i used v. 2.10.2-beta.0 and now it takes 25s to switch from one dataset to the other.

But it looks fast when switching to
image
Is there a special treatment for the first and second options ?

what you say is very interesting. When in ayah count mode, values range from 5 to 290, whereas in other modes value is from 1 to 114. Is slow down caused by small numbers?

I just deployed the latest version using 2.10.2-beta.0 and it is indeed very fast for Ayah mode

I tried using a wide range locally and got zero diff in performance.
Is this the only difference between datasets ?
Could you try using a hardcoded dataset for the ayah count to see if the speed is the same ?

Edit: I answered too quickly, there might be an impact in the perf of the placement algo when the difference between min and max is big. I'm investigating. A solution would be to clamp font sizes under a certain threshold

Try setting a lower value to config.style.chart.words.maxFontSize (to 64 or 50)
Basically having a big max font size makes finding placement for small words take more attempts, which is not ideal for large datasets. A sweet spot needs to be found with the config to balance the font size vs dataset length. The default maxFontSize is set to 100, which is quite big.

I can't really do much more than that right now, sorry.

You can upgrade to 2.10.2:

• slight performance improvement (but you'll have to tweak max font size when using a large dataset, or reduce the number of words to show)
• improvements for word selection on hover

Also using Suspense with a loader is better than blocking the main thread;)