Product Compare
NVIDIA GeForce RTX 4090 vs AMD Radeon Instinct MI325X
| GPU | GPU | |
|---|---|---|
| Graphics Card | ||
| Release Date | Sep 2022 | Oct 2024 |
| Generation | GeForce 40 | Radeon Instinct |
| Type | Desktop | Professional |
| Bus Interface | PCIe 4.0 x16 | PCIe 5.0 x16 |
| Clock Speeds | ||
| Base Clock | 2235 MHz | 1000 MHz |
| Boost Clock | 2520 MHz | 2100 MHz |
| Memory Clock | 1313 MHz | 1500 MHz |
| Memory | ||
| Memory Size | 24GB | 256GB |
| Memory Type | GDDR6X | HBM3e |
| Memory Bus | 384bit | 8192bit |
| Bandwidth | 1.01TB/s | 6.14TB/s |
| Render Config | ||
| SM Count | 128 | |
| Shading Units | 16384 | 19456 |
| TMUs | 512 | 1216 |
| ROPs | 176 | 0 |
| Tensor Cores | 512 | 1216 |
| RT Cores | 128 | |
| L1 Cache | 128 KB (per SM) | 16 KB (per CU) |
| L2 Cache | 72 MB | 16 MB |
| Compute Units | 304 | |
| L3 Cache | 256 MB | |
| Theoretical Performance | ||
| Pixel Rate | 443.5 GPixel/s | 0 MPixel/s |
| Texture Rate | 1290.2 GTexel/s | 2553.6 GTexel/s |
| FP16 (half) | 82.58 TFLOPS | 81.72 TFLOPS |
| FP32 (float) | 82.58 TFLOPS | 81.72 TFLOPS |
| FP64 (double) | 1290.2 GFLOPS | 40.86 TFLOPS |
| Graphics Processor | ||
| GPU Name | AD102 | Aqua Vanjaram |
| GPU Variant | AD102-300-A1 | |
| Architecture | Ada Lovelace | CDNA 3.0 |
| Foundry | TSMC | TSMC |
| Process Size | 5 nm | 5 nm |
| Transistors | 76.3 billion | 153 billion |
| Die Size | 609 mm² | 1017 mm² |
| Board Design | ||
| TDP | 450W | 1000W |
| Suggested PSU | 850 W | 1400 W |
| Outputs | 1x HDMI 2.13x DisplayPort 1.4a | No outputs |
| Power Connectors | 1x 16-pin | None |
| 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 | |
| Shader Model | 6.8 | N/A |