Quick/Comparisons
61 | Gaming | 9 |
62 | Workstation | 12 |
54 | AI/ML |
Compare with something else?
Compare GeForce RTX 4070 with another →13 |
46 | Energy Efficiency | 86 |
60 | Quick Comparison Final Score | 23 |
| AD104 | Graphics Processor | Krackan Point |
| 7168 | Cores | 512 |
| 224 / 80 | TMUs / ROPs | 32 / 16 |
| 6153.24 | Blender GPU | 193.82 |
| 131 | Forza Horizon 5 | 24 |
| 176 | The Witcher 3 | 18 |
| 212 | Counter-Strike 2 | 26 |
| 145 | Far Cry 6 | 26 |
| 106 | Hogwarts Legacy | 12 |
| 165 | Call of Duty | 27 |
| 100 | Ghost of Tsushima | 12 |
| 131 | Cyberpunk 2077 | 13 |
| 226 | Tomb Raider | 22 |
| 155 | Average FPS | 20 |
| 183 | 1080p High | 26 |
| 155 | 1080p Ultra | 20 |
| 120 | 1440p Ultra | 14 |
| 70 | 4K Ultra | - |
| Low | Margin of Error | Low |
| 206598 | GB6 Compute Score | 22911 |
| 359.5 img/sec | Background Blur | 57.2 img/sec |
| 210 img/sec | Face Detection | 30.5 img/sec |
| 7.61 Gpixels/sec | Horizon Detection | 0.83 Gpixels/sec |
| 10.4 Gpixels/sec | Edge Detection | 1.16 Gpixels/sec |
| 11.3 Gpixels/sec | Gaussian Blur | 0.93 Gpixels/sec |
| 1.97 Gpixels/sec | Feature Matching | 0.35 Gpixels/sec |
| 969.3 Gpixels/sec | Stereo Matching | 97.8 Gpixels/sec |
| 29061.4 FPS | Particle Physics | 3933.5 FPS |
| OpenCL | API | OpenCL |
| 29964 | G3D Mark Score | 4877 |
| 1193 | G2D Mark | 916 |
| 271 FPS | DirectX 11 | 37 FPS |
| 110 FPS | DirectX 12 | 24 FPS |
| 16985 Ops/s | GPU Compute | 2421 Ops/s |
| 30706 | GB6 ML Single Precision | 6947 |
| 43851 | GB6 ML Half Precision | 10338 |
| 23776 | GB6 ML Quantized | 5334 |
| 12146 | Image Classification (SP) | 2022 |
| 23932 | Image Segmentation (HP) | 3987 |
| 36321 | Image Super Resolution (Q) | 6657 |
| 52836 | Face Detection (HP) | 10337 |
| 179788 | Pose Estimation (Q) | 28437 |
| 3371 | Text Classification (SP) | 2326 |
| 5017 | Machine Translation (HP) | 4011 |
| 15164 | Object Detection (SP) | 1995 |
| 62500 | Depth Estimation (Q) | 12031 |
| 357489 | Style Transfer (SP) | 56710 |
| ONNX | Framework | ONNX |
| DirectML | Backend | DirectML |
| 198 GPixel/s | Pixel Fill Rate | 48 GPixel/s |
| 554 GTexel/s | Texture Fill Rate | 96 GTexel/s |
| 35.5 TFLOPS | FLOPS (FP32) | 3.1 TFLOPS |
| GDDR6X | Memory Type | System Shared |
| 12 GB | Memory Size | — |
| 1313 MHz | Memory Clock | — |
| 21000 Mbps | Effective Memory Speed | — |
| 192-bit | Bus | — |
| No | ECC | — |
| 504.2 GB/s | Memory Bandwidth | — |
| PCIe 4.0 x16 | Interface | PCIe 4.0 x16 |
| 220 W | TGP | 15-54 W (configurable) |
| TSMC | Manufacturing | TSMC |
| 5 nm | Fabrication Process | 4 nm |
| 294 mm² | Die Size | 233 mm² |
| 35 billion | Transistor Count | 34 billion |
| 119.05 MTr/mm² | Transistor Density | 145.92 MTr/mm² |
| 12 | DirectX | 12 |
| 1.3 | Vulkan | 1.4 |
| 4.6 | OpenGL | 4.6 |
| 3 | OpenCL | 2.1 |
| 8.9 | CUDA | No |
| Yes | Ray Tracing | Yes |
| DLSS 3 | DLSS | No |
| 1.4a | DisplayPort | 2.1 |
| Nvidia | Vendor | Amd |
| Discrete | Build | Integrated |
| January 17, 2024 | Released | January 7, 2025 |
| Desktop | Case | Laptop |
| Gaming | Purpose | Generic |
| Mid-range | Segment | Mid-range |
| Ada Lovelace | Architecture | RDNA 3.5 |
| AD104 | GPU Codename | Krackan Point |
| -Intel Core i7 14700Kor above | Recommended CPU | -AMD Ryzen AI 7 350or above |
| 21568 | Steel Nomad Lite Score | 2112 |
| 20882 | Time Spy | 2565 |
| 98166 | Solar Bay | 11270 |
| 13122 | Port Royal | 1103 |
| 50092 | Fire Strike | 6835 |
| 41630 | Wild Life Extreme | 5150 |
| 164777 | Night Raid | 29003 |
| 1980 MHz | Base Clock | 600 MHz |
| 2475 MHz | Boost Clock | 3000 MHz |
| 7168 | Shading Units | 512 |
| 224 | Texture Mapping Units (TMUs) | 32 |
| 80 | Render Output Units (ROPs) | 16 |
| 56 | Compute Units (Pipelines) | 8 |
| 224 | Tensor Cores | No |
| 56 | Ray-tracing Cores | 8 |
| 128KB per cluster | L1 Cache | 512KB per cluster |
| 48MB shared | L2 Cache | 2MB shared |
| 2 IPC | Instructions Per Cycle | 2 IPC |