System Requirements
Minimum Requirements
Section titled “Minimum Requirements”| Component | Minimum | Recommended |
|---|---|---|
| RAM | 4 GB | 16 GB or more for large whole-slide images |
| CPU | 2 cores, 64-bit x86-64 (Windows/Linux) or Apple Silicon arm64 (macOS) | 4+ cores for faster parallel pipeline execution |
| Free disk space | ~5 GB | More if you plan to store multiple AI model files locally |
| Display | 1280×800 | 1920×1080 or larger |
Supported Operating Systems
Section titled “Supported Operating Systems”| Operating system | Notes |
|---|---|
| Windows | Windows 10 or later, x86-64 |
| Linux | A modern 64-bit distribution (glibc 2.31 or newer - e.g. Ubuntu 20.04+, Debian 11+, Fedora 34+); see the Installation page for required GUI libraries |
| macOS | macOS 13 (Ventura) or later, Apple Silicon (arm64) |
GPU / CUDA Requirements
Section titled “GPU / CUDA Requirements”AI segmentation steps (Stardist, UNet, Cellpose) run on CPU by default and work on every supported platform. For significantly faster inference, download a CUDA build (Windows or Linux only - see Downloads) if your machine has:
| Component | Requirement |
|---|---|
| GPU | An NVIDIA GPU with compute capability 6.0 or higher |
| Driver | An NVIDIA driver new enough to support CUDA 12.x |
| VRAM | 4 GB minimum; more for large images or batches of high-resolution tiles |
If you don’t have a CUDA-capable GPU, install the regular CPU build - AI segmentation still works, just slower.
Scaling with Image Size
Section titled “Scaling with Image Size”EVAnalyzer streams large images in tiles rather than loading them fully into memory, but very large whole-slide or multi-well-plate datasets still benefit from more RAM and a faster disk (SSD recommended) - see Image Formats for how tiled/pyramidal formats are handled.