Product Compare
NVIDIA Tesla K40st vs NVIDIA H100 SXM5 96 GB
| GPU | GPU | |
|---|---|---|
| Graphics Card | ||
| Release Date | Nov 2013 | Mar 2023 |
| Generation | Tesla | Server Hopper |
| Type | Professional | AI GPU |
| Bus Interface | PCIe 3.0 x16 | PCIe 5.0 x16 |
| Clock Speeds | ||
| Memory Clock | 1502 MHz | 1313 MHz |
| Base Clock | 1350 MHz | |
| Boost Clock | 1980 MHz | |
| Memory | ||
| Memory Size | 12GB | 96GB |
| Memory Type | GDDR5 | HBM3 |
| Memory Bus | 384bit | 5120bit |
| Bandwidth | 288.4GB/s | 3.36TB/s |
| Render Config | ||
| Shading Units | 2880 | 16896 |
| TMUs | 240 | 528 |
| ROPs | 48 | 24 |
| L1 Cache | 16 KB (per SMX) | 256 KB (per SM) |
| L2 Cache | 1536 KB | 50 MB |
| SM Count | 132 | |
| Tensor Cores | 528 | |
| Theoretical Performance | ||
| Pixel Rate | 34.50 GPixel/s | 47.52 GPixel/s |
| Texture Rate | 138.0 GTexel/s | 1045 GTexel/s |
| FP32 (float) | 3.312 TFLOPS | 66.91 TFLOPS |
| FP64 (double) | 1104 GFLOPS | 33.45 TFLOPS |
| FP16 (half) | 267.6 TFLOPS | |
| Graphics Processor | ||
| GPU Name | GK110B | GH100 |
| Architecture | Kepler | Hopper |
| Foundry | TSMC | TSMC |
| Process Size | 28 nm | 5 nm |
| Transistors | 7.08 billion | 80 billion |
| Die Size | 561 mm² | 814 mm² |
| Board Design | ||
| TDP | 245W | 700W |
| Suggested PSU | 550 W | 1100 W |
| Outputs | No outputs | No outputs |
| Power Connectors | 8-pin EPS | |
| 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 | 9.0 |
| Shader Model | 5.1 | N/A |