Quick/Comparisons
The GeForce RTX 4090's superior gaming performance is due to its high clock speeds and 24GB GDDR6X memory, making it the clear winner in this category.
The GeForce RTX 4090's strong workstation performance and 24GB GDDR6X memory make it the top choice for content creation tasks.
The GeForce RTX 4090's capable AI/ML compute capabilities, thanks to its Ada Lovelace architecture, give it a slight edge over the Radeon RX 9070 XT in this category.
92 | Gaming | 73 |
91 | Workstation | 62 |
76 | AI/ML | 67 |
43 | Energy Efficiency | 49 |
89 | Quick Comparison Final Score | 68 |
| AD102 | Graphics Processor | Navi 48 |
| 16384 | Cores | 4096 |
| 512 / 176 | TMUs / ROPs | 256 / 128 |
| 11688.48 | Blender GPU | 3176.42 |
| 203 | Forza Horizon 5 | 169 |
| 270 | The Witcher 3 | 236 |
| 245 | Counter-Strike 2 | 224 |
| 183 | Far Cry 6 | 178 |
| 165 | Hogwarts Legacy | 132 |
| 269 | Call of Duty | 234 |
| 150 | Ghost of Tsushima | 135 |
| 204 | Cyberpunk 2077 | 166 |
| 283 | Tomb Raider | 268 |
| 219 | Average FPS | 194 |
| 256 | 1080p High | 229 |
| 219 | 1080p Ultra | 194 |
| 176 | 1440p Ultra | 150 |
| 107 | 4K Ultra | 89 |
| Low | Margin of Error | Low |
| 2.38 Gpixels/sec | Feature Matching | 1.42 Gpixels/sec |
| 1950 Gpixels/sec | Stereo Matching | 651.4 Gpixels/sec |
| 48090.4 FPS | Particle Physics | 23630 FPS |
| OpenCL | API | OpenCL |
| 316301 | GB6 Compute Score | 182816 |
| 318.9 img/sec | Background Blur | 427.4 img/sec |
| 222.8 img/sec | Face Detection | 251 img/sec |
| 14.2 Gpixels/sec | Horizon Detection | 6.77 Gpixels/sec |
| 21 Gpixels/sec | Edge Detection | 9.42 Gpixels/sec |
| 23.9 Gpixels/sec | Gaussian Blur | 9.35 Gpixels/sec |
| 38071 | G3D Mark Score | 26904 |
| 1303 | G2D Mark | 1319 |
| 324 FPS | DirectX 11 | 295 FPS |
| 151 FPS | DirectX 12 | 77 FPS |
| 26193 Ops/s | GPU Compute | 15995 Ops/s |
| 42962 | GB6 ML Single Precision | 39190 |
| 59636 | GB6 ML Half Precision | 55727 |
| 31670 | GB6 ML Quantized | 33599 |
| 15600 | Image Classification (SP) | 18001 |
| 36657 | Image Segmentation (HP) | 25174 |
| 48114 | Image Super Resolution (Q) | 45714 |
| 78111 | Face Detection (HP) | 74776 |
| 289639 | Pose Estimation (Q) | 196137 |
| 4074 | Text Classification (SP) | 5163 |
| 6349 | Machine Translation (HP) | 8677 |
| 21186 | Object Detection (SP) | 19602 |
| 75599 | Depth Estimation (Q) | 74879 |
| 601428 | Style Transfer (SP) | 411302 |
| ONNX | Framework | ONNX |
| DirectML | Backend | DirectML |
| 444 GPixel/s | Pixel Fill Rate | 380 GPixel/s |
| 1290 GTexel/s | Texture Fill Rate | 760 GTexel/s |
| 82.6 TFLOPS | FLOPS (FP32) | 48.7 TFLOPS |
| GDDR6X | Memory Type | GDDR6 |
| 24 GB | Memory Size | 16 GB |
| 2625 MHz | Memory Clock | 2518 MHz |
| 21000 Mbps | Effective Memory Speed | 20100 Mbps |
| 384-bit | Bus | 256-bit |
| No | ECC | No |
| 1010 GB/s | Memory Bandwidth | 640 GB/s |
| PCIe 4.0 x16 | Interface | PCIe 5.0 x16 |
| 450 W | TGP | 304 W |
| TSMC | Manufacturing | TSMC |
| 5 nm | Fabrication Process | 4 nm |
| 609 mm² | Die Size | 356.5 mm² |
| 76.3 billion | Transistor Count | 53.9 billion |
| 125.29 MTr/mm² | Transistor Density | 151.19 MTr/mm² |
| 12 | DirectX | 12 |
| 1.3 | Vulkan | 1.3 |
| 4.6 | OpenGL | 4.6 |
| 3 | OpenCL | 2.2 |
| 8.9 | CUDA | No |
| Yes | Ray Tracing | Yes |
| DLSS 3 | DLSS | No |
| 1.4a | DisplayPort | 2.1a |
| Nvidia | Vendor | Amd |
| Discrete | Build | Discrete |
| September 20, 2022 | Released | January 7, 2025 |
| Desktop | Case | Desktop |
| Gaming | Purpose | Gaming |
| High-end | Segment | High-end |
| Ada Lovelace | Architecture | RDNA 4.0 |
| AD102 | GPU Codename | Navi 48 |
| -Intel Core i9 14900Kor above | Recommended CPU | -Intel Core Ultra 5 245Kor above |
| 42169 | Steel Nomad Lite Score | 26180 |
| 36311 | Time Spy | 29701 |
| 187386 | Solar Bay | 121225 |
| 26128 | Port Royal | 18525 |
| 72387 | Fire Strike | 66381 |
| 85230 | Wild Life Extreme | 54685 |
| 195880 | Night Raid | 195698 |
| 2235 MHz | Base Clock | 1660 MHz |
| 2520 MHz | Boost Clock | 2970 MHz |
| 16384 | Shading Units | 4096 |
| 512 | Texture Mapping Units (TMUs) | 256 |
| 176 | Render Output Units (ROPs) | 128 |
| 128 | Compute Units (Pipelines) | 64 |
| 512 | Tensor Cores | 128 |
| 128 | Ray-tracing Cores | 64 |
| 128KB per cluster | L1 Cache | 128KB per cluster |
| 72MB shared | L2 Cache | 4MB shared |
| 2 IPC | Instructions Per Cycle | 4 IPC |
Compare with something else?
Compare GeForce RTX 4090 with another →