Quick/Comparisons
54 | Gaming | 16 |
55 | Workstation | 10 |
48 | AI/ML |
Compare with something else?
Compare GeForce RTX 4070 with another →7 |
44 | Energy Efficiency | 94 |
53 | Quick Comparison Final Score | 26 |
| AD104 | Graphics Processor | — |
| 5888 | Cores | 512 |
| 184 / 64 | TMUs / ROPs | 32 / 16 |
| 5394.14 | Blender GPU | — |
| 114 | Forza Horizon 5 | — |
| 163 | The Witcher 3 | — |
| 202 | Counter-Strike 2 | — |
| 135 | Far Cry 6 | — |
| 96 | Hogwarts Legacy | — |
| 143 | Call of Duty | — |
| 89 | Ghost of Tsushima | — |
| 115 | Cyberpunk 2077 | — |
| 196 | Tomb Raider | — |
| 139 | Average FPS | — |
| 165 | 1080p High | — |
| 139 | 1080p Ultra | — |
| 107 | 1440p Ultra | — |
| 63 | 4K Ultra | — |
| Low | Margin of Error | — |
| 167522 | GB6 Compute Score | 16321 |
| 233 img/sec | Background Blur | 37.6 img/sec |
| 141.7 img/sec | Face Detection | 17.2 img/sec |
| 6.62 Gpixels/sec | Horizon Detection | 0.3 Gpixels/sec |
| 9.19 Gpixels/sec | Edge Detection | 0.63 Gpixels/sec |
| 9.2 Gpixels/sec | Gaussian Blur | 0.94 Gpixels/sec |
| 1.78 Gpixels/sec | Feature Matching | 0.22 Gpixels/sec |
| 845.4 Gpixels/sec | Stereo Matching | 60 Gpixels/sec |
| 25043.9 FPS | Particle Physics | 3695.9 FPS |
| OpenCL | API | OpenCL |
| 26921 | G3D Mark Score | — |
| 1168 | G2D Mark | — |
| 243 FPS | DirectX 11 | — |
| 103 FPS | DirectX 12 | — |
| 14623 Ops/s | GPU Compute | — |
| 27166 | GB6 ML Single Precision | 3770 |
| 40963 | GB6 ML Half Precision | 5146 |
| 20332 | GB6 ML Quantized | 2824 |
| 10246 | Image Classification (SP) | 2902 |
| 26153 | Image Segmentation (HP) | 5473 |
| 32054 | Image Super Resolution (Q) | 2967 |
| 51337 | Face Detection (HP) | 6068 |
| 124617 | Pose Estimation (Q) | 8220 |
| 3703 | Text Classification (SP) | 609 |
| 5484 | Machine Translation (HP) | 1146 |
| 13134 | Object Detection (SP) | 2766 |
| 50548 | Depth Estimation (Q) | 6405 |
| 263477 | Style Transfer (SP) | 26796 |
| ONNX | Framework | ONNX |
| DirectML | Backend | DirectML |
| 158 GPixel/s | Pixel Fill Rate | 28-32 GPixel/s |
| 455 GTexel/s | Texture Fill Rate | 56-64 GTexel/s |
| 29.1 TFLOPS | FLOPS (FP32) | 1.8-2 TFLOPS |
| GDDR6X | Memory Type | System Shared |
| 12 GB | Memory Size | — |
| 1313 MHz | Memory Clock | — |
| 21000 Mbps | Effective Memory Speed | — |
| 192-bit | Bus | — |
| No | ECC | — |
| 504 GB/s | Memory Bandwidth | — |
| PCIe 4.0 x16 | Interface | Ring Bus |
| 200 W | TGP | 14 W |
| TSMC | Manufacturing | TSMC |
| 5 nm | Fabrication Process | 5 nm |
| 294 mm² | Die Size | — |
| 35 billion | Transistor Count | — |
| 119.05 MTr/mm² | Transistor Density | — |
| 12 | DirectX | 12 |
| 1.3 | Vulkan | 1.3 |
| 4.6 | OpenGL | 4.6 |
| 3 | OpenCL | 3 |
| 8.9 | CUDA | — |
| Yes | Ray Tracing | Yes |
| DLSS 3 | DLSS | No |
| 1.4a | DisplayPort | 2.1 |
| Nvidia | Vendor | Intel |
| Discrete | Build | Integrated |
| April 13, 2023 | Released | December 14, 2023 |
| Desktop | Case | Laptop |
| Gaming | Purpose | Generic |
| Mid-range | Segment | Mid-range |
| Ada Lovelace | Architecture | Xe LPG |
| AD104 | GPU Codename | — |
| -Intel Core i7 14700Kor above | Recommended CPU | -Intel Core Ultra 5 125Uor above |
| 18089 | Steel Nomad Lite Score | — |
| 17806 | Time Spy | — |
| 82784 | Solar Bay | — |
| 11154 | Port Royal | — |
| 44075 | Fire Strike | — |
| 35470 | Wild Life Extreme | — |
| 153508 | Night Raid | — |
| 1920 MHz | Base Clock | 300 MHz |
| 2475 MHz | Boost Clock | 1750-2000 MHz (configurable) |
| 5888 | Shading Units | 512 |
| 184 | Texture Mapping Units (TMUs) | 32 |
| 64 | Render Output Units (ROPs) | 16 |
| 46 | Compute Units (Pipelines) | 64 |
| 184 | Tensor Cores | — |
| 46 | Ray-tracing Cores | 4 |
| 128KB per cluster | L1 Cache | 768KB shared |
| 36MB shared | L2 Cache | — |
| 2 IPC | Instructions Per Cycle | 2 IPC |