Quick/Comparisons
14 | Gaming | 17 |
23 | Workstation | 27 |
10 | AI/ML |
Compare with something else?
Compare Apple M3 GPU with another →17 |
95 | Energy Efficiency | 61 |
30 | Quick Comparison Final Score | 27 |
| — | Graphics Processor | GA107 |
| 1280 | Cores | 2048 |
| 80 / 40 | TMUs / ROPs | 64 / 32 |
| 155.27 | Blender GPU | 1142.3 |
| — | Forza Horizon 5 | 43 |
| — | The Witcher 3 | 53 |
| — | Counter-Strike 2 | 79 |
| — | Far Cry 6 | 55 |
| — | Hogwarts Legacy | 31 |
| — | Call of Duty | 39 |
| — | Ghost of Tsushima | 26 |
| — | Cyberpunk 2077 | 36 |
| — | Tomb Raider | 61 |
| — | Average FPS | 47 |
| — | 1080p High | 59 |
| — | 1080p Ultra | 47 |
| — | 1440p Ultra | 32 |
| — | 4K Ultra | 17 |
| — | Margin of Error | High |
| 30027 | GB6 Compute Score | 48591 |
| 60.3 img/sec | Background Blur | 80.6 img/sec |
| 40.8 img/sec | Face Detection | 47.7 img/sec |
| 1.21 Gpixels/sec | Horizon Detection | 2.14 Gpixels/sec |
| 1.64 Gpixels/sec | Edge Detection | 2.82 Gpixels/sec |
| 1.2 Gpixels/sec | Gaussian Blur | 2 Gpixels/sec |
| 0.37 Gpixels/sec | Feature Matching | 0.61 Gpixels/sec |
| 102.6 Gpixels/sec | Stereo Matching | 199.8 Gpixels/sec |
| 3293.3 FPS | Particle Physics | 6124.5 FPS |
| OpenCL | API | OpenCL |
| — | G3D Mark Score | 9487 |
| — | G2D Mark | 524 |
| — | DirectX 11 | 65 FPS |
| — | DirectX 12 | 47 FPS |
| — | GPU Compute | 3856 Ops/s |
| 4120 | GB6 ML Single Precision | 9684 |
| 6992 | GB6 ML Half Precision | 18365 |
| 5638 | GB6 ML Quantized | 6700 |
| 1579 | Image Classification (SP) | 4109 |
| 2944 | Image Segmentation (HP) | 10426 |
| 4425 | Image Super Resolution (Q) | 8368 |
| 6661 | Face Detection (HP) | 18130 |
| 15020 | Pose Estimation (Q) | 32595 |
| 2984 | Text Classification (SP) | 2340 |
| 5876 | Machine Translation (HP) | 3225 |
| 1826 | Object Detection (SP) | 4684 |
| 9339 | Depth Estimation (Q) | 16324 |
| 22291 | Style Transfer (SP) | 70507 |
| Core ML | Framework | ONNX |
| CPU | Backend | DirectML |
| 55 GPixel/s | Pixel Fill Rate | 34-56 GPixel/s |
| 110 GTexel/s | Texture Fill Rate | 68-111 GTexel/s |
| 3.5 TFLOPS | FLOPS (FP32) | 4.3-7.1 TFLOPS |
| System Shared | Memory Type | GDDR6 |
| — | Memory Size | 4 GB |
| 6400 MHz | Memory Clock | 1375 MHz |
| 12800 Mbps | Effective Memory Speed | 11000 Mbps |
| 128-bit | Bus | 128-bit |
| No | ECC | No |
| 102.4 GB/s | Memory Bandwidth | 176 GB/s |
| Custom | Interface | PCIe 4.0 x8 |
| 15 W | TGP | 35-95 W (configurable) |
| TSMC | Manufacturing | Samsung |
| 3 nm | Fabrication Process | 8 nm |
| — | Die Size | 200 mm² |
| 13.8 billion | Transistor Count | 8.7 billion |
| — | Transistor Density | 43.5 MTr/mm² |
| — | DirectX | 12 |
| — | Vulkan | 1.3 |
| — | OpenGL | 4.6 |
| — | OpenCL | 3 |
| — | CUDA | 8.6 |
| Yes | Ray Tracing | Yes |
| No | DLSS | DLSS 2 |
| — | DisplayPort | 1.4a |
| Apple | Vendor | Nvidia |
| Integrated | Build | Discrete |
| October 31, 2023 | Released | March 30, 2022 |
| Laptop | Case | Laptop |
| Professional | Purpose | Professional |
| Mid-range | Segment | Entry-level |
| Apple M GPU | Architecture | Ampere |
| — | GPU Codename | GA107 |
| -Apple M3or above | Recommended CPU | -Intel Core i7 12800Hor above |
| 3478 | Steel Nomad Lite Score | 5042 |
| — | Time Spy | 4727 |
| 13529 | Solar Bay | 21233 |
| — | Port Royal | 762 |
| — | Fire Strike | 11952 |
| 8229 | Wild Life Extreme | — |
| — | Night Raid | — |
| 500 MHz | Base Clock | 712-1530 MHz (configurable) |
| 1380 MHz | Boost Clock | 1057-1740 MHz (configurable) |
| 1280 | Shading Units | 2048 |
| 80 | Texture Mapping Units (TMUs) | 64 |
| 40 | Render Output Units (ROPs) | 32 |
| 160 | Compute Units (Pipelines) | 16 |
| — | Tensor Cores | 64 |
| — | Ray-tracing Cores | 16 |
| — | L1 Cache | 128KB per cluster |
| — | L2 Cache | 2MB shared |
| 2 IPC | Instructions Per Cycle | 2 IPC |