System Requirements
Data Sensors ships as a native Apple Silicon (arm64) build on macOS - there is no Intel Mac build - and as a 64-bit build on Windows 10 or later.
The numbers below come from real measurements, not estimates: idle memory, memory after loading a large file, and memory/CPU during a Model Competition run. Model Competition is deliberately the heaviest realistic scenario, since every competitor you select - Random Forest, XGBoost, and each Neural Network auto-tune variant - trains in its own real thread at the same time, rather than one after another.
| Minimum | Recommended | |
|---|---|---|
| RAM | 4 GB | 8 GB (16 GB for Model Competition on large files) |
| Disk space | 500 MB free | 1 GB free |
| CPU | Any 64-bit Apple Silicon or x86_64 processor from the last ~5 years | 4+ cores |
The installed app itself stays under 500 MB either way - about 310 MB on macOS, about 340 MB on Windows. The RAM figures above are mostly about the measurement data and trained models you work with, not the application shell.
Model Competition is the one scenario worth planning extra headroom for. Because every competitor trains in parallel rather than in sequence, memory and CPU use scale with how many competitors you select and how large your training data is, not just with the app's own baseline footprint. A representative run - 60 channels, 180,000 samples, 7 competitors including 5 Neural Network auto-tune variants - peaked at about 1.8 GB of memory and used every available CPU core.
To make this visible while you work rather than something you have to guess at, Data Sensors shows a live RAM/CPU usage readout - split into what the app itself is using, what everything else on your machine is using, and what's left - in both the Training Queue and the Model Competition wizard.