Performance & large batches
Keep big batches fast and the rest of your system responsive.
When a run feels slow, or a big import makes the rest of your machine sluggish, the fix is almost always a few settings on the Performance tab — not a faster computer. This page is a quick recipe: pick the right speed/quality trade-offs, then pace large bulk imports and watched folders so nothing stalls.
Each fix below is a bold problem followed by what to change. Performance changes are staged — they're held until you click Apply in the bottom action bar, and then take effect on the next job — so you can change one, apply it, run a batch, and adjust.
Screenshot
The Settings window with the Performance tab open, showing Parallel OCR jobs, Adaptive performance, Page image quality, and OCR speed. — shot troubleshooting-performance-01
Tune for speed
Open Settings and select the Performance tab. Work through these in order — the first two control how much your machine takes on at once; the last two control how hard each page is worked.
Batches finish too slowly — raise Parallel OCR jobs
Parallel OCR jobs sets how many pages Scanix processes at the same time, from 1 to 16. More jobs finish a batch faster but use more CPU and memory.
- On a laptop, 2–4 is comfortable.
- On a workstation or server, push higher.
If you'd rather not tune the number by hand, apply a Quick profile (Workstation, Server, or Laptop) — Scanix detects your CPU and marks the recommended one Auto, then sets sensible values for you.
Screenshot
The Parallel OCR jobs control set toward the upper end of its range, with the Quick profiles card above it. — shot troubleshooting-performance-02
A big run bogs down the rest of your machine — leave Adaptive performance on
Turn Adaptive performance On. Scanix then automatically scales workers up or down based on live CPU and memory headroom, keeping the rest of your system responsive while still finishing batches quickly.
Leave it on as your default. It lets you set Parallel OCR jobs ambitiously without starving everything else when the machine gets busy.
Clean printed pages are reading slower than they need to — set OCR speed to Fast
OCR speed trades off how fast OCR runs against how much detail it captures:
| Setting | Trade-off | Best for |
|---|---|---|
| Fast | Up to 5× faster | High-volume batches of clean printed pages. |
| Balanced | Everyday default | Good speed with reliable accuracy. |
| Best accuracy | Slower | Reads tiny print and lower-quality scans more reliably. |
For a large stack of clean, printed documents, choose Fast. If results come back with mistakes — tiny print, faded or low-quality scans — step up to Balanced or Best accuracy.
Pages render heavier than they need to — match Page image quality to the source
Page image quality is the resolution PDF pages are rendered at before OCR. Higher resolution is sharper but slower and larger; the trick is to use just enough for your documents.
| Setting | Resolution | Best for |
|---|---|---|
| Draft | 150 dpi | Quick text preview, smallest files. |
| Standard | 300 dpi | Recommended for clean printed documents. |
| High | 450 dpi | Small fonts, faded scans, stamps. |
| Maximum | 600 dpi | Archival quality — slowest and largest. |
For most batches, Standard (300 dpi) is the sweet spot. Only move to High or Maximum when small fonts or faded scans are actually being misread — those settings cost speed and disk on every page.
A fast, clean-batch recipe
For a clean, high-volume run: set OCR speed to Fast, keep Page image quality at Standard (300 dpi), raise Parallel OCR jobs toward the top of your machine's comfort range — or just apply the Workstation or Server quick profile — and leave Adaptive performance On.
Large bulk imports and watch-folder rhythm
Two situations need their own pacing: dropping a very large pile of files in at once, and letting a watched (hot) folder ingest continuously.
A huge one-off import makes the app feel unresponsive. A massive drag-and-drop renders and queues a lot of pages at once. Keep Adaptive performance On so Scanix throttles itself under pressure, and lean on OCR speed → Fast plus Page image quality → Standard so each page is cheaper to process. If you don't need every page on screen immediately, importing in a couple of smaller drops keeps the interface snappier than one enormous batch.
A watched folder either reacts sluggishly or hammers the disk. Watched-folder pacing lives on its own Watch folders tab in Settings, separate from the throughput controls above. There you set the Scan rhythm — how often the folder is rescanned — from Fast through Battery saver:
- Pick a faster rhythm when you want files picked up the moment they land.
- Pick a more relaxed rhythm to be gentler on disks and network shares, or when the folder lives on a slow share.
The same Performance settings (parallel jobs, OCR speed, page quality) govern how quickly each picked-up file is then read, so tuning them above also speeds up everything a hot folder ingests.
Click Apply, then the next job picks it up
Performance changes are staged: click Apply in the bottom action bar to commit a new value. After that, OCR speed, Page image quality, and Parallel OCR jobs take effect on the next job or watcher tick — no restart needed. For the full reference, including which control needs a restart, see Resources & performance.
Troubleshooting
Next steps
Resources & performance
The full reference for every Performance, GPU, and Watch-folder setting and its values.
Import files & images
How drag-and-drop and file imports work, so large batches start out efficient.