Quick/Comparisons
The GeForce RTX 5070 Ti's high gaming score of 76 is significantly better than the Radeon 660M's score of 7, making it the clear winner for gaming use cases.
The GeForce RTX 5070 Ti's workstation score of 74 is notably higher than the Radeon 660M's score of 13, making it the better choice for content creation tasks.
The GeForce RTX 5070 Ti's AI/ML score of 66 is substantially higher than the Radeon 660M's score of 7, making it the preferred option for AI and machine learning applications.
76 | Gaming | 7 |
74 | Workstation | 13 |
66 | AI/ML | 7 |
49 | Energy Efficiency | 88 |
74 | Quick Comparison Final Score | 22 |
| GB203-200-A1 | Graphics Processor | Rembrandt |
| 8960 | Cores | 384 |
| 280 / 96 | TMUs / ROPs | 24 / 16 |
| 7584.67 | Blender GPU | 92.56 |
| 166 | Forza Horizon 5 | 12 |
| 213 | The Witcher 3 | 10 |
| 219 | Counter-Strike 2 | 20 |
| 159 | Far Cry 6 | 19 |
| 135 | Hogwarts Legacy | - |
| 205 | Call of Duty | 15 |
| 121 | Ghost of Tsushima | 9 |
| 174 | Cyberpunk 2077 | 9 |
| 260 | Tomb Raider | 13 |
| 184 | Average FPS | 13 |
| 217 | 1080p High | 17 |
| 184 | 1080p Ultra | 13 |
| 148 | 1440p Ultra | 9 |
| 86 | 4K Ultra | - |
| Low | Margin of Error | Low |
| 240508 | GB6 Compute Score | 14065 |
| 297.9 img/sec | Background Blur | 36 img/sec |
| 218.6 img/sec | Face Detection | 16.4 img/sec |
| 10.1 Gpixels/sec | Horizon Detection | 0.48 Gpixels/sec |
| 13.9 Gpixels/sec | Edge Detection | 0.65 Gpixels/sec |
| 14.8 Gpixels/sec | Gaussian Blur | 0.53 Gpixels/sec |
| 2.03 Gpixels/sec | Feature Matching | 0.18 Gpixels/sec |
| 1290 Gpixels/sec | Stereo Matching | 46.5 Gpixels/sec |
| 35438.3 FPS | Particle Physics | 2056.5 FPS |
| OpenCL | API | OpenCL |
| 32425 | G3D Mark Score | 3207 |
| 1329 | G2D Mark | 645 |
| 304 FPS | DirectX 11 | 54 FPS |
| 126 FPS | DirectX 12 | 27 FPS |
| 18089 Ops/s | GPU Compute | 1418 Ops/s |
| 36504 | GB6 ML Single Precision | 3903 |
| 64459 | GB6 ML Half Precision | 4745 |
| 27312 | GB6 ML Quantized | 2977 |
| 14913 | Image Classification (SP) | 1465 |
| 45199 | Image Segmentation (HP) | 1706 |
| 39360 | Image Super Resolution (Q) | 4266 |
| 84188 | Face Detection (HP) | 5115 |
| 161400 | Pose Estimation (Q) | 12992 |
| 4010 | Text Classification (SP) | 1116 |
| 7973 | Machine Translation (HP) | 1859 |
| 18934 | Object Detection (SP) | 1574 |
| 67761 | Depth Estimation (Q) | 6538 |
| 335879 | Style Transfer (SP) | 26656 |
| ONNX | Framework | ONNX |
| DirectML | Backend | DirectML |
| 235 GPixel/s | Pixel Fill Rate | 29-30 GPixel/s |
| 687 GTexel/s | Texture Fill Rate | 43-46 GTexel/s |
| 43.9 TFLOPS | FLOPS (FP32) | 1.4-1.5 TFLOPS |
| GDDR7 | Memory Type | System Shared |
| 16 GB | Memory Size | — |
| 1750 MHz | Memory Clock | — |
| 28000 Mbps | Effective Memory Speed | — |
| 256-bit | Bus | — |
| No | ECC | No |
| 896 GB/s | Memory Bandwidth | — |
| PCIe 5.0 x16 | Interface | PCIe 4.0 x8 |
| 300 W | TGP | 15 W |
| TSMC | Manufacturing | TSMC |
| 4 nm | Fabrication Process | 6 nm |
| 378 mm² | Die Size | 208 mm² |
| 45.6 billion | Transistor Count | 13.1 billion |
| 120.63 MTr/mm² | Transistor Density | 62.98 MTr/mm² |
| 12.2 | DirectX | 12 |
| 1.4 | Vulkan | 1.2 |
| 4.6 | OpenGL | 4.6 |
| 3 | OpenCL | 2 |
| 12 | CUDA | — |
| Yes | Ray Tracing | Yes |
| DLSS 4 | DLSS | No |
| 2.1b | DisplayPort | — |
| Nvidia | Vendor | Amd |
| Discrete | Build | Integrated |
| January 7, 2025 | Released | January 1, 2022 |
| Desktop | Case | Laptop |
| Gaming | Purpose | Generic |
| Mid-range | Segment | Entry-level |
| Blackwell 2.0 | Architecture | RDNA 2.0 |
| GB203-200-A1 | GPU Codename | Rembrandt |
| -Intel Core Ultra 5 245Kor above | Recommended CPU | -AMD Ryzen 5 6600Hor above |
| 29989 | Steel Nomad Lite Score | 1344 |
| 27815 | Time Spy | 1515 |
| 136964 | Solar Bay | 6396 |
| 19613 | Port Royal | 586 |
| 67382 | Fire Strike | 4870 |
| 57734 | Wild Life Extreme | 2676 |
| 189734 | Night Raid | 19259 |
| 2295 MHz | Base Clock | 400 MHz |
| 2452 MHz | Boost Clock | 1800-1900 MHz (configurable) |
| 8960 | Shading Units | 384 |
| 280 | Texture Mapping Units (TMUs) | 24 |
| 96 | Render Output Units (ROPs) | 16 |
| 70 | Compute Units (Pipelines) | 6 |
| 280 | Tensor Cores | — |
| 70 | Ray-tracing Cores | 6 |
| 128KB per cluster | L1 Cache | 128KB per cluster |
| 48MB shared | L2 Cache | 1024KB shared |
| 2 IPC | Instructions Per Cycle | 2 IPC |
Compare with something else?
Compare GeForce RTX 5070 with another →