AMD Radeon HD 6970 vs NVIDIA Tesla D870
Graphics Processor
| GPU model | Cayman | G80 |
|---|---|---|
| GPU variant | Cayman XT (215-0807007) | — |
| Architecture | TeraScale 3 | Tesla |
| Foundry | TSMC | TSMC |
| Manufacturing process | 40 nm | 90 nm |
| Number of transistors | 2,640 million | 681 million |
| Die size | 389 mm² | 484 mm² |
Graphics Card
| Release date | Dec 14th, 2010 | May 2nd, 2007 |
|---|---|---|
| Generation | Northern Islands | Tesla |
| Production | End-of-life | End-of-life |
| Launch price | 369 USD | 7,499 USD |
| Interface | PCIe 2.0 x16 | PCIe 1.0 x16 |
| Reviews | 80 in our database | — |
| Predecessor | Evergreen | — |
| Successor | Southern Islands | — |
Clocks
| GPU clock | 880 MHz | 600 MHz |
|---|---|---|
| Memory clock | 1375 MHz 5.5 Gbps effective | 800 MHz 1600 Mbps effective |
| Shader clock | — | 1350 MHz |
Memory Configuration
| Memory size | 2 GB | 1536 MB |
|---|---|---|
| Memory type | GDDR5 | GDDR3 |
| Memory bus width | 256 bit | 384 bit |
| Bandwidth | 176.0 GB/s | 76.80 GB/s |
Render Configuration
| Shading units | 1536 | 128 |
|---|---|---|
| TMUs | 96 | 32 |
| ROPs | 32 | 24 |
| Compute units | 24 | — |
| Cache L1 | 8 KB (per CU) | — |
| Cache L2 | 512 KB | 96 KB |
| SM count | — | 16 |
Performance
| Pixel rate | 28.16 GPixel/s | 14.40 GPixel/s |
|---|---|---|
| Texture rate | 84.48 GTexel/s | 38.40 GTexel/s |
| FP32 (float) performance | 2.703 TFLOPS | 345.6 GFLOPS |
| FP64 (double) performance | 675.8 GFLOPS (1:4) | — |
Dimensions & Outputs
| Slot width | Dual-slot | Dual-slot |
|---|---|---|
| Length | 286 mm 11.3 inches | — |
| Width | 126 mm 5 inches | — |
| Height | 42 mm 1.7 inches | — |
| TDP | 250 W | 520 W |
| Suggested PSU | 600 W | 900 W |
| Outputs | 2x DVI1x HDMI2x mini-DisplayPort | No outputs |
| Power connectors | 1x 6-pin + 1x 8-pin | — |
| Board number | C200-47 | — |
API Support & Features
| DirectX | 11.2 (11_0) | 11.1 (10_0) |
|---|---|---|
| OpenGL | 4.4 | 3.3 |
| OpenCL | 1.2 | 1.1 (1.0) |
| Vulkan | — | — |
| Shader model | 5.0 | 4.0 |
| CUDA | — | 1.0 |
