Quick/Comparisons
50 | Gaming | 12 |
59 | Workstation | 16 |
41 | AI/ML |
Compare with something else?
Compare Apple M3 Ultra with another →18 |
61 | Energy Efficiency | 89 |
52 | Quick Comparison Final Score | 26 |
| Custom | Graphics Processor | Arrow Lake iGPU |
| 10240 | Cores | 1024 |
| 640 / 320 | TMUs / ROPs | 64 / 32 |
| 7360.18 | Blender GPU | 765.53 |
| — | Forza Horizon 5 | 29 |
| — | The Witcher 3 | 33 |
| — | Counter-Strike 2 | 33 |
| — | Far Cry 6 | 35 |
| — | Hogwarts Legacy | 14 |
| — | Call of Duty | 27 |
| — | Ghost of Tsushima | 17 |
| — | Cyberpunk 2077 | 20 |
| — | Tomb Raider | 35 |
| — | Average FPS | 27 |
| — | 1080p High | 35 |
| — | 1080p Ultra | 27 |
| — | 1440p Ultra | 19 |
| — | 4K Ultra | 10 |
| — | Margin of Error | Low |
| 147719 | GB6 Compute Score | 42502 |
| 204.5 img/sec | Background Blur | 90 img/sec |
| 126 img/sec | Face Detection | 39.5 img/sec |
| 6.53 Gpixels/sec | Horizon Detection | 1.21 Gpixels/sec |
| 10.7 Gpixels/sec | Edge Detection | 1.52 Gpixels/sec |
| 9.5 Gpixels/sec | Gaussian Blur | 2.41 Gpixels/sec |
| 1.05 Gpixels/sec | Feature Matching | 0.51 Gpixels/sec |
| 560.3 Gpixels/sec | Stereo Matching | 161.8 Gpixels/sec |
| 25267.8 FPS | Particle Physics | 9155.9 FPS |
| OpenCL | API | OpenCL |
| — | G3D Mark Score | 6620 |
| — | G2D Mark | 835 |
| — | DirectX 11 | 47 FPS |
| — | DirectX 12 | 29 FPS |
| — | GPU Compute | 2992 Ops/s |
| 21356 | GB6 ML Single Precision | 8764 |
| 23530 | GB6 ML Half Precision | 12046 |
| 21706 | GB6 ML Quantized | 5815 |
| 8094 | Image Classification (SP) | 5224 |
| 24595 | Image Segmentation (HP) | 10501 |
| 25190 | Image Super Resolution (Q) | 6436 |
| 38337 | Face Detection (HP) | 15277 |
| 111021 | Pose Estimation (Q) | 19008 |
| 2786 | Text Classification (SP) | 1566 |
| 4339 | Machine Translation (HP) | 2969 |
| 7668 | Object Detection (SP) | 6172 |
| 37488 | Depth Estimation (Q) | 14526 |
| 244483 | Style Transfer (SP) | 62049 |
| Core ML | Framework | ONNX |
| GPU | Backend | DirectML |
| 442 GPixel/s | Pixel Fill Rate | 72-75 GPixel/s |
| 883 GTexel/s | Texture Fill Rate | 144-150 GTexel/s |
| 28.3 TFLOPS | FLOPS (FP32) | 4.6-4.8 TFLOPS |
| System Shared | Memory Type | System Shared |
| — | Memory Size | — |
| 6400 MHz | Memory Clock | — |
| — | Effective Memory Speed | — |
| 1024-bit | Bus | — |
| No | ECC | — |
| 819.3 GB/s | Memory Bandwidth | — |
| Custom | Interface | PCIe 4.0 x8 |
| 140 W | TGP | 35 W |
| TSMC | Manufacturing | TSMC |
| 3 nm | Fabrication Process | 5 nm |
| — | Die Size | — |
| 184 billion | Transistor Count | — |
| — | Transistor Density | — |
| — | DirectX | 12.2 |
| — | Vulkan | 1.3 |
| — | OpenGL | 4.6 |
| — | OpenCL | 3 |
| — | CUDA | No |
| Yes | Ray Tracing | Yes |
| No | DLSS | No |
| — | DisplayPort | 2.1 |
| Apple | Vendor | Intel |
| Integrated | Build | Integrated |
| March 5, 2025 | Released | January 6, 2025 |
| Desktop | Case | Laptop |
| Professional | Purpose | Generic |
| High-end | Segment | Entry-level |
| Apple M GPU | Architecture | Xe+ |
| Custom | GPU Codename | Arrow Lake iGPU |
| -Apple M3 Ultraor above | Recommended CPU | -Intel Core Ultra 5 235Hor above |
| 19491 | Steel Nomad Lite Score | 3255 |
| — | Time Spy | 3793 |
| 73690 | Solar Bay | 14650 |
| — | Port Royal | 1990 |
| — | Fire Strike | 8598 |
| 44759 | Wild Life Extreme | 6648 |
| — | Night Raid | 38856 |
| — | Base Clock | 300 MHz |
| 1380 MHz | Boost Clock | 2250-2350 MHz (configurable) |
| 10240 | Shading Units | 1024 |
| 640 | Texture Mapping Units (TMUs) | 64 |
| 320 | Render Output Units (ROPs) | 32 |
| 1280 | Compute Units (Pipelines) | 8 |
| — | Tensor Cores | 128 |
| — | Ray-tracing Cores | 8 |
| — | L1 Cache | — |
| — | L2 Cache | 4MB shared |
| 2 IPC | Instructions Per Cycle | 2 IPC |