Quick/Comparisons
12 | Gaming | 42 |
21 | Workstation | 48 |
11 | AI/ML |
Compare with something else?
Compare GeForce GTX 970 with another →37 |
10 | Energy Efficiency | 37 |
15 | Quick Comparison Final Score | 42 |
| GM204 | Graphics Processor | GA103 (GN20-E8) |
| 1664 | Cores | 7424 |
| 104 / 56 | TMUs / ROPs | 232 / 96 |
| 286.89 | Blender GPU | 3502.46 |
| 32 | Forza Horizon 5 | 92 |
| 35 | The Witcher 3 | 136 |
| 57 | Counter-Strike 2 | 181 |
| 39 | Far Cry 6 | 111 |
| 18 | Hogwarts Legacy | 79 |
| 31 | Call of Duty | 106 |
| 17 | Ghost of Tsushima | 72 |
| 24 | Cyberpunk 2077 | 86 |
| 44 | Tomb Raider | 143 |
| 33 | Average FPS | 112 |
| 44 | 1080p High | 135 |
| 33 | 1080p Ultra | 112 |
| 23 | 1440p Ultra | 83 |
| 12 | 4K Ultra | 47 |
| Low | Margin of Error | Low |
| 28989 | GB6 Compute Score | 133659 |
| 77.1 img/sec | Background Blur | 200.9 img/sec |
| 38.1 img/sec | Face Detection | 131.2 img/sec |
| 1.28 Gpixels/sec | Horizon Detection | 6.49 Gpixels/sec |
| 2.08 Gpixels/sec | Edge Detection | 8.9 Gpixels/sec |
| 1.85 Gpixels/sec | Gaussian Blur | 6.34 Gpixels/sec |
| 0.17 Gpixels/sec | Feature Matching | 1.2 Gpixels/sec |
| 92.9 Gpixels/sec | Stereo Matching | 571 Gpixels/sec |
| 2443 FPS | Particle Physics | 17240.3 FPS |
| OpenCL | API | OpenCL |
| 9639 | G3D Mark Score | 18948 |
| 763 | G2D Mark | 817 |
| 71 FPS | DirectX 11 | 161 FPS |
| 41 FPS | DirectX 12 | 87 FPS |
| 4080 Ops/s | GPU Compute | 8302 Ops/s |
| 6791 | GB6 ML Single Precision | 18415 |
| 6738 | GB6 ML Half Precision | 29828 |
| 5128 | GB6 ML Quantized | 11425 |
| 1852 | Image Classification (SP) | 6797 |
| 3622 | Image Segmentation (HP) | 20808 |
| 8987 | Image Super Resolution (Q) | 15194 |
| 7074 | Face Detection (HP) | 37084 |
| 43726 | Pose Estimation (Q) | 82268 |
| 1446 | Text Classification (SP) | 2863 |
| 1576 | Machine Translation (HP) | 4135 |
| 2425 | Object Detection (SP) | 8559 |
| 12902 | Depth Estimation (Q) | 28330 |
| 86714 | Style Transfer (SP) | 184129 |
| ONNX | Framework | ONNX |
| DirectML | Backend | DirectML |
| 66 GPixel/s | Pixel Fill Rate | 108-153 GPixel/s |
| 123 GTexel/s | Texture Fill Rate | 261-369 GTexel/s |
| 3.9 TFLOPS | FLOPS (FP32) | 16.7-23.6 TFLOPS |
| GDDR5 | Memory Type | GDDR6 |
| 4 GB | Memory Size | 16 GB |
| 1753 MHz | Memory Clock | 2000 MHz |
| 7000 Mbps | Effective Memory Speed | 16000 Mbps |
| 256-bit | Bus | 256-bit |
| No | ECC | No |
| 224.4 GB/s | Memory Bandwidth | 512 GB/s |
| PCIe 3.0 x16 | Interface | PCIe 4.0 x16 |
| 148 W | TGP | 80-175 W (configurable) |
| TSMC | Manufacturing | Samsung |
| 28 nm | Fabrication Process | 8 nm |
| 398 mm² | Die Size | 496 mm² |
| 5 billion | Transistor Count | 22 billion |
| 12.56 MTr/mm² | Transistor Density | 44.35 MTr/mm² |
| 12 | DirectX | 12 |
| 1.3 | Vulkan | 1.2 |
| 4.6 | OpenGL | 4.6 |
| 3 | OpenCL | 3 |
| 5.2 | CUDA | 8.6 |
| No | Ray Tracing | Yes |
| No | DLSS | DLSS 2 |
| 1.2 | DisplayPort | 1.4a |
| Nvidia | Vendor | Nvidia |
| Discrete | Build | Discrete |
| September 19, 2014 | Released | January 1, 2022 |
| Desktop | Case | Laptop |
| Gaming | Purpose | Gaming |
| Mid-range | Segment | High-end |
| Maxwell 2.0 | Architecture | Ampere |
| GM204 | GPU Codename | GA103 (GN20-E8) |
| — | Recommended CPU | -Intel Core i9 12900Hor above |
| 3577 | Steel Nomad Lite Score | 12913 |
| 3666 | Time Spy | 12864 |
| — | Solar Bay | 56305 |
| 4353 | Port Royal | 8092 |
| 12102 | Fire Strike | 30858 |
| 6972 | Wild Life Extreme | 26307 |
| 47591 | Night Raid | 98582 |
| 1050 MHz | Base Clock | 585-1230 MHz (configurable) |
| 1178 MHz | Boost Clock | 1125-1590 MHz (configurable) |
| 1664 | Shading Units | 7424 |
| 104 | Texture Mapping Units (TMUs) | 232 |
| 56 | Render Output Units (ROPs) | 96 |
| 13 | Compute Units (Pipelines) | 58 |
| No | Tensor Cores | 232 |
| No | Ray-tracing Cores | 58 |
| 48KB per cluster | L1 Cache | 128KB per cluster |
| 2MB shared | L2 Cache | 4MB shared |
| 2 IPC | Instructions Per Cycle | 2 IPC |