Quick/Comparisons
43 | Gaming | 14 |
44 | Workstation | 21 |
35 | AI/ML |
Compare with something else?
Compare GeForce RTX 3070 with another →11 |
35 | Energy Efficiency | 68 |
42 | Quick Comparison Final Score | 25 |
| GA104 | Graphics Processor | TU117 |
| 5888 | Cores | 896 |
| 184 / 96 | TMUs / ROPs | 56 / 32 |
| 3368.42 | Blender GPU | 417.51 |
| 99 | Forza Horizon 5 | — |
| 143 | The Witcher 3 | — |
| 194 | Counter-Strike 2 | — |
| 121 | Far Cry 6 | — |
| 80 | Hogwarts Legacy | — |
| 113 | Call of Duty | — |
| 74 | Ghost of Tsushima | — |
| 94 | Cyberpunk 2077 | — |
| 153 | Tomb Raider | — |
| 119 | Average FPS | — |
| 141 | 1080p High | — |
| 119 | 1080p Ultra | — |
| 88 | 1440p Ultra | — |
| 50 | 4K Ultra | — |
| Low | Margin of Error | — |
| 126195 | GB6 Compute Score | 33110 |
| 161.6 img/sec | Background Blur | 61.6 img/sec |
| 117.2 img/sec | Face Detection | 31.3 img/sec |
| 5.89 Gpixels/sec | Horizon Detection | 1.17 Gpixels/sec |
| 8.05 Gpixels/sec | Edge Detection | 1.48 Gpixels/sec |
| 6.65 Gpixels/sec | Gaussian Blur | 1.8 Gpixels/sec |
| 1.25 Gpixels/sec | Feature Matching | 0.43 Gpixels/sec |
| 550.6 Gpixels/sec | Stereo Matching | 149.1 Gpixels/sec |
| 17613.1 FPS | Particle Physics | 4185.6 FPS |
| OpenCL | API | OpenCL |
| 22128 | G3D Mark Score | 7038 |
| 1005 | G2D Mark | 482 |
| 182 FPS | DirectX 11 | 51 FPS |
| 84 FPS | DirectX 12 | 32 FPS |
| 11062 Ops/s | GPU Compute | 2678 Ops/s |
| 19789 | GB6 ML Single Precision | 6235 |
| 33013 | GB6 ML Half Precision | 4946 |
| 12443 | GB6 ML Quantized | 4599 |
| 7624 | Image Classification (SP) | 2490 |
| 23420 | Image Segmentation (HP) | 3507 |
| 17898 | Image Super Resolution (Q) | 6427 |
| 40535 | Face Detection (HP) | 5241 |
| 83624 | Pose Estimation (Q) | 21102 |
| 2724 | Text Classification (SP) | 1520 |
| 4481 | Machine Translation (HP) | 2350 |
| 10304 | Object Detection (SP) | 2982 |
| 30418 | Depth Estimation (Q) | 10665 |
| 180089 | Style Transfer (SP) | 43183 |
| ONNX | Framework | ONNX |
| DirectML | Backend | DirectML |
| 166 GPixel/s | Pixel Fill Rate | 45 GPixel/s |
| 317 GTexel/s | Texture Fill Rate | 78 GTexel/s |
| 20.3 TFLOPS | FLOPS (FP32) | 2.5 TFLOPS |
| GDDR6 | Memory Type | GDDR6 |
| 8 GB | Memory Size | 4 GB |
| 1750 MHz | Memory Clock | 1250 MHz |
| 14000 Mbps | Effective Memory Speed | 10000 Mbps |
| 256-bit | Bus | 128-bit |
| No | ECC | — |
| 448 GB/s | Memory Bandwidth | 160 GB/s |
| PCIe 4.0 x16 | Interface | PCIe 3.0 x16 |
| 220 W | TGP | 40 W |
| Samsung | Manufacturing | TSMC |
| 8 nm | Fabrication Process | 12 nm |
| 392 mm² | Die Size | 200 mm² |
| 17 billion | Transistor Count | 4.7 billion |
| 43.37 MTr/mm² | Transistor Density | 23.5 MTr/mm² |
| 12 | DirectX | 12 |
| 1.3 | Vulkan | 1.3 |
| 4.6 | OpenGL | 4.6 |
| 3 | OpenCL | 3 |
| 8.6 | CUDA | 7.5 |
| Yes | Ray Tracing | No |
| DLSS 2 | DLSS | No |
| 1.4a | DisplayPort | 1.4 |
| Nvidia | Vendor | Nvidia |
| Discrete | Build | Discrete |
| October 22, 2020 | Released | April 12, 2021 |
| Desktop | Case | Laptop |
| Gaming | Purpose | Professional |
| Mid-range | Segment | Entry-level |
| Ampere | Architecture | Turing |
| GA104 | GPU Codename | TU117 |
| -Intel Core i7 12700Kor above | Recommended CPU | -Intel Core i7 11800Hor above |
| 14153 | Steel Nomad Lite Score | — |
| 13643 | Time Spy | — |
| 64426 | Solar Bay | — |
| 8311 | Port Royal | — |
| 34054 | Fire Strike | — |
| 28486 | Wild Life Extreme | — |
| 122039 | Night Raid | — |
| 1500 MHz | Base Clock | 930 MHz |
| 1725 MHz | Boost Clock | 1395 MHz |
| 5888 | Shading Units | 896 |
| 184 | Texture Mapping Units (TMUs) | 56 |
| 96 | Render Output Units (ROPs) | 32 |
| 46 | Compute Units (Pipelines) | 14 |
| 184 | Tensor Cores | — |
| 46 | Ray-tracing Cores | — |
| 128KB per cluster | L1 Cache | 64KB per cluster |
| 4MB shared | L2 Cache | 1MB shared |
| 2 IPC | Instructions Per Cycle | 2 IPC |