Quick/Comparisons
50 | Gaming | 39 |
59 | Workstation | 45 |
41 | AI/ML |
Compare with something else?
Compare Apple M3 Ultra with another →35 |
61 | Energy Efficiency | 39 |
52 | Quick Comparison Final Score | 40 |
| Custom | Graphics Processor | GA104 (GN20-E7) |
| 10240 | Cores | 6144 |
| 640 / 320 | TMUs / ROPs | 192 / 96 |
| 7360.18 | Blender GPU | 3258.81 |
| — | Forza Horizon 5 | 84 |
| — | The Witcher 3 | 127 |
| — | Counter-Strike 2 | 165 |
| — | Far Cry 6 | 104 |
| — | Hogwarts Legacy | 70 |
| — | Call of Duty | 98 |
| — | Ghost of Tsushima | 66 |
| — | Cyberpunk 2077 | 84 |
| — | Tomb Raider | 129 |
| — | Average FPS | 103 |
| — | 1080p High | 125 |
| — | 1080p Ultra | 103 |
| — | 1440p Ultra | 76 |
| — | 4K Ultra | 44 |
| — | Margin of Error | Low |
| 147719 | GB6 Compute Score | 128827 |
| 204.5 img/sec | Background Blur | 211 img/sec |
| 126 img/sec | Face Detection | 123.6 img/sec |
| 6.53 Gpixels/sec | Horizon Detection | 5.78 Gpixels/sec |
| 10.7 Gpixels/sec | Edge Detection | 8.03 Gpixels/sec |
| 9.5 Gpixels/sec | Gaussian Blur | 6.25 Gpixels/sec |
| 1.05 Gpixels/sec | Feature Matching | 1.17 Gpixels/sec |
| 560.3 Gpixels/sec | Stereo Matching | 543.2 Gpixels/sec |
| 25267.8 FPS | Particle Physics | 17774.9 FPS |
| OpenCL | API | OpenCL |
| — | G3D Mark Score | 16147 |
| — | G2D Mark | 630 |
| — | DirectX 11 | 145 FPS |
| — | DirectX 12 | 71 FPS |
| — | GPU Compute | 7126 Ops/s |
| 21356 | GB6 ML Single Precision | 18354 |
| 23530 | GB6 ML Half Precision | 29767 |
| 21706 | GB6 ML Quantized | 11487 |
| 8094 | Image Classification (SP) | 6773 |
| 24595 | Image Segmentation (HP) | 19587 |
| 25190 | Image Super Resolution (Q) | 15553 |
| 38337 | Face Detection (HP) | 35885 |
| 111021 | Pose Estimation (Q) | 82446 |
| 2786 | Text Classification (SP) | 2830 |
| 4339 | Machine Translation (HP) | 4019 |
| 7668 | Object Detection (SP) | 8380 |
| 37488 | Depth Estimation (Q) | 28597 |
| 244483 | Style Transfer (SP) | 184715 |
| Core ML | Framework | ONNX |
| GPU | Backend | DirectML |
| 442 GPixel/s | Pixel Fill Rate | 120-164 GPixel/s |
| 883 GTexel/s | Texture Fill Rate | 239-328 GTexel/s |
| 28.3 TFLOPS | FLOPS (FP32) | 15.3-21 TFLOPS |
| System Shared | Memory Type | GDDR6 |
| — | Memory Size | 8-16 GB |
| 6400 MHz | Memory Clock | 1500-1750 MHz |
| — | Effective Memory Speed | 12000-14000 Mbps |
| 1024-bit | Bus | 256-bit |
| No | ECC | No |
| 819.3 GB/s | Memory Bandwidth | 448 GB/s |
| Custom | Interface | PCIe 4.0 x16 |
| 140 W | TGP | 80-165 W (configurable) |
| TSMC | Manufacturing | Samsung |
| 3 nm | Fabrication Process | 8 nm |
| — | Die Size | 392 mm² |
| 184 billion | Transistor Count | 17.4 billion |
| — | Transistor Density | 44.39 MTr/mm² |
| — | DirectX | 12 |
| — | Vulkan | 1.2 |
| — | OpenGL | 4.6 |
| — | OpenCL | 2 |
| — | CUDA | 8.6 |
| Yes | Ray Tracing | Yes |
| No | DLSS | DLSS 2 |
| — | DisplayPort | 1.4a |
| Apple | Vendor | Nvidia |
| Integrated | Build | Discrete |
| March 5, 2025 | Released | January 12, 2021 |
| Desktop | Case | Laptop |
| Professional | Purpose | Gaming |
| High-end | Segment | High-end |
| Apple M GPU | Architecture | Ampere |
| Custom | GPU Codename | GA104 (GN20-E7) |
| -Apple M3 Ultraor above | Recommended CPU | -Intel Core i9 12900Hor above |
| 19491 | Steel Nomad Lite Score | 11592 |
| — | Time Spy | 11991 |
| 73690 | Solar Bay | 50371 |
| — | Port Royal | 7201 |
| — | Fire Strike | 29067 |
| 44759 | Wild Life Extreme | 24760 |
| — | Night Raid | 81949 |
| — | Base Clock | 780-1350 MHz (configurable) |
| 1380 MHz | Boost Clock | 1245-1710 MHz (configurable) |
| 10240 | Shading Units | 6144 |
| 640 | Texture Mapping Units (TMUs) | 192 |
| 320 | Render Output Units (ROPs) | 96 |
| 1280 | Compute Units (Pipelines) | 48 |
| — | Tensor Cores | 192 |
| — | Ray-tracing Cores | 48 |
| — | L1 Cache | 128KB per cluster |
| — | L2 Cache | 4MB shared |
| 2 IPC | Instructions Per Cycle | 2 IPC |