Quick/Comparisons
Both GPUs deliver identical gaming performance, making them tied in this use case.
The GeForce RTX 3070 and GeForce RTX 4060 Ti 8GB share the same workstation score, resulting in a tie.
The GeForce RTX 4060 Ti 8GB outperforms the GeForce RTX 3070 in AI/ML tasks by 4 points, thanks to its higher AI/ML score of 39.
43 | Gaming | 43 |
44 | Workstation | 44 |
35 | AI/ML |
Compare with something else?
Compare GeForce RTX 3070 with another →39 |
35 | Energy Efficiency | 50 |
42 | Quick Comparison Final Score | 43 |
| GA104 | Graphics Processor | AD106 |
| 5888 | Cores | 4352 |
| 184 / 96 | TMUs / ROPs | 136 / 48 |
| 3368.42 | Blender GPU | — |
| 99 | Forza Horizon 5 | 94 |
| 143 | The Witcher 3 | 140 |
| 194 | Counter-Strike 2 | 191 |
| 121 | Far Cry 6 | 114 |
| 80 | Hogwarts Legacy | 77 |
| 113 | Call of Duty | 112 |
| 74 | Ghost of Tsushima | 73 |
| 94 | Cyberpunk 2077 | 92 |
| 153 | Tomb Raider | 151 |
| 119 | Average FPS | 116 |
| 141 | 1080p High | 139 |
| 119 | 1080p Ultra | 116 |
| 88 | 1440p Ultra | 86 |
| 50 | 4K Ultra | 50 |
| Low | Margin of Error | Low |
| 126195 | GB6 Compute Score | 127835 |
| 161.6 img/sec | Background Blur | 198.3 img/sec |
| 117.2 img/sec | Face Detection | 126.2 img/sec |
| 5.89 Gpixels/sec | Horizon Detection | 4.35 Gpixels/sec |
| 8.05 Gpixels/sec | Edge Detection | 5.69 Gpixels/sec |
| 6.65 Gpixels/sec | Gaussian Blur | 6.72 Gpixels/sec |
| 1.25 Gpixels/sec | Feature Matching | 1.53 Gpixels/sec |
| 550.6 Gpixels/sec | Stereo Matching | 656.3 Gpixels/sec |
| 17613.1 FPS | Particle Physics | 19149.2 FPS |
| OpenCL | API | OpenCL |
| 22128 | G3D Mark Score | 22618 |
| 1005 | G2D Mark | 1096 |
| 182 FPS | DirectX 11 | 202 FPS |
| 84 FPS | DirectX 12 | 86 FPS |
| 11062 Ops/s | GPU Compute | 12047 Ops/s |
| 19789 | GB6 ML Single Precision | 22080 |
| 33013 | GB6 ML Half Precision | 36358 |
| 12443 | GB6 ML Quantized | 17354 |
| 7624 | Image Classification (SP) | 9359 |
| 23420 | Image Segmentation (HP) | 21463 |
| 17898 | Image Super Resolution (Q) | 26023 |
| 40535 | Face Detection (HP) | 42533 |
| 83624 | Pose Estimation (Q) | 99698 |
| 2724 | Text Classification (SP) | 3181 |
| 4481 | Machine Translation (HP) | 4956 |
| 10304 | Object Detection (SP) | 11009 |
| 30418 | Depth Estimation (Q) | 43518 |
| 180089 | Style Transfer (SP) | 203667 |
| ONNX | Framework | ONNX |
| DirectML | Backend | DirectML |
| 166 GPixel/s | Pixel Fill Rate | 122 GPixel/s |
| 317 GTexel/s | Texture Fill Rate | 345 GTexel/s |
| 20.3 TFLOPS | FLOPS (FP32) | 22.1 TFLOPS |
| GDDR6 | Memory Type | GDDR6 |
| 8 GB | Memory Size | 8 GB |
| 1750 MHz | Memory Clock | 2250 MHz |
| 14000 Mbps | Effective Memory Speed | 18000 Mbps |
| 256-bit | Bus | 128-bit |
| No | ECC | No |
| 448 GB/s | Memory Bandwidth | 288 GB/s |
| PCIe 4.0 x16 | Interface | PCIe 4.0 x8 |
| 220 W | TGP | 160 W |
| Samsung | Manufacturing | TSMC |
| 8 nm | Fabrication Process | 5 nm |
| 392 mm² | Die Size | 188 mm² |
| 17 billion | Transistor Count | 22 billion |
| 43.37 MTr/mm² | Transistor Density | 117.02 MTr/mm² |
| 12 | DirectX | 12 |
| 1.3 | Vulkan | 1.3 |
| 4.6 | OpenGL | 4.6 |
| 3 | 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 |
| October 22, 2020 | Released | May 18, 2023 |
| Desktop | Case | Desktop |
| Gaming | Purpose | Gaming |
| Mid-range | Segment | Mid-range |
| Ampere | Architecture | Ada Lovelace |
| GA104 | GPU Codename | AD106 |
| -Intel Core i7 12700Kor above | Recommended CPU | -Intel Core i5 14600Kor above |
| 14153 | Steel Nomad Lite Score | 13592 |
| 13643 | Time Spy | 13484 |
| 64426 | Solar Bay | 62383 |
| 8311 | Port Royal | 8103 |
| 34054 | Fire Strike | 34407 |
| 28486 | Wild Life Extreme | 25596 |
| 122039 | Night Raid | 136385 |
| 1500 MHz | Base Clock | 2310 MHz |
| 1725 MHz | Boost Clock | 2535 MHz |
| 5888 | Shading Units | 4352 |
| 184 | Texture Mapping Units (TMUs) | 136 |
| 96 | Render Output Units (ROPs) | 48 |
| 46 | Compute Units (Pipelines) | 34 |
| 184 | Tensor Cores | 136 |
| 46 | Ray-tracing Cores | 34 |
| 128KB per cluster | L1 Cache | 128KB per cluster |
| 4MB shared | L2 Cache | 32MB shared |
| 2 IPC | Instructions Per Cycle | 2 IPC |