
| Slot width | Dual-slot | Dual-slot |
|---|---|---|
| Length | 170 mm 6.7 inches | — |
| TDP | 75 W | 800 W |
| Suggested PSU | 250 W | 1200 W |
| Outputs | 1x DVI1x HDMI1x DisplayPort | No outputs |
| Power connectors | None | — |
| Board number | C994 | — |
API Support & Features
| DirectX | 12 (12_0) | 11.1 (10_0) |
|---|---|---|
| OpenGL | 4.6 | 3.3 |
| OpenCL | 2.1 | 1.1 (1.0) |
| Vulkan | 1.2 | — |
| Shader model | 6.4 | 4.0 |
| CUDA | — | 1.0 |
