|
Analysis |
- 32 GB RAM (minimum)
- 16 GB RAM per CPU core (recommended)
|
- Advanced analysis operations use tiling and multithreading and therefore require more memory.
- More memory is required to fully utilize all available CPU cores.
|
|
Machine Learning/
Deep Learning
(GPU acceleration) |
|
- Ensure that the GPU Support Package is selected during installation.
|
|
Deep Learning Trainer |
- NVIDIA GPU with latest driver
- CUDA compute capabilities (version 7.5 or higher)
see https://developer.nvidia.com/cuda-gpus whether your GPU supports it.
- GPU that provides at least 8GB of memory (highly recommended)
- Python 3 environment (optional installer component)
|
|
|
AI model store Deep Learning Segmentation
(Instance segmentation) |
|
|
|
– Docker Desktop
(Tested successfully with version 4.77.0) |
- Windows 11 64-bit
- Enterprise, Pro or Education version 23H2 (build 22631) or higher.
- BIOS-level Hardware Virtualization support must be enabled
- 64 bit processor with Second Level Address Translation (SLAT)
- WSL 2 feature on Windows has to be enabled, Version 2.1.5 or later (see https://docs.docker.com/desktop/wsl/)
- If your version of Windows does not support the WSL 2 feature: Hyper-V and Containers Windows Features must be enabled.
- NVIDIA GPU with latest drivers supporting WSL 2 GPU Paravirtualization.
- GPU support is only available on Windows with the WSL2 backend.
- CUDA Compute Capability ≥ 7.0 (Volta or newer)
- At least 8 GB CPU RAM
|
|
|
– Alternative setup
(If Docker Desktop or WSL2 cannot be used locally, you can use a second machine to run Docker loads remotely.) |
- Separate Linux machine (a real PC or a VM) e.g. Debian or Ubuntu
- NVIDIA GPU (8GB RAM or higher) with CUDA Compute Capability ≥ 7.0 (Volta or newer)
- Docker engine, configured for Docker Remote Access and NVIDIA Container Toolkit
|
|
|
AI-assisted mode of the Draw Tool |
- NVIDIA GPU (8GB or more recommended)
- GPU Support Package (optional) is installed
- Support for AI-assisted Draw Tool (optional) with additional larger AI models is highly recommended
|
- Note: If these requirements are not met, calculations will default to the CPU, resulting in slower performance and reduced resolution. A smaller AI model will be used, which may decrease accuracy, especially if the GPU has less than 8GB of memory or the optional larger models are not installed.
|