Quick/Comparisons
70 | Gaming | 33 |
72 | Workstation | 51 |
59 | AI/ML |
Compare with something else?
Compare GeForce RTX 4070 with another →33 |
46 | Energy Efficiency | 39 |
69 | Quick Comparison Final Score | 39 |
| AD103 | Graphics Processor | GA104 |
| 8448 | Cores | 6144 |
| 264 / 96 | TMUs / ROPs | 192 / 96 |
| 7089.35 | Blender GPU | 2928.42 |
| 150 | Forza Horizon 5 | 72 |
| 198 | The Witcher 3 | 104 |
| 220 | Counter-Strike 2 | 140 |
| 152 | Far Cry 6 | 89 |
| 118 | Hogwarts Legacy | 57 |
| 188 | Call of Duty | 77 |
| 111 | Ghost of Tsushima | 54 |
| 155 | Cyberpunk 2077 | 67 |
| 249 | Tomb Raider | 103 |
| 171 | Average FPS | 85 |
| 200 | 1080p High | 104 |
| 171 | 1080p Ultra | 85 |
| 134 | 1440p Ultra | 62 |
| 79 | 4K Ultra | 34 |
| Low | Margin of Error | High |
| 237068 | GB6 Compute Score | 111855 |
| 331.7 img/sec | Background Blur | 172.7 img/sec |
| 211.6 img/sec | Face Detection | 103.8 img/sec |
| 9.63 Gpixels/sec | Horizon Detection | 5.34 Gpixels/sec |
| 13.7 Gpixels/sec | Edge Detection | 7.2 Gpixels/sec |
| 13.9 Gpixels/sec | Gaussian Blur | 5.61 Gpixels/sec |
| 2.15 Gpixels/sec | Feature Matching | 1.05 Gpixels/sec |
| 1210 Gpixels/sec | Stereo Matching | 474.4 Gpixels/sec |
| 33566 FPS | Particle Physics | 14218.2 FPS |
| OpenCL | API | OpenCL |
| 31798 | G3D Mark Score | 16426 |
| 1240 | G2D Mark | 663 |
| 277 FPS | DirectX 11 | 138 FPS |
| 119 FPS | DirectX 12 | 75 FPS |
| 18264 Ops/s | GPU Compute | 7003 Ops/s |
| 32793 | GB6 ML Single Precision | 15655 |
| 47412 | GB6 ML Half Precision | 24490 |
| 24566 | GB6 ML Quantized | 10684 |
| 12970 | Image Classification (SP) | 6590 |
| 26364 | Image Segmentation (HP) | 12226 |
| 37217 | Image Super Resolution (Q) | 16682 |
| 58720 | Face Detection (HP) | 30795 |
| 177604 | Pose Estimation (Q) | 71291 |
| 3815 | Text Classification (SP) | 2115 |
| 5829 | Machine Translation (HP) | 3207 |
| 15675 | Object Detection (SP) | 8583 |
| 62447 | Depth Estimation (Q) | 25256 |
| 357918 | Style Transfer (SP) | 149822 |
| ONNX | Framework | ONNX |
| DirectML | Backend | DirectML |
| 251 GPixel/s | Pixel Fill Rate | 120-170 GPixel/s |
| 689 GTexel/s | Texture Fill Rate | 239-340 GTexel/s |
| 44.1 TFLOPS | FLOPS (FP32) | 15.3-21.7 TFLOPS |
| GDDR6X | Memory Type | GDDR6 |
| 16 GB | Memory Size | 16 GB |
| 1313 MHz | Memory Clock | 1750 MHz |
| 21000 Mbps | Effective Memory Speed | 14000 Mbps |
| 256-bit | Bus | 256-bit |
| No | ECC | Yes |
| 672.3 GB/s | Memory Bandwidth | 448 GB/s |
| PCIe 4.0 x16 | Interface | PCIe 4.0 x16 |
| 285 W | TGP | 80-165 W (configurable) |
| TSMC | Manufacturing | Samsung |
| 5 nm | Fabrication Process | 8 nm |
| 379 mm² | Die Size | 392 mm² |
| 45 billion | Transistor Count | 17.4 billion |
| 118.73 MTr/mm² | Transistor Density | 44.39 MTr/mm² |
| 12 | DirectX | 12 |
| 1.3 | Vulkan | 1.3 |
| 4.6 | OpenGL | 4.6 |
| 3 | OpenCL | 3 |
| 8.9 | CUDA | 8.6 |
| Yes | Ray Tracing | Yes |
| DLSS 3 | DLSS | DLSS 2 |
| 1.4a | DisplayPort | 1.4a |
| Nvidia | Vendor | Nvidia |
| Discrete | Build | Discrete |
| January 24, 2024 | Released | April 12, 2021 |
| Desktop | Case | Laptop |
| Gaming | Purpose | Professional |
| High-end | Segment | High-end |
| Ada Lovelace | Architecture | Ampere |
| AD103 | GPU Codename | GA104 |
| -Intel Core i7 14700Kor above | Recommended CPU | -Intel Core i9 12900HXor above |
| 25741 | Steel Nomad Lite Score | 9479 |
| 24222 | Time Spy | 9291 |
| 116542 | Solar Bay | 43275 |
| 15802 | Port Royal | 5797 |
| 56340 | Fire Strike | 25901 |
| 49901 | Wild Life Extreme | 20653 |
| 176609 | Night Raid | 61009 |
| 2340 MHz | Base Clock | 780-1215 MHz (configurable) |
| 2610 MHz | Boost Clock | 1245-1770 MHz (configurable) |
| 8448 | Shading Units | 6144 |
| 264 | Texture Mapping Units (TMUs) | 192 |
| 96 | Render Output Units (ROPs) | 96 |
| 66 | Compute Units (Pipelines) | 48 |
| 264 | Tensor Cores | 192 |
| 66 | Ray-tracing Cores | 48 |
| 128KB per cluster | L1 Cache | 128KB per cluster |
| 48MB shared | L2 Cache | 4MB shared |
| 2 IPC | Instructions Per Cycle | 2 IPC |