Quick/Comparisons
14 | Gaming | 35 |
24 | Workstation | 41 |
| — | AI/ML |
Compare with something else?
Compare Apple M4 GPU with another →97 | Energy Efficiency | 27 |
32 | Quick Comparison Final Score | 37 |
| Custom | Graphics Processor | TU104 |
| 1024 | Cores | 2944 |
| 64 / 32 | TMUs / ROPs | 184 / 64 |
| 1060.26 | Blender GPU | 2335.41 |
| — | Forza Horizon 5 | 81 |
| — | The Witcher 3 | 120 |
| — | Counter-Strike 2 | 160 |
| — | Far Cry 6 | 99 |
| — | Hogwarts Legacy | 65 |
| — | Call of Duty | 93 |
| — | Ghost of Tsushima | 62 |
| — | Cyberpunk 2077 | 75 |
| — | Tomb Raider | 123 |
| — | Average FPS | 98 |
| — | 1080p High | 119 |
| — | 1080p Ultra | 98 |
| — | 1440p Ultra | 72 |
| — | 4K Ultra | 40 |
| — | Margin of Error | Low |
| 30791 | GB6 Compute Score | 118637 |
| 57.4 img/sec | Background Blur | 191.8 img/sec |
| 41.3 img/sec | Face Detection | 115.9 img/sec |
| 1.21 Gpixels/sec | Horizon Detection | 4.74 Gpixels/sec |
| 1.74 Gpixels/sec | Edge Detection | 6.2 Gpixels/sec |
| 1.12 Gpixels/sec | Gaussian Blur | 6.56 Gpixels/sec |
| 0.43 Gpixels/sec | Feature Matching | 1.3 Gpixels/sec |
| 94.9 Gpixels/sec | Stereo Matching | 499.7 Gpixels/sec |
| 3902.5 FPS | Particle Physics | 15858.8 FPS |
| OpenCL | API | OpenCL |
| — | G3D Mark Score | 18607 |
| — | G2D Mark | 900 |
| — | DirectX 11 | 159 FPS |
| — | DirectX 12 | 71 FPS |
| — | GPU Compute | 7761 Ops/s |
| — | GB6 ML Single Precision | 17223 |
| — | GB6 ML Half Precision | 28694 |
| — | GB6 ML Quantized | 11608 |
| — | Image Classification (SP) | 7122 |
| — | Image Segmentation (HP) | 17897 |
| — | Image Super Resolution (Q) | 16120 |
| — | Face Detection (HP) | 34245 |
| — | Pose Estimation (Q) | 69601 |
| — | Text Classification (SP) | 2181 |
| — | Machine Translation (HP) | 4328 |
| — | Object Detection (SP) | 8748 |
| — | Depth Estimation (Q) | 28373 |
| — | Style Transfer (SP) | 154510 |
| — | Framework | ONNX |
| — | Backend | DirectML |
| 47 GPixel/s | Pixel Fill Rate | 109 GPixel/s |
| 94 GTexel/s | Texture Fill Rate | 315 GTexel/s |
| 3 TFLOPS | FLOPS (FP32) | 10.1 TFLOPS |
| System Shared | Memory Type | GDDR6 |
| — | Memory Size | 8 GB |
| 7500 MHz | Memory Clock | 1750 MHz |
| 15000 Mbps | Effective Memory Speed | 14000 Mbps |
| 128-bit | Bus | 256-bit |
| No | ECC | No |
| 120 GB/s | Memory Bandwidth | 448 GB/s |
| Custom | Interface | PCIe 3.0 x16 |
| 15 W | TGP | 215 W |
| TSMC | Manufacturing | TSMC |
| 3 nm | Fabrication Process | 12 nm |
| — | Die Size | 545 mm² |
| — | Transistor Count | 13 billion |
| — | Transistor Density | 23.85 MTr/mm² |
| — | DirectX | 12 |
| — | Vulkan | 1.3 |
| — | OpenGL | 4.6 |
| — | OpenCL | 3 |
| — | CUDA | 7.5 |
| Yes | Ray Tracing | Yes |
| No | DLSS | DLSS 2 |
| — | DisplayPort | 1.4a |
| Apple | Vendor | Nvidia |
| Integrated | Build | Discrete |
| October 29, 2024 | Released | September 20, 2018 |
| Laptop | Case | Desktop |
| Professional | Purpose | Gaming |
| Mid-range | Segment | High-end |
| Apple M GPU | Architecture | Turing |
| Custom | GPU Codename | TU104 |
| -Apple M4 (8-Core)or above | Recommended CPU | -AMD Ryzen 9 9900Xor above |
| 3450 | Steel Nomad Lite Score | 10886 |
| — | Time Spy | 11079 |
| 13908 | Solar Bay | 49857 |
| — | Port Royal | 6459 |
| — | Fire Strike | 27671 |
| 7928 | Wild Life Extreme | 22290 |
| — | Night Raid | 100751 |
| 500 MHz | Base Clock | 1515 MHz |
| 1470 MHz | Boost Clock | 1710 MHz |
| 1024 | Shading Units | 2944 |
| 64 | Texture Mapping Units (TMUs) | 184 |
| 32 | Render Output Units (ROPs) | 64 |
| 128 | Compute Units (Pipelines) | 46 |
| — | Tensor Cores | 368 |
| — | Ray-tracing Cores | 46 |
| — | L1 Cache | 64KB per cluster |
| — | L2 Cache | 4MB shared |
| 2 IPC | Instructions Per Cycle | 2 IPC |