Key Differences
Advantages of GeForce RTX 5090
Advantages of Radeon RX 9070 XT
Specifications
Performance
| TGP | 575 W | 304 W |
Storage Details
| Bus | 512-bit | 256-bit |
Ports
| DisplayPort | 2.1b | 2.1a |
Other
| Forza Horizon 5 | 337243228184 | 245169145111 |
| The Witcher 3 | 365335232124 | 25923717392 |
| Counter-Strike 2 | 36926521198 | 32722616883 |
| Far Cry 6 | 238204194119 | 201174160101 |
| Hogwarts Legacy | 24920016694 | 15813310761 |
| CoD: Modern Warfare III | 334322270180 | 248240178119 |
| Ghost of Tsushima | 221182168115 | 16013311974 |
| Cyberpunk 2077 | 25222516376 | 18716611757 |
| Shadow of the Tomb Raider | 326309230126 | 279269204109 |
| 1080p High | 299 | 229 |
| 1080p Ultra | 254 | 194 |
| 1440p Ultra | 207 | 152 |
| 4K Ultra | 124 | 90 |
| Margin of Error | Low | Low |
| Time Spy | 47023 | 29924 |
| Solar Bay | 233922 | 118881 |
| Port Royal | 37919 | 18600 |
| Fire Strike | 87386 | 66456 |
| Wild Life Extreme | 109309 | 54280 |
| Night Raid | 207511 | 198779 |
| Background Blur | 333 img/sec | 427.4 img/sec |
| Face Detection | 298.5 img/sec | 251 img/sec |
| Horizon Detection | 21.7 Gpixels/sec | 6.77 Gpixels/sec |
| Edge Detection | 33.7 Gpixels/sec | 9.42 Gpixels/sec |
| Gaussian Blur | 34.5 Gpixels/sec | 9.35 Gpixels/sec |
| Feature Matching | 2.53 Gpixels/sec | 1.42 Gpixels/sec |
| Stereo Matching | 2850 Gpixels/sec | 651.4 Gpixels/sec |
| Particle Physics | 59146.9 FPS | 23630 FPS |
| API | OpenGL | OpenGL |
| G2D Mark | 1409 | 1327 |
| DirectX 11 | 341 FPS | 294 FPS |
| DirectX 12 | 177 FPS | 77 FPS |
| GPU Compute | 25439 Ops/s | 15964 Ops/s |
| Aztec Ruins Normal Tier | 1568.7 FPS | 1794.9 FPS |
| Car Chase | 1292 FPS | 1498 FPS |
| Manhattan 3.1.1 (1440p) | 1244.6 FPS | 1199.3 FPS |
| T-Rex | 5493.7 FPS | 5080.1 FPS |
| Tessellation | 13760.7 FPS | 6928.5 FPS |
| ALU 2 | 15791.9 FPS | 9196.7 FPS |
| Driver Overhead 2 | 332.1 FPS | 339.1 FPS |
| OS | Windows | Windows |
| Image Classification (SP) | 18948 | 18001 |
| Image Segmentation (HP) | 55109 | 25174 |
| Image Super Resolution (Q) | 58478 | 45714 |
| Face Detection (HP) | 114685 | 74776 |
| Pose Estimation (Q) | 344729 | 196137 |
| Text Classification (SP) | 4464 | 5163 |
| Machine Translation (HP) | 8184 | 8677 |
| Object Detection (SP) | 27802 | 19602 |
| Depth Estimation (Q) | 85572 | 74879 |
| Style Transfer (SP) | 687080 | 411302 |
| Framework | ONNX | ONNX |
| Backend | DirectML | DirectML |
| Vendor | Nvidia | Amd |
| Build | Discrete | Discrete |
| Released | January 7, 2025 | January 7, 2025 |
| Launch price (MSRP) | $1999 | $599 |
| Case | Desktop | Desktop |
| Purpose | Gaming | Gaming |
| Segment | High-end | High-end |
| Architecture | Blackwell 2.0 | RDNA 4.0 |
| GPU Codename | GB202-300 | Navi 48 |
| Base Clock | 2017 MHz | 1660 MHz |
| Boost Clock | 2407 MHz | 2970 MHz |
| Shading Units | 21760 | 4096 |
| Texture Mapping Units (TMUs) | 680 | 256 |
| Render Output Units (ROPs) | 176 | 128 |
| Compute Units (Pipelines) | 170 | 64 |
| Tensor Cores | 680 | 128 |
| Ray-tracing Cores | 170 | 64 |
| L1 Cache | 128KB per cluster | 128KB per cluster |
| L2 Cache | 96MB shared | 4MB shared |
| Instructions Per Cycle | 2 IPC | 4 IPC |
| Pixel Fill Rate | 424 GPixel/s | 380 GPixel/s |
| Texture Fill Rate | 1637 GTexel/s | 760 GTexel/s |
| Interface | PCIe 5.0 x16 | PCIe 5.0 x16 |
| Manufacturing | TSMC | TSMC |
| Fabrication Process | 4 nm | 4 nm |
| Die Size | 750 mm² | 356.5 mm² |
| Transistor Count | 92.2 billion | 53.9 billion |
| Transistor Density | 122.93 MTr/mm² | 151.19 MTr/mm² |
| Memory Type | GDDR7 | GDDR6 |
| Memory Size | 32 GB | 16 GB |
| Memory Clock | 1750 MHz | 2518 MHz |
| Effective Memory Speed | 28000 Mbps | 20100 Mbps |
| ECC | No | No |
| DirectX | 12.2 | 12 |
| Vulkan | 1.4 | 1.3 |
| OpenGL | 4.6 | 4.6 |
| OpenCL | 3.0 | 2.2 |
| CUDA | 12.0 | No |
| Ray Tracing | Yes | Yes |
| DLSS | DLSS 4 | No |
| Rival Equivalent | - | GeForce RTX 5070 Ti |