Quick/Comparisons
The GeForce RTX 5070 Ti takes the lead in gaming performance, thanks to its higher score of 76, outpacing the GeForce RTX 3090 Ti's score of 66.
Both the GeForce RTX 3090 Ti and GeForce RTX 5070 Ti deliver solid workstation performance, with the GeForce RTX 3090 Ti scoring 73 and the GeForce RTX 5070 Ti scoring 74, resulting in a tie.
The GeForce RTX 5070 Ti excels in AI/ML tasks, boasting a higher score of 66 compared to the GeForce RTX 3090 Ti's score of 52, making it the better choice for AI-intensive workloads.
66 | Gaming | 76 |
73 | Workstation | 74 |
52 | AI/ML | 66 |
37 | Energy Efficiency | 49 |
66 | Quick Comparison Final Score | 74 |
| GA102 | Graphics Processor | GB203-200-A1 |
| 10752 | Cores | 8960 |
| 336 / 112 | TMUs / ROPs | 280 / 96 |
| 6057.83 | Blender GPU | 7584.67 |
| 141 | Forza Horizon 5 | 166 |
| 182 | The Witcher 3 | 213 |
| 210 | Counter-Strike 2 | 219 |
| 147 | Far Cry 6 | 159 |
| 112 | Hogwarts Legacy | 135 |
| 173 | Call of Duty | 205 |
| 108 | Ghost of Tsushima | 121 |
| 143 | Cyberpunk 2077 | 174 |
| 243 | Tomb Raider | 260 |
| 162 | Average FPS | 184 |
| 128 | 1440p Ultra | 148 |
| 75 | 4K Ultra | 86 |
| Low | Margin of Error | Low |
| 193 | 1080p High | 217 |
| 162 | 1080p Ultra | 184 |
| 232083 | GB6 Compute Score | 240508 |
| 362.7 img/sec | Background Blur | 297.9 img/sec |
| 227.9 img/sec | Face Detection | 218.6 img/sec |
| 11.3 Gpixels/sec | Horizon Detection | 10.1 Gpixels/sec |
| 15.9 Gpixels/sec | Edge Detection | 13.9 Gpixels/sec |
| 12.1 Gpixels/sec | Gaussian Blur | 14.8 Gpixels/sec |
| 1.56 Gpixels/sec | Feature Matching | 2.03 Gpixels/sec |
| 1080 Gpixels/sec | Stereo Matching | 1290 Gpixels/sec |
| 31024.6 FPS | Particle Physics | 35438.3 FPS |
| OpenCL | API | OpenCL |
| 29288 | G3D Mark Score | 32425 |
| 1226 | G2D Mark | 1329 |
| 238 FPS | DirectX 11 | 304 FPS |
| 122 FPS | DirectX 12 | 126 FPS |
| 16569 Ops/s | GPU Compute | 18089 Ops/s |
| 26701 | GB6 ML Single Precision | 36504 |
| 38573 | GB6 ML Half Precision | 64459 |
| 15995 | GB6 ML Quantized | 27312 |
| 10246 | Image Classification (SP) | 14913 |
| 22830 | Image Segmentation (HP) | 45199 |
| 23532 | Image Super Resolution (Q) | 39360 |
| 48657 | Face Detection (HP) | 84188 |
| 147241 | Pose Estimation (Q) | 161400 |
| 2917 | Text Classification (SP) | 4010 |
| 4427 | Machine Translation (HP) | 7973 |
| 13323 | Object Detection (SP) | 18934 |
| 40055 | Depth Estimation (Q) | 67761 |
| 298683 | Style Transfer (SP) | 335879 |
| ONNX | Framework | ONNX |
| DirectML | Backend | DirectML |
| 208 GPixel/s | Pixel Fill Rate | 235 GPixel/s |
| 625 GTexel/s | Texture Fill Rate | 687 GTexel/s |
| 40 TFLOPS | FLOPS (FP32) | 43.9 TFLOPS |
| GDDR6X | Memory Type | GDDR7 |
| 24 GB | Memory Size | 16 GB |
| 1313 MHz | Memory Clock | 1750 MHz |
| 21000 Mbps | Effective Memory Speed | 28000 Mbps |
| 384-bit | Bus | 256-bit |
| No | ECC | No |
| 1008 GB/s | Memory Bandwidth | 896 GB/s |
| PCIe 4.0 x16 | Interface | PCIe 5.0 x16 |
| 450 W | TGP | 300 W |
| Samsung | Manufacturing | TSMC |
| 8 nm | Fabrication Process | 4 nm |
| 628 mm² | Die Size | 378 mm² |
| 28 billion | Transistor Count | 45.6 billion |
| 44.59 MTr/mm² | Transistor Density | 120.63 MTr/mm² |
| 12 | DirectX | 12.2 |
| 1.3 | Vulkan | 1.4 |
| 4.6 | OpenGL | 4.6 |
| 3 | OpenCL | 3 |
| 8.6 | CUDA | 12 |
| Yes | Ray Tracing | Yes |
| DLSS 2 | DLSS | DLSS 4 |
| 1.4a | DisplayPort | 2.1b |
| Nvidia | Vendor | Nvidia |
| Discrete | Build | Discrete |
| March 29, 2022 | Released | January 7, 2025 |
| Desktop | Case | Desktop |
| Gaming | Purpose | Gaming |
| High-end | Segment | Mid-range |
| Ampere | Architecture | Blackwell 2.0 |
| GA102 | GPU Codename | GB203-200-A1 |
| -Intel Core i9 12900Kor above | Recommended CPU | -Intel Core Ultra 5 245Kor above |
| 24378 | Steel Nomad Lite Score | 29989 |
| 21854 | Time Spy | 27815 |
| 109167 | Solar Bay | 136964 |
| 14888 | Port Royal | 19613 |
| 51943 | Fire Strike | 67382 |
| 48864 | Wild Life Extreme | 57734 |
| 156500 | Night Raid | 189734 |
| 1560 MHz | Base Clock | 2295 MHz |
| 1860 MHz | Boost Clock | 2452 MHz |
| 10752 | Shading Units | 8960 |
| 336 | Texture Mapping Units (TMUs) | 280 |
| 112 | Render Output Units (ROPs) | 96 |
| 84 | Compute Units (Pipelines) | 70 |
| 336 | Tensor Cores | 280 |
| 84 | Ray-tracing Cores | 70 |
| 128KB per cluster | L1 Cache | 128KB per cluster |
| 6MB shared | L2 Cache | 48MB shared |
| 2 IPC | Instructions Per Cycle | 2 IPC |
Compare with something else?
Compare GeForce RTX 3090 with another →