Quick/Comparisons
43 | Gaming | 32 |
44 | Workstation | 33 |
35 | AI/ML |
Compare with something else?
Compare GeForce RTX 3070 with another →41 |
35 | Energy Efficiency | 66 |
42 | Quick Comparison Final Score | 39 |
| GA104 | Graphics Processor | Strix Halo |
| 5888 | Cores | 2560 |
| 184 / 96 | TMUs / ROPs | 160 / 64 |
| 3368.42 | Blender GPU | 1335.18 |
| 99 | Forza Horizon 5 | 77 |
| 143 | The Witcher 3 | 101 |
| 194 | Counter-Strike 2 | 138 |
| 121 | Far Cry 6 | 92 |
| 80 | Hogwarts Legacy | 61 |
| 113 | Call of Duty | 89 |
| 74 | Ghost of Tsushima | 64 |
| 94 | Cyberpunk 2077 | 65 |
| 153 | Tomb Raider | 115 |
| 119 | Average FPS | 89 |
| 141 | 1080p High | 114 |
| 119 | 1080p Ultra | 89 |
| 88 | 1440p Ultra | 64 |
| 50 | 4K Ultra | 34 |
| Low | Margin of Error | Low |
| 126195 | GB6 Compute Score | 93515 |
| 161.6 img/sec | Background Blur | 232.1 img/sec |
| 117.2 img/sec | Face Detection | 122.5 img/sec |
| 5.89 Gpixels/sec | Horizon Detection | 2.92 Gpixels/sec |
| 8.05 Gpixels/sec | Edge Detection | 4.07 Gpixels/sec |
| 6.65 Gpixels/sec | Gaussian Blur | 3.49 Gpixels/sec |
| 1.25 Gpixels/sec | Feature Matching | 1.08 Gpixels/sec |
| 550.6 Gpixels/sec | Stereo Matching | 396.3 Gpixels/sec |
| 17613.1 FPS | Particle Physics | 13016.5 FPS |
| OpenCL | API | OpenCL |
| 22128 | G3D Mark Score | 18115 |
| 1005 | G2D Mark | 1172 |
| 182 FPS | DirectX 11 | 157 FPS |
| 84 FPS | DirectX 12 | 84 FPS |
| 11062 Ops/s | GPU Compute | 9062 Ops/s |
| 19789 | GB6 ML Single Precision | 22565 |
| 33013 | GB6 ML Half Precision | 30098 |
| 12443 | GB6 ML Quantized | 18100 |
| 7624 | Image Classification (SP) | 11176 |
| 23420 | Image Segmentation (HP) | 14027 |
| 17898 | Image Super Resolution (Q) | 27198 |
| 40535 | Face Detection (HP) | 40613 |
| 83624 | Pose Estimation (Q) | 92878 |
| 2724 | Text Classification (SP) | 4158 |
| 4481 | Machine Translation (HP) | 6473 |
| 10304 | Object Detection (SP) | 7431 |
| 30418 | Depth Estimation (Q) | 41825 |
| 180089 | Style Transfer (SP) | 190221 |
| ONNX | Framework | ONNX |
| DirectML | Backend | DirectML |
| 166 GPixel/s | Pixel Fill Rate | 186 GPixel/s |
| 317 GTexel/s | Texture Fill Rate | 464 GTexel/s |
| 20.3 TFLOPS | FLOPS (FP32) | 14.8 TFLOPS |
| GDDR6 | Memory Type | System Shared |
| 8 GB | Memory Size | — |
| 1750 MHz | Memory Clock | — |
| 14000 Mbps | Effective Memory Speed | — |
| 256-bit | Bus | — |
| No | ECC | — |
| 448 GB/s | Memory Bandwidth | — |
| PCIe 4.0 x16 | Interface | PCIe 5.0 x16 |
| 220 W | TGP | 45-120 W (configurable) |
| Samsung | Manufacturing | TSMC |
| 8 nm | Fabrication Process | 4 nm |
| 392 mm² | Die Size | 308 mm² |
| 17 billion | Transistor Count | — |
| 43.37 MTr/mm² | Transistor Density | — |
| 12 | DirectX | 12.2 |
| 1.3 | Vulkan | 1.4 |
| 4.6 | OpenGL | 4.6 |
| 3 | OpenCL | 2.1 |
| 8.6 | CUDA | No |
| Yes | Ray Tracing | Yes |
| DLSS 2 | DLSS | No |
| 1.4a | DisplayPort | 2.1a |
| Nvidia | Vendor | Amd |
| Discrete | Build | Integrated |
| October 22, 2020 | Released | January 7, 2025 |
| Desktop | Case | Laptop |
| Gaming | Purpose | Generic |
| Mid-range | Segment | Mid-range |
| Ampere | Architecture | RDNA 3.5 |
| GA104 | GPU Codename | Strix Halo |
| -Intel Core i7 12700Kor above | Recommended CPU | -AMD Ryzen AI Max+ 395or above |
| 14153 | Steel Nomad Lite Score | 9818 |
| 13643 | Time Spy | 10010 |
| 64426 | Solar Bay | 46934 |
| 8311 | Port Royal | 5512 |
| 34054 | Fire Strike | 26722 |
| 28486 | Wild Life Extreme | 19043 |
| 122039 | Night Raid | 100428 |
| 1500 MHz | Base Clock | 1295 MHz |
| 1725 MHz | Boost Clock | 2900 MHz |
| 5888 | Shading Units | 2560 |
| 184 | Texture Mapping Units (TMUs) | 160 |
| 96 | Render Output Units (ROPs) | 64 |
| 46 | Compute Units (Pipelines) | 40 |
| 184 | Tensor Cores | No |
| 46 | Ray-tracing Cores | 40 |
| 128KB per cluster | L1 Cache | 512KB per cluster |
| 4MB shared | L2 Cache | 2MB shared |
| 2 IPC | Instructions Per Cycle | 2 IPC |