Product Compare
AMD Radeon HD 6990M vs NVIDIA Tesla T4
| GPU | GPU | |
|---|---|---|
| Graphics Card | ||
| Release Date | Jul 2011 | Sep 2018 |
| Generation | Vancouver | Tesla |
| Type | Mobile | Professional |
| Bus Interface | MXM-B (3.0) | PCIe 3.0 x16 |
| Clock Speeds | ||
| Memory Clock | 900 MHz | 1250 MHz |
| Base Clock | 585 MHz | |
| Boost Clock | 1590 MHz | |
| Memory | ||
| Memory Size | 2GB | 16GB |
| Memory Type | GDDR5 | GDDR6 |
| Memory Bus | 256bit | 256bit |
| Bandwidth | 115.2GB/s | 320.0GB/s |
| Render Config | ||
| Compute Units | 14 | |
| Shading Units | 1120 | 2560 |
| TMUs | 56 | 160 |
| ROPs | 32 | 64 |
| L1 Cache | 8 KB (per CU) | 64 KB (per SM) |
| L2 Cache | 512 KB | 4 MB |
| SM Count | 40 | |
| Tensor Cores | 320 | |
| RT Cores | 40 | |
| Theoretical Performance | ||
| Pixel Rate | 22.88 GPixel/s | 101.8 GPixel/s |
| Texture Rate | 40.04 GTexel/s | 254.4 GTexel/s |
| FP32 (float) | 1.602 TFLOPS | 8.141 TFLOPS |
| FP16 (half) | 65.13 TFLOPS | |
| FP64 (double) | 254.4 GFLOPS | |
| Graphics Processor | ||
| GPU Name | Blackcomb | TU104 |
| GPU Variant | Blackcomb XTX(216-0811030) | TU104-895-A1 |
| Architecture | TeraScale 2 | Turing |
| Foundry | TSMC | TSMC |
| Process Size | 40 nm | 12 nm |
| Transistors | 1.7 billion | 13.6 billion |
| Die Size | 212 mm² | 545 mm² |
| Board Design | ||
| TDP | 100W | 70W |
| Outputs | No outputs | No outputs |
| Power Connectors | None | None |
| Suggested PSU | 250 W | |
| Graphics Features | ||
| DirectX | 11.2 (11_0) | 12 Ultimate (12_2) |
| OpenGL | 4.4 | 4.6 |
| OpenCL | 1.2 | 3.0 |
| Vulkan | N/A | 1.3 |
| Shader Model | 5.0 | 6.6 |
| CUDA | 7.5 | |