Product Compare
NVIDIA L40 CNX vs NVIDIA RTX 4000 SFF Ada Generation
| GPU | GPU | |
|---|---|---|
| Graphics Card | ||
| Release Date | Oct 2022 | Nov 2018 |
| Generation | Tesla Ada | Radeon Instinct |
| Type | Professional | Professional |
| Bus Interface | PCIe 4.0 x16 | PCIe 4.0 x16 |
| Clock Speeds | ||
| Base Clock | 1005 MHz | 1200 MHz |
| Boost Clock | 2475 MHz | 1746 MHz |
| Memory Clock | 2250 MHz | 1000 MHz |
| Memory | ||
| Memory Size | 24GB | 16GB |
| Memory Type | GDDR6 | HBM2 |
| Memory Bus | 384bit | 4096bit |
| Bandwidth | 864.0GB/s | 1.02TB/s |
| Render Config | ||
| SM Count | 142 | 48 |
| Shading Units | 18176 | 3840 |
| TMUs | 568 | 240 |
| ROPs | 192 | 64 |
| Tensor Cores | 568 | 192 |
| RT Cores | 142 | 48 |
| L1 Cache | 128 KB (per SM) | 16 KB (per CU) |
| L2 Cache | 48 MB | 4 MB |
| Compute Units | 60 | |
| Theoretical Performance | ||
| Pixel Rate | 475.2 GPixel/s | 111.7 GPixel/s |
| Texture Rate | 1406 GTexel/s | 419.0 GTexel/s |
| FP16 (half) | 89.97 TFLOPS | 26.82 TFLOPS |
| FP32 (float) | 89.97 TFLOPS | 13.41 TFLOPS |
| FP64 (double) | 1406 GFLOPS | 6.705 TFLOPS |
| Graphics Processor | ||
| GPU Name | AD102 | Vega 20 |
| GPU Variant | AD102-???-A1 | Vega 20 GLXT |
| Architecture | Ada Lovelace | GCN 5.1 |
| Foundry | TSMC | TSMC |
| Process Size | 4 nm | 7 nm |
| Transistors | 76.3 billion | 13.23 billion |
| Die Size | 608 mm² | 331 mm² |
| Board Design | ||
| TDP | 300W | 300W |
| Suggested PSU | 700 W | 700 W |
| Outputs | 1x HDMI 2.13x DisplayPort 1.4a | 1x mini-DisplayPort 1.4a |
| Power Connectors | 1x 16-pin | 2x 8-pin |
| Graphics Features | ||
| DirectX | 12 Ultimate (12_2) | 12 (12_1) |
| OpenGL | 4.6 | 4.6 |
| OpenCL | 3.0 | 2.1 |
| Vulkan | 1.3 | 1.3 |
| CUDA | 8.9 | 8.9 |
| Shader Model | 6.6 | 6.7 |