Product Compare
NVIDIA B200 vs NVIDIA L40S
| GPU | GPU | |
|---|---|---|
| Graphics Card | ||
| Release Date | Jan 2024 | Oct 2022 |
| Generation | Server Blackwell | Server Ada |
| Type | Desktop | Professional |
| Bus Interface | PCIe 5.0 x16 | PCIe 4.0 x16 |
| Clock Speeds | ||
| Base Clock | 700 MHz | 1065 MHz |
| Boost Clock | 1965 MHz | 2520 MHz |
| Memory Clock | 2000 MHz | 2250 MHz |
| Memory | ||
| Memory Size | 90GB | 48GB |
| Memory Type | HBM3e | GDDR6 |
| Memory Bus | 4096bit | 384bit |
| Bandwidth | 4.10TB/s | 864.0GB/s |
| Render Config | ||
| SM Count | 148 | 142 |
| Shading Units | 18944 | 18176 |
| TMUs | 592 | 568 |
| ROPs | 24 | 192 |
| Tensor Cores | 592 | 568 |
| L1 Cache | 256 KB (per SM) | 128 KB (per SM) |
| L2 Cache | 50 MB | 48 MB |
| RT Cores | 142 | |
| Theoretical Performance | ||
| Pixel Rate | 47.16 GPixel/s | 483.8 GPixel/s |
| Texture Rate | 1163.3 GTexel/s | 1431.4 GTexel/s |
| FP16 (half) | 1191.2 TFLOPS | 91.61 TFLOPS |
| FP32 (float) | 74.45 TFLOPS | 91.61 TFLOPS |
| FP64 (double) | 37.22 TFLOPS | 1431.4 GFLOPS |
| Graphics Processor | ||
| GPU Name | GB100 | AD102 |
| Architecture | Blackwell | Ada Lovelace |
| Foundry | TSMC | TSMC |
| Process Size | 5 nm | 5 nm |
| Transistors | 104 billion | 76.3 billion |
| Die Size | Unknown | 609 mm² |
| GPU Variant | AD102-896-A1 | |
| Board Design | ||
| TDP | 1000W | 350W |
| Suggested PSU | 1400 W | 750 W |
| Outputs | No outputs | 4x DisplayPort 1.4a |
| Power Connectors | 1x 16-pin | |
| Graphics Features | ||
| DirectX | N/A | 12 Ultimate (12_2) |
| OpenGL | N/A | 4.6 |
| OpenCL | 3.0 | 3.0 |
| Vulkan | N/A | 1.4 |
| CUDA | 10.0 | 8.9 |
| Shader Model | N/A | 6.9 |