Quick/Comparisons
61 | Gaming | 12 |
62 | Workstation | 23 |
54 | AI/ML |
Compare with something else?
Compare GeForce RTX 4070 with another →14 |
46 | Energy Efficiency | 72 |
60 | Quick Comparison Final Score | 25 |
| AD104 | Graphics Processor | GA107S |
| 7168 | Cores | 2048 |
| 224 / 80 | TMUs / ROPs | 64 / 32 |
| 6153.24 | Blender GPU | 692.94 |
| 131 | Forza Horizon 5 | 24 |
| 176 | The Witcher 3 | 26 |
| 212 | Counter-Strike 2 | 42 |
| 145 | Far Cry 6 | 33 |
| 106 | Hogwarts Legacy | 14 |
| 165 | Call of Duty | 26 |
| 100 | Ghost of Tsushima | 13 |
| 131 | Cyberpunk 2077 | 18 |
| 226 | Tomb Raider | 36 |
| 155 | Average FPS | 26 |
| 183 | 1080p High | 35 |
| 155 | 1080p Ultra | 26 |
| 120 | 1440p Ultra | 18 |
| 70 | 4K Ultra | 9 |
| Low | Margin of Error | Medium |
| 206598 | GB6 Compute Score | 40574 |
| 359.5 img/sec | Background Blur | 62.4 img/sec |
| 210 img/sec | Face Detection | 35.7 img/sec |
| 7.61 Gpixels/sec | Horizon Detection | 1.62 Gpixels/sec |
| 10.4 Gpixels/sec | Edge Detection | 2.11 Gpixels/sec |
| 11.3 Gpixels/sec | Gaussian Blur | 2.05 Gpixels/sec |
| 1.97 Gpixels/sec | Feature Matching | 0.52 Gpixels/sec |
| 969.3 Gpixels/sec | Stereo Matching | 185.1 Gpixels/sec |
| 29061.4 FPS | Particle Physics | 5502.5 FPS |
| OpenCL | API | OpenCL |
| 29964 | G3D Mark Score | 6570 |
| 1193 | G2D Mark | 416 |
| 271 FPS | DirectX 11 | 40 FPS |
| 110 FPS | DirectX 12 | 38 FPS |
| 16985 Ops/s | GPU Compute | 2911 Ops/s |
| 30706 | GB6 ML Single Precision | 7597 |
| 43851 | GB6 ML Half Precision | 15030 |
| 23776 | GB6 ML Quantized | 5337 |
| 12146 | Image Classification (SP) | 3298 |
| 23932 | Image Segmentation (HP) | 8358 |
| 36321 | Image Super Resolution (Q) | 6219 |
| 52836 | Face Detection (HP) | 14319 |
| 179788 | Pose Estimation (Q) | 22995 |
| 3371 | Text Classification (SP) | 2226 |
| 5017 | Machine Translation (HP) | 3163 |
| 15164 | Object Detection (SP) | 3768 |
| 62500 | Depth Estimation (Q) | 12079 |
| 357489 | Style Transfer (SP) | 53139 |
| ONNX | Framework | ONNX |
| DirectML | Backend | DirectML |
| 198 GPixel/s | Pixel Fill Rate | 24-55 GPixel/s |
| 554 GTexel/s | Texture Fill Rate | 48-109 GTexel/s |
| 35.5 TFLOPS | FLOPS (FP32) | 3.1-7 TFLOPS |
| GDDR6X | Memory Type | GDDR6 |
| 12 GB | Memory Size | 4 GB |
| 1313 MHz | Memory Clock | 1500 MHz |
| 21000 Mbps | Effective Memory Speed | 12000 Mbps |
| 192-bit | Bus | 64-bit |
| No | ECC | No |
| 504.2 GB/s | Memory Bandwidth | 96 GB/s |
| PCIe 4.0 x16 | Interface | PCIe 4.0 x8 |
| 220 W | TGP | 20-60 W (configurable) |
| TSMC | Manufacturing | Samsung |
| 5 nm | Fabrication Process | 8 nm |
| 294 mm² | Die Size | 200 mm² |
| 35 billion | Transistor Count | 8.7 billion |
| 119.05 MTr/mm² | Transistor Density | 43.5 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 17, 2024 | Released | March 22, 2022 |
| Desktop | Case | Laptop |
| Gaming | Purpose | Professional |
| Mid-range | Segment | Entry-level |
| Ada Lovelace | Architecture | Ampere |
| AD104 | GPU Codename | GA107S |
| -Intel Core i7 14700Kor above | Recommended CPU | -Intel Core i7 12700Hor above |
| 21568 | Steel Nomad Lite Score | 3021 |
| 20882 | Time Spy | 3023 |
| 98166 | Solar Bay | 13710 |
| 13122 | Port Royal | 625 |
| 50092 | Fire Strike | 10265 |
| 41630 | Wild Life Extreme | 4053 |
| 164777 | Night Raid | 38312 |
| 1980 MHz | Base Clock | 370-890 MHz (configurable) |
| 2475 MHz | Boost Clock | 745-1710 MHz (configurable) |
| 7168 | Shading Units | 2048 |
| 224 | Texture Mapping Units (TMUs) | 64 |
| 80 | Render Output Units (ROPs) | 32 |
| 56 | Compute Units (Pipelines) | 16 |
| 224 | Tensor Cores | 64 |
| 56 | Ray-tracing Cores | 16 |
| 128KB per cluster | L1 Cache | 128KB per cluster |
| 48MB shared | L2 Cache | 2MB shared |
| 2 IPC | Instructions Per Cycle | 2 IPC |