Product Compare
NVIDIA Tesla K40st vs AMD Radeon Instinct MI250
| GPU | GPU | |
|---|---|---|
| Graphics Card | ||
| Release Date | Nov 2013 | Nov 2021 |
| Generation | Tesla | Radeon Instinct |
| Type | Professional | Professional |
| Bus Interface | PCIe 3.0 x16 | PCIe 4.0 x16 |
| Clock Speeds | ||
| Memory Clock | 1502 MHz | 1600 MHz |
| Base Clock | 1000 MHz | |
| Boost Clock | 1700 MHz | |
| Memory | ||
| Memory Size | 12GB | 128GB |
| Memory Type | GDDR5 | HBM2e |
| Memory Bus | 384bit | 8192bit |
| Bandwidth | 288.4GB/s | 3277GB/s |
| Render Config | ||
| Shading Units | 2880 | 13312 |
| TMUs | 240 | 832 |
| ROPs | 48 | 0 |
| L1 Cache | 16 KB (per SMX) | 16 KB (per CU) |
| L2 Cache | 1536 KB | 16 MB |
| Compute Units | 208 | |
| Theoretical Performance | ||
| Pixel Rate | 34.50 GPixel/s | 0 MPixel/s |
| Texture Rate | 138.0 GTexel/s | 1414 GTexel/s |
| FP32 (float) | 3.312 TFLOPS | 45.26 TFLOPS |
| FP64 (double) | 1104 GFLOPS | 45.26 TFLOPS |
| FP16 (half) | 362.1 TFLOPS | |
| Graphics Processor | ||
| GPU Name | GK110B | Aldebaran |
| Architecture | Kepler | CDNA 2.0 |
| Foundry | TSMC | TSMC |
| Process Size | 28 nm | 6 nm |
| Transistors | 7.08 billion | 58.2 billion |
| Die Size | 561 mm² | Unknown |
| GPU Variant | Aldebaran | |
| Board Design | ||
| TDP | 245W | 500W |
| Suggested PSU | 550 W | 900 W |
| Outputs | No outputs | No outputs |
| Power Connectors | 2x 8-pin | |
| Graphics Features | ||
| DirectX | 12 (11_1) | N/A |
| OpenGL | 4.6 | N/A |
| OpenCL | 3.0 | 3.0 |
| Vulkan | 1.1 | N/A |
| CUDA | 3.5 | |
| Shader Model | 5.1 | N/A |