Product Compare
NVIDIA L40S vs AMD Radeon RX 570
| GPU | GPU | |
|---|---|---|
| Graphics Card | ||
| Release Date | Oct 2022 | Apr 2017 |
| Generation | Server Ada | Polaris |
| Type | Professional | Desktop |
| Bus Interface | PCIe 4.0 x16 | PCIe 3.0 x16 |
| Clock Speeds | ||
| Base Clock | 1065 MHz | 1168 MHz |
| Boost Clock | 2520 MHz | 1244 MHz |
| Memory Clock | 2250 MHz | 1750 MHz |
| Memory | ||
| Memory Size | 48GB | 4GB |
| Memory Type | GDDR6 | GDDR5 |
| Memory Bus | 384bit | 256bit |
| Bandwidth | 864.0GB/s | 224.0GB/s |
| Render Config | ||
| SM Count | 142 | |
| Shading Units | 18176 | 2048 |
| TMUs | 568 | 128 |
| ROPs | 192 | 32 |
| Tensor Cores | 568 | |
| RT Cores | 142 | |
| L1 Cache | 128 KB (per SM) | 16 KB (per CU) |
| L2 Cache | 48 MB | 2 MB |
| Compute Units | 32 | |
| Theoretical Performance | ||
| Pixel Rate | 483.8 GPixel/s | 39.81 GPixel/s |
| Texture Rate | 1431.4 GTexel/s | 159.2 GTexel/s |
| FP16 (half) | 91.61 TFLOPS | 5.095 TFLOPS |
| FP32 (float) | 91.61 TFLOPS | 5.095 TFLOPS |
| FP64 (double) | 1431.4 GFLOPS | 318.5 GFLOPS |
| Graphics Processor | ||
| GPU Name | AD102 | Polaris 20 |
| GPU Variant | AD102-896-A1 | Polaris 20 XL(215-0910052) |
| Architecture | Ada Lovelace | GCN 4.0 |
| Foundry | TSMC | GlobalFoundries |
| Process Size | 5 nm | 14 nm |
| Transistors | 76.3 billion | 5.7 billion |
| Die Size | 609 mm² | 232 mm² |
| Board Design | ||
| TDP | 350W | 150W |
| Suggested PSU | 750 W | 450 W |
| Outputs | 4x DisplayPort 1.4a | 1x DVI1x HDMI 2.0b3x DisplayPort 1.4a |
| Power Connectors | 1x 16-pin | 1x 6-pin |
| Graphics Features | ||
| DirectX | 12 Ultimate (12_2) | 12 (12_0) |
| OpenGL | 4.6 | 4.6 |
| OpenCL | 3.0 | 2.1 |
| Vulkan | 1.4 | 1.2 |
| CUDA | 8.9 | |
| Shader Model | 6.9 | 6.4 |