Quick/Comparisons
The Radeon RX 9070 XT takes the lead in gaming performance with a score of 73, narrowly edging out the Radeon RX 7900 XT's 72.
The Radeon RX 7900 XT's workstation score of 70 surpasses the Radeon RX 9070 XT's 62, making it the better choice for content creation.
The Radeon RX 7900 XT and Radeon RX 9070 XT are neck and neck in AI/ML performance, with scores of 68 and 67 respectively, making them essentially tied.
72 | Gaming | 73 |
70 | Workstation | 62 |
68 | AI/ML | 67 |
45 | Energy Efficiency | 49 |
70 | Quick Comparison Final Score | 68 |
| Navi 31 | Graphics Processor | Navi 48 |
| 5376 | Cores | 4096 |
| 336 / 192 | TMUs / ROPs | 256 / 128 |
| 3548.59 | Blender GPU | 3176.42 |
| 157 | Forza Horizon 5 | 169 |
| 228 | The Witcher 3 | 236 |
| 223 | Counter-Strike 2 | 224 |
| 173 | Far Cry 6 | 178 |
| 126 | Hogwarts Legacy | 132 |
| 222 | Call of Duty | 234 |
| 129 | Ghost of Tsushima | 135 |
| 158 | Cyberpunk 2077 | 166 |
| 247 | Tomb Raider | 268 |
| 185 | Average FPS | 194 |
| 219 | 1080p High | 229 |
| 185 | 1080p Ultra | 194 |
| 144 | 1440p Ultra | 150 |
| 85 | 4K Ultra | 89 |
| Low | Margin of Error | Low |
| 203754 | GB6 Compute Score | 182816 |
| 419.8 img/sec | Background Blur | 427.4 img/sec |
| 240.5 img/sec | Face Detection | 251 img/sec |
| 8.72 Gpixels/sec | Horizon Detection | 6.77 Gpixels/sec |
| 12 Gpixels/sec | Edge Detection | 9.42 Gpixels/sec |
| 9.88 Gpixels/sec | Gaussian Blur | 9.35 Gpixels/sec |
| 1.71 Gpixels/sec | Feature Matching | 1.42 Gpixels/sec |
| 701.7 Gpixels/sec | Stereo Matching | 651.4 Gpixels/sec |
| 26631.4 FPS | Particle Physics | 23630 FPS |
| OpenCL | API | OpenCL |
| 29015 | G3D Mark Score | 26904 |
| 1241 | G2D Mark | 1319 |
| 340 FPS | DirectX 11 | 295 FPS |
| 111 FPS | DirectX 12 | 77 FPS |
| 16851 Ops/s | GPU Compute | 15995 Ops/s |
| 39099 | GB6 ML Single Precision | 39190 |
| 47298 | GB6 ML Half Precision | 55727 |
| 30875 | GB6 ML Quantized | 33599 |
| 16951 | Image Classification (SP) | 18001 |
| 23234 | Image Segmentation (HP) | 25174 |
| 41128 | Image Super Resolution (Q) | 45714 |
| 67615 | Face Detection (HP) | 74776 |
| 237026 | Pose Estimation (Q) | 196137 |
| 4961 | Text Classification (SP) | 5163 |
| 7043 | Machine Translation (HP) | 8677 |
| 18737 | Object Detection (SP) | 19602 |
| 67746 | Depth Estimation (Q) | 74879 |
| 457919 | Style Transfer (SP) | 411302 |
| ONNX | Framework | ONNX |
| DirectML | Backend | DirectML |
| 460 GPixel/s | Pixel Fill Rate | 380 GPixel/s |
| 804 GTexel/s | Texture Fill Rate | 760 GTexel/s |
| 51.5 TFLOPS | FLOPS (FP32) | 48.7 TFLOPS |
| GDDR6 | Memory Type | GDDR6 |
| 20 GB | Memory Size | 16 GB |
| 2500 MHz | Memory Clock | 2518 MHz |
| 20000 Mbps | Effective Memory Speed | 20100 Mbps |
| 320-bit | Bus | 256-bit |
| No | ECC | No |
| 800 GB/s | Memory Bandwidth | 640 GB/s |
| PCIe 4.0 x16 | Interface | PCIe 5.0 x16 |
| 300 W | TGP | 304 W |
| TSMC | Manufacturing | TSMC |
| 5 nm | Fabrication Process | 4 nm |
| 529 mm² | Die Size | 356.5 mm² |
| 57 billion | Transistor Count | 53.9 billion |
| 107.75 MTr/mm² | Transistor Density | 151.19 MTr/mm² |
| 12 | DirectX | 12 |
| 1.3 | Vulkan | 1.3 |
| 4.6 | OpenGL | 4.6 |
| 2.2 | OpenCL | 2.2 |
| — | CUDA | No |
| Yes | Ray Tracing | Yes |
| No | DLSS | No |
| 2.1 | DisplayPort | 2.1a |
| Amd | Vendor | Amd |
| Discrete | Build | Discrete |
| December 13, 2023 | Released | January 7, 2025 |
| Desktop | Case | Desktop |
| Gaming | Purpose | Gaming |
| High-end | Segment | High-end |
| RDNA 3.0 | Architecture | RDNA 4.0 |
| Navi 31 | GPU Codename | Navi 48 |
| -Intel Core i7 13700Kor above | Recommended CPU | -Intel Core Ultra 5 245Kor above |
| 24675 | Steel Nomad Lite Score | 26180 |
| 27067 | Time Spy | 29701 |
| 104912 | Solar Bay | 121225 |
| 14358 | Port Royal | 18525 |
| 67256 | Fire Strike | 66381 |
| 48319 | Wild Life Extreme | 54685 |
| 176267 | Night Raid | 195698 |
| 1387 MHz | Base Clock | 1660 MHz |
| 2394 MHz | Boost Clock | 2970 MHz |
| 5376 | Shading Units | 4096 |
| 336 | Texture Mapping Units (TMUs) | 256 |
| 192 | Render Output Units (ROPs) | 128 |
| — | Compute Units (Pipelines) | 64 |
| — | Tensor Cores | 128 |
| 84 | Ray-tracing Cores | 64 |
| 256KB per cluster | L1 Cache | 128KB per cluster |
| 6MB shared | L2 Cache | 4MB shared |
| 4 IPC | Instructions Per Cycle | 4 IPC |
Compare with something else?
Compare Radeon RX 7900 with another →