Quick/Comparisons
46 | Gaming | 35 |
49 | Workstation | 43 |
39 | AI/ML |
Compare with something else?
Compare GeForce RTX 3070 with another →31 |
35 | Energy Efficiency | 41 |
46 | Quick Comparison Final Score | 37 |
| GA104 | Graphics Processor | DG2-512 |
| 6144 | Cores | 4096 |
| 192 / 96 | TMUs / ROPs | 256 / 128 |
| 3746.83 | Blender GPU | 2016.55 |
| 104 | Forza Horizon 5 | 77 |
| 143 | The Witcher 3 | 124 |
| 198 | Counter-Strike 2 | 118 |
| 125 | Far Cry 6 | 97 |
| 85 | Hogwarts Legacy | 59 |
| 120 | Call of Duty | 79 |
| 81 | Ghost of Tsushima | 57 |
| 102 | Cyberpunk 2077 | 80 |
| 161 | Tomb Raider | 106 |
| 124 | Average FPS | 89 |
| 150 | 1080p High | 106 |
| 124 | 1080p Ultra | 89 |
| 93 | 1440p Ultra | 65 |
| 54 | 4K Ultra | 37 |
| Low | Margin of Error | Low |
| 136096 | GB6 Compute Score | 113935 |
| 197 img/sec | Background Blur | 175 img/sec |
| 128.1 img/sec | Face Detection | 98 img/sec |
| 6.56 Gpixels/sec | Horizon Detection | 3.63 Gpixels/sec |
| 8.81 Gpixels/sec | Edge Detection | 5.51 Gpixels/sec |
| 6.47 Gpixels/sec | Gaussian Blur | 8.51 Gpixels/sec |
| 1.26 Gpixels/sec | Feature Matching | 0.89 Gpixels/sec |
| 609.3 Gpixels/sec | Stereo Matching | 492.6 Gpixels/sec |
| 18205 FPS | Particle Physics | 24819.2 FPS |
| OpenCL | API | OpenCL |
| 23260 | G3D Mark Score | 13358 |
| 1061 | G2D Mark | 761 |
| 192 FPS | DirectX 11 | 78 FPS |
| 91 FPS | DirectX 12 | 75 FPS |
| 11452 Ops/s | GPU Compute | 5850 Ops/s |
| 22516 | GB6 ML Single Precision | 13344 |
| 36377 | GB6 ML Half Precision | 14119 |
| 14215 | GB6 ML Quantized | 10082 |
| 8615 | Image Classification (SP) | 5750 |
| 24659 | Image Segmentation (HP) | 16175 |
| 20631 | Image Super Resolution (Q) | 12997 |
| 43834 | Face Detection (HP) | 21409 |
| 98347 | Pose Estimation (Q) | 67820 |
| 3162 | Text Classification (SP) | 849 |
| 4911 | Machine Translation (HP) | 2627 |
| 11498 | Object Detection (SP) | 7465 |
| 34719 | Depth Estimation (Q) | 24186 |
| 196833 | Style Transfer (SP) | 183238 |
| ONNX | Framework | ONNX |
| DirectML | Backend | DirectML |
| 170 GPixel/s | Pixel Fill Rate | 307 GPixel/s |
| 340 GTexel/s | Texture Fill Rate | 614 GTexel/s |
| 21.7 TFLOPS | FLOPS (FP32) | 19.7 TFLOPS |
| GDDR6X | Memory Type | GDDR6 |
| 8 GB | Memory Size | 16 GB |
| 1188 MHz | Memory Clock | 2000 MHz |
| 19000 Mbps | Effective Memory Speed | 16000 Mbps |
| 256-bit | Bus | 256-bit |
| No | ECC | No |
| 608.3 GB/s | Memory Bandwidth | 512 GB/s |
| PCIe 4.0 x16 | Interface | PCIe 4.0 x16 |
| 290 W | TGP | 225 W |
| Samsung | Manufacturing | TSMC |
| 8 nm | Fabrication Process | 6 nm |
| 392 mm² | Die Size | 406 mm² |
| 17 billion | Transistor Count | 21 billion |
| 43.37 MTr/mm² | Transistor Density | 51.72 MTr/mm² |
| 12 | DirectX | 12 |
| 1.3 | Vulkan | 1.3 |
| 4.6 | OpenGL | 4.6 |
| 3 | OpenCL | 3 |
| 8.6 | CUDA | — |
| Yes | Ray Tracing | Yes |
| DLSS 2 | DLSS | No |
| 1.4a | DisplayPort | — |
| Nvidia | Vendor | Intel |
| Discrete | Build | Discrete |
| June 10, 2021 | Released | October 12, 2022 |
| Desktop | Case | Desktop |
| Gaming | Purpose | Gaming |
| Mid-range | Segment | Mid-range |
| Ampere | Architecture | Generation 12.7 |
| GA104 | GPU Codename | DG2-512 |
| -Intel Core i7 12700Kor above | Recommended CPU | -Intel Core i5 12500or above |
| 15402 | Steel Nomad Lite Score | 12648 |
| 14880 | Time Spy | 13554 |
| 69900 | Solar Bay | 51114 |
| 8884 | Port Royal | 7231 |
| 37238 | Fire Strike | 34163 |
| 31911 | Wild Life Extreme | 26266 |
| 134012 | Night Raid | 122588 |
| 1575 MHz | Base Clock | 2100 MHz |
| 1770 MHz | Boost Clock | 2400 MHz |
| 6144 | Shading Units | 4096 |
| 192 | Texture Mapping Units (TMUs) | 256 |
| 96 | Render Output Units (ROPs) | 128 |
| 48 | Compute Units (Pipelines) | — |
| 192 | Tensor Cores | 512 |
| 48 | Ray-tracing Cores | 32 |
| 128KB per cluster | L1 Cache | — |
| 4MB shared | L2 Cache | 16MB shared |
| 2 IPC | Instructions Per Cycle | 2 IPC |