Quick/Comparisons
14 | Gaming | 5 |
23 | Workstation | 13 |
10 | AI/ML |
Compare with something else?
Compare Apple M3 GPU with another →6 |
95 | Energy Efficiency | 75 |
30 | Quick Comparison Final Score | 19 |
| — | Graphics Processor | — |
| 1280 | Cores | 448 |
| 80 / 40 | TMUs / ROPs | 28 / 8 |
| 155.27 | Blender GPU | — |
| — | Forza Horizon 5 | 10 |
| — | The Witcher 3 | — |
| — | Counter-Strike 2 | 14 |
| — | Far Cry 6 | 11 |
| — | Hogwarts Legacy | - |
| — | Call of Duty | 13 |
| — | Ghost of Tsushima | - |
| — | Cyberpunk 2077 | - |
| — | Tomb Raider | 11 |
| — | Average FPS | 10 |
| — | 1080p High | 13 |
| — | 1080p Ultra | 10 |
| — | 1440p Ultra | - |
| — | 4K Ultra | - |
| — | Margin of Error | Low |
| 30027 | GB6 Compute Score | 14408 |
| 60.3 img/sec | Background Blur | 33.7 img/sec |
| 40.8 img/sec | Face Detection | 15.4 img/sec |
| 1.21 Gpixels/sec | Horizon Detection | 0.35 Gpixels/sec |
| 1.64 Gpixels/sec | Edge Detection | 0.76 Gpixels/sec |
| 1.2 Gpixels/sec | Gaussian Blur | 0.69 Gpixels/sec |
| 0.37 Gpixels/sec | Feature Matching | 0.21 Gpixels/sec |
| 102.6 Gpixels/sec | Stereo Matching | 54.7 Gpixels/sec |
| 3293.3 FPS | Particle Physics | 1957.6 FPS |
| OpenCL | API | OpenCL |
| — | G3D Mark Score | 2250 |
| — | G2D Mark | 568 |
| — | DirectX 11 | 16 FPS |
| — | DirectX 12 | 11 FPS |
| — | GPU Compute | 1215 Ops/s |
| 4120 | GB6 ML Single Precision | 3157 |
| 6992 | GB6 ML Half Precision | 4191 |
| 5638 | GB6 ML Quantized | 2365 |
| 1579 | Image Classification (SP) | 1323 |
| 2944 | Image Segmentation (HP) | 1675 |
| 4425 | Image Super Resolution (Q) | 2919 |
| 6661 | Face Detection (HP) | 4386 |
| 15020 | Pose Estimation (Q) | 9348 |
| 2984 | Text Classification (SP) | 1024 |
| 5876 | Machine Translation (HP) | 1888 |
| 1826 | Object Detection (SP) | 1452 |
| 9339 | Depth Estimation (Q) | 5272 |
| 22291 | Style Transfer (SP) | 19585 |
| Core ML | Framework | ONNX |
| CPU | Backend | DirectML |
| 55 GPixel/s | Pixel Fill Rate | 12-16 GPixel/s |
| 110 GTexel/s | Texture Fill Rate | 42-56 GTexel/s |
| 3.5 TFLOPS | FLOPS (FP32) | 1.3-1.8 TFLOPS |
| System Shared | Memory Type | System Shared |
| — | Memory Size | System RAM Dependent |
| 6400 MHz | Memory Clock | System RAM Dependent |
| 12800 Mbps | Effective Memory Speed | System RAM Dependent |
| 128-bit | Bus | — |
| No | ECC | No |
| 102.4 GB/s | Memory Bandwidth | — |
| Custom | Interface | IGP |
| 15 W | TGP | 10-45 W (configurable) |
| TSMC | Manufacturing | TSMC |
| 3 nm | Fabrication Process | 7 nm |
| — | Die Size | — |
| 13.8 billion | Transistor Count | — |
| — | Transistor Density | — |
| — | DirectX | 12.1 |
| — | Vulkan | — |
| — | OpenGL | 4.6 |
| — | OpenCL | 2.2 |
| — | CUDA | — |
| Yes | Ray Tracing | No |
| No | DLSS | No |
| — | DisplayPort | — |
| Apple | Vendor | Amd |
| Integrated | Build | Integrated |
| October 31, 2023 | Released | January 7, 2020 |
| Laptop | Case | Laptop |
| Professional | Purpose | Generic |
| Mid-range | Segment | Entry-level |
| Apple M GPU | Architecture | Raven |
| — | GPU Codename | — |
| -Apple M3or above | Recommended CPU | -AMD Ryzen 5 5500Uor above |
| 3478 | Steel Nomad Lite Score | 943 |
| — | Time Spy | 1068 |
| 13529 | Solar Bay | — |
| — | Port Royal | — |
| — | Fire Strike | 3396 |
| 8229 | Wild Life Extreme | 2015 |
| — | Night Raid | 13093 |
| 500 MHz | Base Clock | 300 MHz |
| 1380 MHz | Boost Clock | 1500-2000 MHz (configurable) |
| 1280 | Shading Units | 448 |
| 80 | Texture Mapping Units (TMUs) | 28 |
| 40 | Render Output Units (ROPs) | 8 |
| 160 | Compute Units (Pipelines) | 7 |
| — | Tensor Cores | No |
| — | Ray-tracing Cores | No |
| — | L1 Cache | — |
| — | L2 Cache | — |
| 2 IPC | Instructions Per Cycle | 2 IPC |