GeForce GTX 970
- Graphics Processor: GM204
- Cores: 1664
- TMUs / ROPs: 104 / 56
Review
Compare GeForce GTX 970 with other GPUs
Games
| Forza Horizon 5 | 58 33 26 16 |
| The Witcher 3 | 46 35 22 11 |
| Counter-Strike 2 | 77 56 37 16 |
| Far Cry 6 | 47 43 29 13 |
| Hogwarts Legacy | 25 22 15 - |
| CoD: Modern Warfare III | 38 32 21 15 |
| Ghost of Tsushima | 24 16 14 - |
| Cyberpunk 2077 | 27 21 13 - |
| Shadow of the Tomb Raider | 49 44 27 14 |
| 1080p High | 43 |
| 1080p Ultra | 34 |
| 1440p Ultra | 23 |
| 4K Ultra | 12 |
| Margin of Error | Low |
Test in Benchmarks
Performance tests in graphics benchmarks and applications3D Mark
| Time Spy | 3665 |
| Port Royal | 4353 |
| Fire Strike | 12101 |
| Wild Life Extreme | 6976 |
| Night Raid | 47583 |
GeekBench 6 OpenCL
| Background Blur | 77.1 img/sec |
| Face Detection | 38.1 img/sec |
| Horizon Detection | 1.28 Gpixels/sec |
| Edge Detection | 2.08 Gpixels/sec |
| Gaussian Blur | 1.85 Gpixels/sec |
| Feature Matching | 0.17 Gpixels/sec |
| Stereo Matching | 92.9 Gpixels/sec |
| Particle Physics | 2443 FPS |
| API | OpenCL |
Passmark Graphics
| G2D Mark | 763 |
| DirectX 11 | 71 FPS |
| DirectX 12 | 41 FPS |
| GPU Compute | 4079 Ops/s |
Artificial Intelligence Tests
Performance in machine learning and artificial intelligence tasksGeekBench 6 ML
| Image Classification (SP) | 1852 |
| Image Segmentation (HP) | 3622 |
| Image Super Resolution (Q) | 8987 |
| Face Detection (HP) | 7074 |
| Pose Estimation (Q) | 43726 |
| Text Classification (SP) | 1446 |
| Machine Translation (HP) | 1576 |
| Object Detection (SP) | 2425 |
| Depth Estimation (Q) | 12902 |
| Style Transfer (SP) | 86714 |
| Framework | ONNX |
| Backend | DirectML |
Specifications
Detailed technical specifications of the GeForce GTX 970General
| Vendor | Nvidia |
| Build | Discrete |
| Released | September 19, 2014 |
| Launch price (MSRP) | $329 |
| Case | Desktop |
| Purpose | Gaming |
| Segment | Mid-range |
| Architecture | Maxwell 2.0 |
| GPU Codename | GM204 |
| Successor | - GeForce RTX 5070 |
Desktop GPU rating - 86th place
Graphics Processing Unit
| Base Clock | 1050 MHz |
| Boost Clock | 1178 MHz |
| Shading Units | 1664 |
| Texture Mapping Units (TMUs) | 104 |
| Render Output Units (ROPs) | 56 |
| Compute Units (Pipelines) | 13 |
| Tensor Cores | No |
| Ray-tracing Cores | No |
| L1 Cache | 48KB per cluster |
| L2 Cache | 2MB shared |
| Instructions Per Cycle | 2 IPC |
Raw Performance
| Pixel Fill Rate | 66 GPixel/s |
| Texture Fill Rate | 123 GTexel/s |
Physical
| Interface | PCIe 3.0 x16 |
| TGP | 148 W |
| Manufacturing | TSMC |
| Fabrication Process | 28 nm |
| Die Size | 398 mm² |
| Transistor Count | 5 billion |
| Transistor Density | 12.56 MTr/mm² |
Memory
| Memory Type | GDDR5 |
| Memory Size | 4 GB |
| Memory Clock | 1753 MHz |
| Effective Memory Speed | 7000 Mbps |
| Bus | 256-bit |
| ECC | No |
API
| DirectX | 12 |
| Vulkan | 1.3 |
| OpenGL | 4.6 |
| OpenCL | 3.0 |
| CUDA | 5.2 |
| Ray Tracing | No |
| DLSS | No |
| DisplayPort | 1.2 |