In practice, the performance boost depends on the task at hand.
Hardware acceleration is of great benefit to many raster-based tasks. Vector operations and specific features like blend ranges are performed on the CPU.
Tools, adjustments, filters and other operations including RAW development will use GPU resources to achieve improved performance.
The benefits are especially noticeable when stacking several Live filter layers together—export times are significantly quicker and canvas previewing is snappier.
As a trade-off, memory requirements are increased and performance may be dependent on the amount of VRAM available to the GPU(s).
The VRAM requirement depends on the complexity of your workflow. Document resolution and bit depth, screen resolution, and layer complexity all contribute to it.
Using a 4K display as a baseline, 1–2 GB of VRAM is sufficient for most light editing. For large amounts of compositing work, consider a GPU with 4GB—especially when working to 16-bit precision.
To perform 32-bit 3D rendering work with many layers, 4 GB is the minimum amount, though we recommend 8 GB.
Affinity support for Metal compute acceleration requires macOS High Sierra 10.13 or later.
Affinity support for safe unplugging of eGPUs requires macOS Mojave 10.14 or later.
Affinity support for OpenCL compute acceleration requires Windows 10.0.19042 (May 2020) or later.
It also requires GPU support for Direct3D 12 Feature Level 12.0, meaning the GPU must feature AMD's GCN (Graphics Core Next), NVIDIA's Maxwell, or Intel HD Graphics 510 (Skylake) or a later microarchitecture.
To assess the benefit of hardware acceleration to your system, use Affinity Photo's built-in benchmark to measure and compare single- and multi-core CPU, single GPU and, where applicable, multi-GPU performance.
The setting is inaccessible if Affinity Photo is unable to detect a compatible GPU on your system.
When enabled, compatible GPUs in use by the setting are listed under it.