Quick/Comparisons
66 | Gaming | 56 |
66 | Workstation | 67 |
57 | AI/ML |
Compare with something else?
Compare GeForce RTX 4070 with another →53 |
44 | Energy Efficiency | 49 |
64 | Quick Comparison Final Score | 57 |
| AD104 | Graphics Processor | GB203-300-A1 |
| 7680 | Cores | 7680 |
| 240 / 80 | TMUs / ROPs | 240 / 96 |
| 6639.7 | Blender GPU | 5699.95 |
| 141 | Forza Horizon 5 | 114 |
| 182 | The Witcher 3 | 157 |
| 216 | Counter-Strike 2 | 204 |
| 148 | Far Cry 6 | 132 |
| 114 | Hogwarts Legacy | 96 |
| 171 | Call of Duty | 143 |
| 106 | Ghost of Tsushima | 87 |
| 142 | Cyberpunk 2077 | 118 |
| 245 | Tomb Raider | 197 |
| 163 | Average FPS | 139 |
| 194 | 1080p High | 166 |
| 163 | 1080p Ultra | 139 |
| 128 | 1440p Ultra | 106 |
| 75 | 4K Ultra | 63 |
| Low | Margin of Error | High |
| 219163 | GB6 Compute Score | 197909 |
| 371.2 img/sec | Background Blur | 267.1 img/sec |
| 235 img/sec | Face Detection | 165.5 img/sec |
| 7.51 Gpixels/sec | Horizon Detection | 8.72 Gpixels/sec |
| 10.6 Gpixels/sec | Edge Detection | 12.3 Gpixels/sec |
| 12.4 Gpixels/sec | Gaussian Blur | 12.6 Gpixels/sec |
| 2.03 Gpixels/sec | Feature Matching | 1.83 Gpixels/sec |
| 1120 Gpixels/sec | Stereo Matching | 942.6 Gpixels/sec |
| 30420.2 FPS | Particle Physics | 25637.1 FPS |
| OpenCL | API | OpenCL |
| 31568 | G3D Mark Score | 23401 |
| 1209 | G2D Mark | 874 |
| 288 FPS | DirectX 11 | 232 FPS |
| 116 FPS | DirectX 12 | 109 FPS |
| 18221 Ops/s | GPU Compute | 8154 Ops/s |
| 32405 | GB6 ML Single Precision | 27844 |
| 48989 | GB6 ML Half Precision | 48544 |
| 24894 | GB6 ML Quantized | 21478 |
| 12496 | Image Classification (SP) | 11723 |
| 33378 | Image Segmentation (HP) | 32142 |
| 37180 | Image Super Resolution (Q) | 30735 |
| 62919 | Face Detection (HP) | 58722 |
| 171125 | Pose Estimation (Q) | 129551 |
| 3630 | Text Classification (SP) | 3539 |
| 6420 | Machine Translation (HP) | 6242 |
| 15804 | Object Detection (SP) | 14495 |
| 58647 | Depth Estimation (Q) | 50690 |
| 331379 | Style Transfer (SP) | 231154 |
| ONNX | Framework | ONNX |
| DirectML | Backend | DirectML |
| 209 GPixel/s | Pixel Fill Rate | 163-180 GPixel/s |
| 626 GTexel/s | Texture Fill Rate | 408-450 GTexel/s |
| 40.1 TFLOPS | FLOPS (FP32) | 26.1-28.8 TFLOPS |
| GDDR6X | Memory Type | GDDR7 |
| 12 GB | Memory Size | 16 GB |
| 1313 MHz | Memory Clock | 1750 MHz |
| 21000 Mbps | Effective Memory Speed | 28000 Mbps |
| 192-bit | Bus | 256-bit |
| No | ECC | No |
| 504 GB/s | Memory Bandwidth | 896 GB/s |
| PCIe 4.0 x16 | Interface | PCIe 5.0 x16 |
| 285 W | TGP | 80-175 W (configurable) |
| TSMC | Manufacturing | TSMC |
| 5 nm | Fabrication Process | 5 nm |
| 294 mm² | Die Size | 378 mm² |
| 35 billion | Transistor Count | 45.6 billion |
| 119.05 MTr/mm² | Transistor Density | 120.63 MTr/mm² |
| 12 | DirectX | 12.2 |
| 1.3 | Vulkan | 1.4 |
| 4.6 | OpenGL | 4.6 |
| 3 | OpenCL | 3 |
| 8.9 | CUDA | 12 |
| Yes | Ray Tracing | Yes |
| DLSS 3 | DLSS | DLSS 4 |
| 1.4a | DisplayPort | 2.1b |
| Mid-range | Segment | High-end |
| Ada Lovelace | Architecture | Blackwell 2.0 |
| AD104 | GPU Codename | GB203-300-A1 |
| -Intel Core i7 14700Kor above | Recommended CPU | -Intel Core Ultra 9 275HXor above |
| Nvidia | Vendor | Nvidia |
| Discrete | Build | Discrete |
| January 3, 2023 | Released | March 18, 2025 |
| Desktop | Case | Laptop |
| Gaming | Purpose | Professional |
| 23618 | Steel Nomad Lite Score | 22035 |
| 22751 | Time Spy | 13961 |
| 108422 | Solar Bay | — |
| 13994 | Port Royal | 12094 |
| 53484 | Fire Strike | 38402 |
| 45397 | Wild Life Extreme | — |
| 171258 | Night Raid | — |
| 2310 MHz | Base Clock | 975-1702 MHz (configurable) |
| 2610 MHz | Boost Clock | 1702-1875 MHz (configurable) |
| 7680 | Shading Units | 7680 |
| 240 | Texture Mapping Units (TMUs) | 240 |
| 80 | Render Output Units (ROPs) | 96 |
| 60 | Compute Units (Pipelines) | 60 |
| 240 | Tensor Cores | 240 |
| 60 | Ray-tracing Cores | 60 |
| 128KB per cluster | L1 Cache | 128KB per cluster |
| 48MB shared | L2 Cache | 64MB shared |
| 2 IPC | Instructions Per Cycle | 2 IPC |