Quick/Comparisons
The Nvidia RTX Pro 6000 Blackwell's exceptional gaming performance makes it the clear winner, with a 53-point gap in scores. Its 99 out of 100 gaming score is unmatched by the GeForce RTX 3070 Ti's 46.
The Nvidia RTX Pro 6000 Blackwell takes the win in content creation with a perfect 100 score, thanks to its strong workstation performance and robust feature set. The GeForce RTX 3070 Ti trails behind with a 49.
The Nvidia RTX Pro 6000 Blackwell leads in AI/ML compute with a 54-point gap in scores, thanks to its capable AI tensor cores and strong performance. The GeForce RTX 3070 Ti's 39 score is respectable but falls short.
46 | Gaming | 99 |
49 | Workstation | 100 |
39 | AI/ML |
Compare with something else?
Compare GeForce RTX 3070 with another →93 |
35 | Energy Efficiency | 49 |
46 | Quick Comparison Final Score | 98 |
| GA104 | Graphics Processor | GB202-300-A1 |
| 6144 | Cores | 24064 |
| 192 / 96 | TMUs / ROPs | 752 / 190 |
| 3746.83 | Blender GPU | 16644.8 |
| 104 | Forza Horizon 5 | 230 |
| 143 | The Witcher 3 | 304 |
| 198 | Counter-Strike 2 | 259 |
| 125 | Far Cry 6 | 198 |
| 85 | Hogwarts Legacy | 192 |
| 120 | Call of Duty | 301 |
| 81 | Ghost of Tsushima | 172 |
| 102 | Cyberpunk 2077 | 221 |
| 161 | Tomb Raider | 303 |
| 124 | Average FPS | 242 |
| 150 | 1080p High | 282 |
| 124 | 1080p Ultra | 242 |
| 93 | 1440p Ultra | 197 |
| 54 | 4K Ultra | 118 |
| Low | Margin of Error | High |
| 136096 | GB6 Compute Score | 419470 |
| 197 img/sec | Background Blur | 366.1 img/sec |
| 128.1 img/sec | Face Detection | 248 img/sec |
| 6.56 Gpixels/sec | Horizon Detection | 23.3 Gpixels/sec |
| 8.81 Gpixels/sec | Edge Detection | 35.2 Gpixels/sec |
| 6.47 Gpixels/sec | Gaussian Blur | 45 Gpixels/sec |
| 1.26 Gpixels/sec | Feature Matching | 2.47 Gpixels/sec |
| 609.3 Gpixels/sec | Stereo Matching | 3040 Gpixels/sec |
| 18205 FPS | Particle Physics | 42574.6 FPS |
| OpenCL | API | OpenCL |
| 23260 | G3D Mark Score | 37305 |
| 1061 | G2D Mark | 1375 |
| 192 FPS | DirectX 11 | 338 FPS |
| 91 FPS | DirectX 12 | 153 FPS |
| 11452 Ops/s | GPU Compute | 26895 Ops/s |
| 22516 | GB6 ML Single Precision | 50053 |
| 36377 | GB6 ML Half Precision | 74036 |
| 14215 | GB6 ML Quantized | 37636 |
| 8615 | Image Classification (SP) | 16903 |
| 24659 | Image Segmentation (HP) | 48518 |
| 20631 | Image Super Resolution (Q) | 54618 |
| 43834 | Face Detection (HP) | 103224 |
| 98347 | Pose Estimation (Q) | 338587 |
| 3162 | Text Classification (SP) | 3894 |
| 4911 | Machine Translation (HP) | 7291 |
| 11498 | Object Detection (SP) | 26498 |
| 34719 | Depth Estimation (Q) | 90853 |
| 196833 | Style Transfer (SP) | 635289 |
| ONNX | Framework | ONNX |
| DirectML | Backend | DirectML |
| 170 GPixel/s | Pixel Fill Rate | 497 GPixel/s |
| 340 GTexel/s | Texture Fill Rate | 1968 GTexel/s |
| 21.7 TFLOPS | FLOPS (FP32) | 126 TFLOPS |
| GDDR6X | Memory Type | GDDR7 |
| 8 GB | Memory Size | 96 GB |
| 1188 MHz | Memory Clock | 1750 MHz |
| 19000 Mbps | Effective Memory Speed | 28000 Mbps |
| 256-bit | Bus | 512-bit |
| No | ECC | Yes |
| 608.3 GB/s | Memory Bandwidth | 1792 GB/s |
| PCIe 4.0 x16 | Interface | PCIe 5.0 x16 |
| 290 W | TGP | 600 W |
| Samsung | Manufacturing | TSMC |
| 8 nm | Fabrication Process | 5 nm |
| 392 mm² | Die Size | 750 mm² |
| 17 billion | Transistor Count | 92.2 billion |
| 43.37 MTr/mm² | Transistor Density | 122.93 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 |
| June 10, 2021 | Released | March 18, 2026 |
| Desktop | Case | Desktop |
| Gaming | Purpose | Professional |
| Mid-range | Segment | High-end |
| Ampere | Architecture | Blackwell 2.0 |
| GA104 | GPU Codename | GB202-300-A1 |
| -Intel Core i7 12700Kor above | Recommended CPU | -AMD Ryzen 9 9955HX3Dor above |
| 15402 | Steel Nomad Lite Score | 50516 |
| 14880 | Time Spy | 42437 |
| 69900 | Solar Bay | 258808 |
| 8884 | Port Royal | 39952 |
| 37238 | Fire Strike | 72183 |
| 31911 | Wild Life Extreme | 110070 |
| 134012 | Night Raid | 216213 |
| 1575 MHz | Base Clock | 1590 MHz |
| 1770 MHz | Boost Clock | 2617 MHz |
| 6144 | Shading Units | 24064 |
| 192 | Texture Mapping Units (TMUs) | 752 |
| 96 | Render Output Units (ROPs) | 190 |
| 48 | Compute Units (Pipelines) | 188 |
| 192 | Tensor Cores | 752 |
| 48 | Ray-tracing Cores | 188 |
| 128KB per cluster | L1 Cache | 128KB per cluster |
| 4MB shared | L2 Cache | 128MB shared |
| 2 IPC | Instructions Per Cycle | 2 IPC |