Quick/Comparisons
34 | Gaming | 41 |
41 | Workstation | 40 |
31 | AI/ML |
Compare with something else?
Compare GeForce RTX 3070 with another →38 |
43 | Energy Efficiency | 56 |
37 | Quick Comparison Final Score | 43 |
| GA104 (GN20-E5) | Graphics Processor | AD106 (GN21-X6) |
| 5120 | Cores | 4608 |
| 160 / 80 | TMUs / ROPs | 144 / 48 |
| 2998.92 | Blender GPU | 3587.08 |
| 79 | Forza Horizon 5 | 90 |
| 114 | The Witcher 3 | 132 |
| 150 | Counter-Strike 2 | 178 |
| 94 | Far Cry 6 | 112 |
| 61 | Hogwarts Legacy | 75 |
| 88 | Call of Duty | 104 |
| 56 | Ghost of Tsushima | 68 |
| 73 | Cyberpunk 2077 | 87 |
| 116 | Tomb Raider | 139 |
| 92 | Average FPS | 109 |
| 114 | 1080p High | 131 |
| 92 | 1080p Ultra | 109 |
| 68 | 1440p Ultra | 81 |
| 38 | 4K Ultra | 46 |
| Low | Margin of Error | Low |
| 118499 | GB6 Compute Score | 115081 |
| 170.3 img/sec | Background Blur | 158.5 img/sec |
| 113.6 img/sec | Face Detection | 116.5 img/sec |
| 5.59 Gpixels/sec | Horizon Detection | 4.35 Gpixels/sec |
| 7.94 Gpixels/sec | Edge Detection | 5.67 Gpixels/sec |
| 6.19 Gpixels/sec | Gaussian Blur | 6.44 Gpixels/sec |
| 1.16 Gpixels/sec | Feature Matching | 1.36 Gpixels/sec |
| 467 Gpixels/sec | Stereo Matching | 528.5 Gpixels/sec |
| 15131.3 FPS | Particle Physics | 16372.3 FPS |
| OpenCL | API | OpenCL |
| 15105 | G3D Mark Score | 19527 |
| 630 | G2D Mark | 767 |
| 137 FPS | DirectX 11 | 178 FPS |
| 63 FPS | DirectX 12 | 85 FPS |
| 6702 Ops/s | GPU Compute | 8389 Ops/s |
| 17250 | GB6 ML Single Precision | 21186 |
| 27936 | GB6 ML Half Precision | 34404 |
| 11249 | GB6 ML Quantized | 16559 |
| 6969 | Image Classification (SP) | 8642 |
| 17142 | Image Segmentation (HP) | 20635 |
| 16426 | Image Super Resolution (Q) | 26378 |
| 31628 | Face Detection (HP) | 39776 |
| 74413 | Pose Estimation (Q) | 105983 |
| 2522 | Text Classification (SP) | 2995 |
| 3639 | Machine Translation (HP) | 4507 |
| 9002 | Object Detection (SP) | 10198 |
| 27790 | Depth Estimation (Q) | 41078 |
| 160005 | Style Transfer (SP) | 190315 |
| ONNX | Framework | ONNX |
| DirectML | Backend | DirectML |
| 103-130 GPixel/s | Pixel Fill Rate | 59-104 GPixel/s |
| 206-259 GTexel/s | Texture Fill Rate | 177-313 GTexel/s |
| 13.2-16.6 TFLOPS | FLOPS (FP32) | 11.3-20 TFLOPS |
| GDDR6 | Memory Type | GDDR6 |
| 8 GB | Memory Size | 8 GB |
| 1500-1750 MHz | Memory Clock | 2000 MHz |
| 12000-14000 Mbps | Effective Memory Speed | 16000 Mbps |
| 256-bit | Bus | 128-bit |
| No | ECC | No |
| 448 GB/s | Memory Bandwidth | 256 GB/s |
| PCIe 4.0 x16 | Interface | PCIe 4.0 x16 |
| 80-150 W (configurable) | TGP | 35-140 W (configurable) |
| Samsung | Manufacturing | TSMC |
| 8 nm | Fabrication Process | 5 nm |
| 392 mm² | Die Size | 186 mm² |
| 17.4 billion | Transistor Count | 22.9 billion |
| 44.39 MTr/mm² | Transistor Density | 123.12 MTr/mm² |
| 12 | DirectX | 12 |
| 1.2 | Vulkan | 1.3 |
| 4.6 | OpenGL | 4.6 |
| 2 | OpenCL | 3 |
| 8.6 | CUDA | 8.9 |
| Yes | Ray Tracing | Yes |
| DLSS 2 | DLSS | DLSS 3 |
| 1.4a | DisplayPort | 1.4a |
| Nvidia | Vendor | Nvidia |
| Discrete | Build | Discrete |
| January 12, 2021 | Released | February 22, 2023 |
| Laptop | Case | Laptop |
| Gaming | Purpose | Gaming |
| High-end | Segment | High-end |
| Ampere | Architecture | Ada Lovelace |
| GA104 (GN20-E5) | GPU Codename | AD106 (GN21-X6) |
| -Intel Core i7 12700Hor above | Recommended CPU | -Intel Core Ultra 7 155Hor above |
| 10214 | Steel Nomad Lite Score | 12729 |
| 10477 | Time Spy | 12518 |
| 45215 | Solar Bay | 54826 |
| 6247 | Port Royal | 7403 |
| 26729 | Fire Strike | 30044 |
| 21137 | Wild Life Extreme | 23244 |
| 82533 | Night Raid | 113720 |
| 780-1215 MHz (configurable) | Base Clock | 735-2070 MHz (configurable) |
| 1290-1620 MHz (configurable) | Boost Clock | 1230-2175 MHz (configurable) |
| 5120 | Shading Units | 4608 |
| 160 | Texture Mapping Units (TMUs) | 144 |
| 80 | Render Output Units (ROPs) | 48 |
| 40 | Compute Units (Pipelines) | 36 |
| 160 | Tensor Cores | 144 |
| 40 | Ray-tracing Cores | 36 |
| 128KB per cluster | L1 Cache | 128KB per cluster |
| 4MB shared | L2 Cache | 32MB shared |
| 2 IPC | Instructions Per Cycle | 2 IPC |