Scanned file -force caching

[Pakal]

I observed that fastest format for scans is djvu. Although, it's still not as fast as I would like it: inside same page scroll is immidiate, switching to new page is ~6s and returning to page already visited(so cached) is ~2s.

Is there any way to "force" precaching pages? I have i62Firefly and checked that there is plenty of free RAM: base RAM usage is 100MB, after caching 20 pages it's ~130MB, so ~50 pages can be safely loaded.

The only 2 straighforward solution I can think of is merging pages(instead of having 50 pages XxY make it one huge page Xx50*Y) and forking djvu_reader.If I would find proper code in boox-opensource/code/src /djvu_reader/ would it suffice to compile it using toolchaing and replace one corresponding binary?

EDIT: I've checked, immidiate scrolling inside page is maintained even with page of 15x previous height, it's not really an universal good solution but at least noteworthy

[Inemi]

It's not normal that with a djvu you have to wait 6s each page.
Did you encoded the document to djvu? Is it a doc with colours?
My bitonal pages are usually 100-200kb, or even less, in hight resolution (2500p x 3500p). In contrast, a pdf page at the same quality should be 1-2Mb, and that explain why djvu is so fast (swithing to a new page is around 0.5s).
I don't understand how your djvu doc can be so complex to require so much memory and cpu.

From my point of view, no need to fork the reader, just re-encode better the document.

cheers

[Pakal]

Inemi: Big thanks! Haven't thought from this side... Pages were 2500x3500 and 100-200KB each, but switching was slow. I resized them to 900x1200 and now switching pages is blazing fast.

[vvilmos]

Is there a djvu reader for Onyx with search capabilities?