Quick/Comparisons
66 | Gaming | 16 |
66 | Workstation | 25 |
57 | AI/ML |
Compare with something else?
Compare GeForce RTX 4070 with another →22 |
44 | Energy Efficiency | 100 |
64 | Quick Comparison Final Score | 33 |
| AD104 | Graphics Processor | Panther Lake |
| 7680 | Cores | 1536 |
| 240 / 80 | TMUs / ROPs | 48 / 24 |
| 6639.7 | Blender GPU | 1268.79 |
| 141 | Forza Horizon 5 | 38 |
| 182 | The Witcher 3 | 55 |
| 216 | Counter-Strike 2 | 60 |
| 148 | Far Cry 6 | 49 |
| 114 | Hogwarts Legacy | 21 |
| 171 | Call of Duty | 32 |
| 106 | Ghost of Tsushima | 25 |
| 142 | Cyberpunk 2077 | 33 |
| 245 | Tomb Raider | 52 |
| 163 | Average FPS | 41 |
| 194 | 1080p High | 52 |
| 163 | 1080p Ultra | 41 |
| 128 | 1440p Ultra | 29 |
| 75 | 4K Ultra | 14 |
| Low | Margin of Error | Low |
| 219163 | GB6 Compute Score | 64546 |
| 371.2 img/sec | Background Blur | — |
| 235 img/sec | Face Detection | — |
| 7.51 Gpixels/sec | Horizon Detection | — |
| 10.6 Gpixels/sec | Edge Detection | — |
| 12.4 Gpixels/sec | Gaussian Blur | — |
| 2.03 Gpixels/sec | Feature Matching | — |
| 1120 Gpixels/sec | Stereo Matching | — |
| 30420.2 FPS | Particle Physics | — |
| OpenCL | API | — |
| 31568 | G3D Mark Score | 9029 |
| 1209 | G2D Mark | 786 |
| 288 FPS | DirectX 11 | 70 FPS |
| 116 FPS | DirectX 12 | 50 FPS |
| 18221 Ops/s | GPU Compute | 3702 Ops/s |
| 32405 | GB6 ML Single Precision | 11871 |
| 48989 | GB6 ML Half Precision | 20120 |
| 24894 | GB6 ML Quantized | 7879 |
| 12496 | Image Classification (SP) | 5487 |
| 33378 | Image Segmentation (HP) | 16864 |
| 37180 | Image Super Resolution (Q) | 8788 |
| 62919 | Face Detection (HP) | 25029 |
| 171125 | Pose Estimation (Q) | 18852 |
| 3630 | Text Classification (SP) | 2115 |
| 6420 | Machine Translation (HP) | 4266 |
| 15804 | Object Detection (SP) | 7449 |
| 58647 | Depth Estimation (Q) | 22207 |
| 331379 | Style Transfer (SP) | 95039 |
| ONNX | Framework | ONNX |
| DirectML | Backend | DirectML |
| 209 GPixel/s | Pixel Fill Rate | 60 GPixel/s |
| 626 GTexel/s | Texture Fill Rate | 120 GTexel/s |
| 40.1 TFLOPS | FLOPS (FP32) | 7.7 TFLOPS |
| GDDR6X | Memory Type | System Shared |
| 12 GB | Memory Size | — |
| 1313 MHz | Memory Clock | — |
| 21000 Mbps | Effective Memory Speed | — |
| 192-bit | Bus | — |
| No | ECC | No |
| 504 GB/s | Memory Bandwidth | — |
| PCIe 4.0 x16 | Interface | PCIe 5.0 |
| 285 W | TGP | — |
| TSMC | Manufacturing | TSMC |
| 5 nm | Fabrication Process | 3 nm |
| 294 mm² | Die Size | — |
| 35 billion | Transistor Count | — |
| 119.05 MTr/mm² | Transistor Density | — |
| 12 | DirectX | 12.1 |
| 1.3 | Vulkan | 1.4 |
| 4.6 | OpenGL | 4.6 |
| 3 | OpenCL | 3 |
| 8.9 | CUDA | — |
| Yes | Ray Tracing | Yes |
| DLSS 3 | DLSS | No |
| 1.4a | DisplayPort | 2.1 |
| Mid-range | Segment | Mid-range |
| Ada Lovelace | Architecture | Celestial |
| AD104 | GPU Codename | Panther Lake |
| -Intel Core i7 14700Kor above | Recommended CPU | -Intel Core Ultra X9 388Hor above |
| Nvidia | Vendor | Intel |
| Discrete | Build | Integrated |
| January 3, 2023 | Released | January 5, 2026 |
| Desktop | Case | Laptop |
| Gaming | Purpose | Gaming |
| 23618 | Steel Nomad Lite Score | 4242 |
| 22751 | Time Spy | 6595 |
| 108422 | Solar Bay | 26399 |
| 13994 | Port Royal | 3886 |
| 53484 | Fire Strike | 16770 |
| 45397 | Wild Life Extreme | 11707 |
| 171258 | Night Raid | 68498 |
| 2310 MHz | Base Clock | 300 MHz |
| 2610 MHz | Boost Clock | 2500 MHz |
| 7680 | Shading Units | 1536 |
| 240 | Texture Mapping Units (TMUs) | 48 |
| 80 | Render Output Units (ROPs) | 24 |
| 60 | Compute Units (Pipelines) | 12 |
| 240 | Tensor Cores | — |
| 60 | Ray-tracing Cores | — |
| 128KB per cluster | L1 Cache | 64KB per cluster |
| 48MB shared | L2 Cache | 16MB shared |
| 2 IPC | Instructions Per Cycle | 2 IPC |