Product Compare
NVIDIA L40S vs NVIDIA H200 SXM 141 GB
| GPU | GPU | |
|---|---|---|
| Graphics Card | ||
| Release Date | Oct 2022 | Nov 2024 |
| Generation | Server Ada | Server Hopper |
| Type | Professional | AI GPU |
| Bus Interface | PCIe 4.0 x16 | PCIe 5.0 x16 |
| Clock Speeds | ||
| Base Clock | 1065 MHz | 1500 MHz |
| Boost Clock | 2520 MHz | 1980 MHz |
| Memory Clock | 2250 MHz | 1593 MHz |
| Memory | ||
| Memory Size | 48GB | 141GB |
| Memory Type | GDDR6 | HBM3e |
| Memory Bus | 384bit | 6144bit |
| Bandwidth | 864.0GB/s | 4.89TB/s |
| Render Config | ||
| SM Count | 142 | 132 |
| Shading Units | 18176 | 16896 |
| TMUs | 568 | 528 |
| ROPs | 192 | 24 |
| Tensor Cores | 568 | 528 |
| RT Cores | 142 | |
| L1 Cache | 128 KB (per SM) | 256 KB (per SM) |
| L2 Cache | 48 MB | 50 MB |
| Theoretical Performance | ||
| Pixel Rate | 483.8 GPixel/s | 47.52 GPixel/s |
| Texture Rate | 1431.4 GTexel/s | 1045.4 GTexel/s |
| FP16 (half) | 91.61 TFLOPS | 267.6 TFLOPS |
| FP32 (float) | 91.61 TFLOPS | 66.91 TFLOPS |
| FP64 (double) | 1431.4 GFLOPS | 33.45 TFLOPS |
| Graphics Processor | ||
| GPU Name | AD102 | GH100 |
| GPU Variant | AD102-896-A1 | |
| Architecture | Ada Lovelace | Hopper |
| Foundry | TSMC | TSMC |
| Process Size | 5 nm | 5 nm |
| Transistors | 76.3 billion | 80 billion |
| Die Size | 609 mm² | 814 mm² |
| Board Design | ||
| TDP | 350W | 700W |
| Suggested PSU | 750 W | 1100 W |
| Outputs | 4x DisplayPort 1.4a | No outputs |
| Power Connectors | 1x 16-pin | 8-pin EPS |
| Graphics Features | ||
| DirectX | 12 Ultimate (12_2) | N/A |
| OpenGL | 4.6 | N/A |
| OpenCL | 3.0 | 3.0 |
| Vulkan | 1.4 | N/A |
| CUDA | 8.9 | 9.0 |
| Shader Model | 6.9 | N/A |