Quick/Comparisons
46 | Gaming | 16 |
49 | Workstation | 10 |
39 | AI/ML |
Compare with something else?
Compare GeForce RTX 3070 with another →7 |
35 | Energy Efficiency | 94 |
46 | Quick Comparison Final Score | 26 |
| GA104 | Graphics Processor | — |
| 6144 | Cores | 512 |
| 192 / 96 | TMUs / ROPs | 32 / 16 |
| 3746.83 | Blender GPU | — |
| 104 | Forza Horizon 5 | — |
| 143 | The Witcher 3 | — |
| 198 | Counter-Strike 2 | — |
| 125 | Far Cry 6 | — |
| 85 | Hogwarts Legacy | — |
| 120 | Call of Duty | — |
| 81 | Ghost of Tsushima | — |
| 102 | Cyberpunk 2077 | — |
| 161 | Tomb Raider | — |
| 124 | Average FPS | — |
| 150 | 1080p High | — |
| 124 | 1080p Ultra | — |
| 93 | 1440p Ultra | — |
| 54 | 4K Ultra | — |
| Low | Margin of Error | — |
| 136096 | GB6 Compute Score | 16321 |
| 197 img/sec | Background Blur | 37.6 img/sec |
| 128.1 img/sec | Face Detection | 17.2 img/sec |
| 6.56 Gpixels/sec | Horizon Detection | 0.3 Gpixels/sec |
| 8.81 Gpixels/sec | Edge Detection | 0.63 Gpixels/sec |
| 6.47 Gpixels/sec | Gaussian Blur | 0.94 Gpixels/sec |
| 1.26 Gpixels/sec | Feature Matching | 0.22 Gpixels/sec |
| 609.3 Gpixels/sec | Stereo Matching | 60 Gpixels/sec |
| 18205 FPS | Particle Physics | 3695.9 FPS |
| OpenCL | API | OpenCL |
| 23260 | G3D Mark Score | — |
| 1061 | G2D Mark | — |
| 192 FPS | DirectX 11 | — |
| 91 FPS | DirectX 12 | — |
| 11452 Ops/s | GPU Compute | — |
| 22516 | GB6 ML Single Precision | 3770 |
| 36377 | GB6 ML Half Precision | 5146 |
| 14215 | GB6 ML Quantized | 2824 |
| 8615 | Image Classification (SP) | 2902 |
| 24659 | Image Segmentation (HP) | 5473 |
| 20631 | Image Super Resolution (Q) | 2967 |
| 43834 | Face Detection (HP) | 6068 |
| 98347 | Pose Estimation (Q) | 8220 |
| 3162 | Text Classification (SP) | 609 |
| 4911 | Machine Translation (HP) | 1146 |
| 11498 | Object Detection (SP) | 2766 |
| 34719 | Depth Estimation (Q) | 6405 |
| 196833 | Style Transfer (SP) | 26796 |
| ONNX | Framework | ONNX |
| DirectML | Backend | DirectML |
| 170 GPixel/s | Pixel Fill Rate | 28-32 GPixel/s |
| 340 GTexel/s | Texture Fill Rate | 56-64 GTexel/s |
| 21.7 TFLOPS | FLOPS (FP32) | 1.8-2 TFLOPS |
| GDDR6X | Memory Type | System Shared |
| 8 GB | Memory Size | — |
| 1188 MHz | Memory Clock | — |
| 19000 Mbps | Effective Memory Speed | — |
| 256-bit | Bus | — |
| No | ECC | — |
| 608.3 GB/s | Memory Bandwidth | — |
| PCIe 4.0 x16 | Interface | Ring Bus |
| 290 W | TGP | 14 W |
| Samsung | Manufacturing | TSMC |
| 8 nm | Fabrication Process | 5 nm |
| 392 mm² | Die Size | — |
| 17 billion | Transistor Count | — |
| 43.37 MTr/mm² | Transistor Density | — |
| 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 | 2.1 |
| Nvidia | Vendor | Intel |
| Discrete | Build | Integrated |
| June 10, 2021 | Released | December 14, 2023 |
| Desktop | Case | Laptop |
| Gaming | Purpose | Generic |
| Mid-range | Segment | Mid-range |
| Ampere | Architecture | Xe LPG |
| GA104 | GPU Codename | — |
| -Intel Core i7 12700Kor above | Recommended CPU | -Intel Core Ultra 5 125Uor above |
| 15402 | Steel Nomad Lite Score | — |
| 14880 | Time Spy | — |
| 69900 | Solar Bay | — |
| 8884 | Port Royal | — |
| 37238 | Fire Strike | — |
| 31911 | Wild Life Extreme | — |
| 134012 | Night Raid | — |
| 1575 MHz | Base Clock | 300 MHz |
| 1770 MHz | Boost Clock | 1750-2000 MHz (configurable) |
| 6144 | Shading Units | 512 |
| 192 | Texture Mapping Units (TMUs) | 32 |
| 96 | Render Output Units (ROPs) | 16 |
| 48 | Compute Units (Pipelines) | 64 |
| 192 | Tensor Cores | — |
| 48 | Ray-tracing Cores | 4 |
| 128KB per cluster | L1 Cache | 768KB shared |
| 4MB shared | L2 Cache | — |
| 2 IPC | Instructions Per Cycle | 2 IPC |