Improve performance for Blueprint & Streams Panel for many entities #8808
Conversation
abey79
added
🚀 performance
|
Web viewer built successfully. If applicable, you should also test it:
Note: This comment is updated whenever you push a commit. |
Wumpf
changed the title
| /// Skipping rendering can increase performances for long lists that are mostly out of view, but | ||
| /// this will prevent any side effects from [`ListItemContent::ui`] from occurring. For this | ||
| /// reason, this is an opt-in optimization. |
There was a problem hiding this comment.
to clarify, could such a side-effect change the vertical size of a list array? If not, then wouldn't it be reasonable to expect this to be opt-in? I mean we're passing a closure after all for drawing a thing - if the thing isn't on screen, why would you expect your closure to be called; lotsa place don't do that
There was a problem hiding this comment.
No, the vertical size is fully determined by ListItem::height. Also, this change only affects the list item itself, not the fact that it might have children in a collapsing section below it. So there cannot be any issue with vertical layout.
And you are probably right that I'm being overly defensive with this opt-in stuff. I really don't have a strong reason to not make it opt-out (or even always on), except of being scared to break things in untested areas 😅
There was a problem hiding this comment.
hmmm I'd really like to make it opt-out in that case, but we can also do this maybe just after the next release then if we're too scared of adding regressions
abey79
force-pushed
the
antoine/list-item-render-offscreen
branch
from
6aaf2d6 to
707e7f3
Compare
Related
What
This PR adds an option to
ListItemto disable rendering offscreen. In that case,ListItemskips some code and in particular does not callListItemContent::ui(). The full sizing computation is still run, so it does not affect layout.Both the blueprint tree and the time panel now opt in to this optimisation.
DNM: chaineded to blueprint tree refactor to minimise conflicts
Benchmark
This optimisation is only measurable in "worst case" scenario, such as fully uncollapsed blueprint tree and streams tree in the air traffic data example. In this case, we shave about 13-15ms off the frame time (90% of which in the time panel) on a M4 Max.
Before/after screenshot (with the view hidden for better frame time stability): (61.8 vs. 47.4ms)