Quick/Comparisons
50 | Gaming | 11 |
59 | Workstation | 20 |
41 | AI/ML |
Compare with something else?
Compare Apple M3 Ultra with another →13 |
61 | Energy Efficiency | 58 |
52 | Quick Comparison Final Score | 15 |
| Custom | Graphics Processor | TU117 |
| 10240 | Cores | 896 |
| 640 / 320 | TMUs / ROPs | 56 / 32 |
| 7360.18 | Blender GPU | 443.47 |
| — | Forza Horizon 5 | 32 |
| — | The Witcher 3 | 33 |
| — | Counter-Strike 2 | 54 |
| — | Far Cry 6 | 41 |
| — | Hogwarts Legacy | 21 |
| — | Call of Duty | 31 |
| — | Ghost of Tsushima | 15 |
| — | Cyberpunk 2077 | 22 |
| — | Tomb Raider | 43 |
| — | Average FPS | 32 |
| — | 1080p High | 42 |
| — | 1080p Ultra | 32 |
| — | 1440p Ultra | 22 |
| — | 4K Ultra | 11 |
| — | Margin of Error | Low |
| 147719 | GB6 Compute Score | 39481 |
| 204.5 img/sec | Background Blur | 66.4 img/sec |
| 126 img/sec | Face Detection | 43.3 img/sec |
| 6.53 Gpixels/sec | Horizon Detection | 1.51 Gpixels/sec |
| 10.7 Gpixels/sec | Edge Detection | 1.68 Gpixels/sec |
| 9.5 Gpixels/sec | Gaussian Blur | 1.87 Gpixels/sec |
| 1.05 Gpixels/sec | Feature Matching | 0.49 Gpixels/sec |
| 560.3 Gpixels/sec | Stereo Matching | 187.9 Gpixels/sec |
| 25267.8 FPS | Particle Physics | 5261.5 FPS |
| OpenCL | API | OpenCL |
| — | G3D Mark Score | 7871 |
| — | G2D Mark | 553 |
| — | DirectX 11 | 58 FPS |
| — | DirectX 12 | 35 FPS |
| — | GPU Compute | 3036 Ops/s |
| 21356 | GB6 ML Single Precision | 7569 |
| 23530 | GB6 ML Half Precision | 5940 |
| 21706 | GB6 ML Quantized | 5581 |
| 8094 | Image Classification (SP) | 3208 |
| 24595 | Image Segmentation (HP) | 4254 |
| 25190 | Image Super Resolution (Q) | 7050 |
| 38337 | Face Detection (HP) | 6849 |
| 111021 | Pose Estimation (Q) | 23562 |
| 2786 | Text Classification (SP) | 1755 |
| 4339 | Machine Translation (HP) | 3096 |
| 7668 | Object Detection (SP) | 3710 |
| 37488 | Depth Estimation (Q) | 12946 |
| 244483 | Style Transfer (SP) | 48894 |
| Core ML | Framework | ONNX |
| GPU | Backend | DirectML |
| 442 GPixel/s | Pixel Fill Rate | 53 GPixel/s |
| 883 GTexel/s | Texture Fill Rate | 93 GTexel/s |
| 28.3 TFLOPS | FLOPS (FP32) | 3 TFLOPS |
| System Shared | Memory Type | GDDR5 |
| — | Memory Size | 4 GB |
| 6400 MHz | Memory Clock | 2001 MHz |
| — | Effective Memory Speed | 8000 Mbps |
| 1024-bit | Bus | 128-bit |
| No | ECC | No |
| 819.3 GB/s | Memory Bandwidth | 128.1 GB/s |
| Custom | Interface | PCIe 3.0 x16 |
| 140 W | TGP | 75 W |
| TSMC | Manufacturing | TSMC |
| 3 nm | Fabrication Process | 12 nm |
| — | Die Size | 200 mm² |
| 184 billion | Transistor Count | 4 billion |
| — | Transistor Density | 20 MTr/mm² |
| — | DirectX | 12 |
| — | Vulkan | 1.3 |
| — | OpenGL | 4.6 |
| — | OpenCL | 3 |
| — | CUDA | 7.5 |
| Yes | Ray Tracing | No |
| No | DLSS | No |
| — | DisplayPort | 1.4a |
| Apple | Vendor | Nvidia |
| Integrated | Build | Discrete |
| March 5, 2025 | Released | April 23, 2019 |
| Desktop | Case | Desktop |
| Professional | Purpose | Gaming |
| High-end | Segment | Entry-level |
| Apple M GPU | Architecture | Turing |
| Custom | GPU Codename | TU117 |
| -Apple M3 Ultraor above | Recommended CPU | -Intel Core i5 9400or above |
| 19491 | Steel Nomad Lite Score | 3419 |
| — | Time Spy | 3563 |
| 73690 | Solar Bay | 12275 |
| — | Port Royal | — |
| — | Fire Strike | 9696 |
| 44759 | Wild Life Extreme | 7350 |
| — | Night Raid | 46189 |
| — | Base Clock | 1485 MHz |
| 1380 MHz | Boost Clock | 1665 MHz |
| 10240 | Shading Units | 896 |
| 640 | Texture Mapping Units (TMUs) | 56 |
| 320 | Render Output Units (ROPs) | 32 |
| 1280 | Compute Units (Pipelines) | 14 |
| — | Tensor Cores | No |
| — | Ray-tracing Cores | No |
| — | L1 Cache | 64KB per cluster |
| — | L2 Cache | 1024KB shared |
| 2 IPC | Instructions Per Cycle | 2 IPC |