Quick/Comparisons
66 | Gaming | 4 |
66 | Workstation | 12 |
57 | AI/ML |
Compare with something else?
Compare GeForce RTX 4070 with another →4 |
44 | Energy Efficiency | 71 |
64 | Quick Comparison Final Score | 17 |
| AD104 | Graphics Processor | Raven |
| 7680 | Cores | 640 |
| 240 / 80 | TMUs / ROPs | 40 / 16 |
| 6639.7 | Blender GPU | 76.12 |
| 141 | Forza Horizon 5 | - |
| 182 | The Witcher 3 | — |
| 216 | Counter-Strike 2 | 13 |
| 148 | Far Cry 6 | - |
| 114 | Hogwarts Legacy | — |
| 171 | Call of Duty | 11 |
| 106 | Ghost of Tsushima | - |
| 142 | Cyberpunk 2077 | — |
| 245 | Tomb Raider | 10 |
| 163 | Average FPS | - |
| 194 | 1080p High | 11 |
| 163 | 1080p Ultra | - |
| 128 | 1440p Ultra | - |
| 75 | 4K Ultra | - |
| Low | Margin of Error | Low |
| 219163 | GB6 Compute Score | 11268 |
| 371.2 img/sec | Background Blur | 25.3 img/sec |
| 235 img/sec | Face Detection | 12.1 img/sec |
| 7.51 Gpixels/sec | Horizon Detection | 0.39 Gpixels/sec |
| 10.6 Gpixels/sec | Edge Detection | 0.62 Gpixels/sec |
| 12.4 Gpixels/sec | Gaussian Blur | 0.49 Gpixels/sec |
| 2.03 Gpixels/sec | Feature Matching | 0.15 Gpixels/sec |
| 1120 Gpixels/sec | Stereo Matching | 38 Gpixels/sec |
| 30420.2 FPS | Particle Physics | 1381 FPS |
| OpenCL | API | OpenCL |
| 31568 | G3D Mark Score | 1500 |
| 1209 | G2D Mark | 338 |
| 288 FPS | DirectX 11 | 9 FPS |
| 116 FPS | DirectX 12 | 8 FPS |
| 18221 Ops/s | GPU Compute | 827 Ops/s |
| 32405 | GB6 ML Single Precision | 2156 |
| 48989 | GB6 ML Half Precision | 2805 |
| 24894 | GB6 ML Quantized | 1656 |
| 12496 | Image Classification (SP) | 938 |
| 33378 | Image Segmentation (HP) | 1144 |
| 37180 | Image Super Resolution (Q) | 2094 |
| 62919 | Face Detection (HP) | 3148 |
| 171125 | Pose Estimation (Q) | 6847 |
| 3630 | Text Classification (SP) | 625 |
| 6420 | Machine Translation (HP) | 1118 |
| 15804 | Object Detection (SP) | 1039 |
| 58647 | Depth Estimation (Q) | 3791 |
| 331379 | Style Transfer (SP) | 13640 |
| ONNX | Framework | ONNX |
| DirectML | Backend | DirectML |
| 209 GPixel/s | Pixel Fill Rate | 21-22 GPixel/s |
| 626 GTexel/s | Texture Fill Rate | 52-56 GTexel/s |
| 40.1 TFLOPS | FLOPS (FP32) | 1.7-1.8 TFLOPS |
| GDDR6X | Memory Type | System Shared |
| 12 GB | Memory Size | System RAM Dependent |
| 1313 MHz | Memory Clock | System RAM Dependent |
| 21000 Mbps | Effective Memory Speed | System RAM Dependent |
| 192-bit | Bus | — |
| No | ECC | No |
| 504 GB/s | Memory Bandwidth | — |
| PCIe 4.0 x16 | Interface | IGP |
| 285 W | TGP | 15 W |
| TSMC | Manufacturing | GlobalFoundries |
| 5 nm | Fabrication Process | 14 nm |
| 294 mm² | Die Size | 210 mm² |
| 35 billion | Transistor Count | 4.94 billion |
| 119.05 MTr/mm² | Transistor Density | 23.52 MTr/mm² |
| 12 | DirectX | 12 |
| 1.3 | Vulkan | 1.3 |
| 4.6 | OpenGL | 4.6 |
| 3 | OpenCL | 2 |
| 8.9 | CUDA | — |
| Yes | Ray Tracing | No |
| DLSS 3 | DLSS | No |
| 1.4a | DisplayPort | — |
| Mid-range | Segment | Entry-level |
| Ada Lovelace | Architecture | GCN 5.0 |
| AD104 | GPU Codename | Raven |
| -Intel Core i7 14700Kor above | Recommended CPU | -AMD Ryzen 7 3700Uor above |
| Nvidia | Vendor | Amd |
| Discrete | Build | Integrated |
| January 3, 2023 | Released | October 26, 2017 |
| Desktop | Case | Laptop |
| Gaming | Purpose | Generic |
| 23618 | Steel Nomad Lite Score | 789 |
| 22751 | Time Spy | 770 |
| 108422 | Solar Bay | — |
| 13994 | Port Royal | — |
| 53484 | Fire Strike | 2597 |
| 45397 | Wild Life Extreme | 1532 |
| 171258 | Night Raid | 9616 |
| 2310 MHz | Base Clock | 300 MHz |
| 2610 MHz | Boost Clock | 1300-1400 MHz (configurable) |
| 7680 | Shading Units | 640 |
| 240 | Texture Mapping Units (TMUs) | 40 |
| 80 | Render Output Units (ROPs) | 16 |
| 60 | Compute Units (Pipelines) | 10 |
| 240 | Tensor Cores | No |
| 60 | Ray-tracing Cores | No |
| 128KB per cluster | L1 Cache | — |
| 48MB shared | L2 Cache | — |
| 2 IPC | Instructions Per Cycle | 2 IPC |