This reminded me, I saw tooltips being a large chunk when I profiled my react app. I should go and check that.

Similarly, adding a modal like this

{isOpen && <Modal isOpen={isOpen} onClose={onClose} />}

instead of

<Modal isOpen={isOpen} onClose={onClose} />

Seems to make the app smoother the more models we had. Rendering the UI (not downloading the code, this is still part of the bundle) only when you need it seems to be a low hanging fruit for optimizing performance.

Any time a library in your code goes from being used by a couple people to used by everyone, you have to periodically audit it from then on.

A set of libraries on our code had hit 20% of response time through years of accretion. A couple months to cut that in half, no architectural or cache changes. Just about the largest and definitely the most cost effective initiative we completed on that team.

Looking at flame charts is only step one. You also need to look at invocation counts, for things that seem to be getting called far more often than they should be. Profiling tools frequently (dare I say consistently) misattribute costs of functions due to pressures on the CPU subsystems. And most of the times I’ve found optimizations that were substantially larger improvements than expected, it’s been from cumulative call count, not run time.

This is one scenario where IMGUI approaches have a small win, even if it's by accident - since GUI elements are constructed on demand in immediate mode, invisible/unused elements won't have tooltip setup run, and the tooltip setup code will probably only run for the control that's showing a tooltip.

(Depending on your IMGUI API you might be setting tooltip text in advance as a constant on every visible control, but that's probably a lot fewer than 38000 controls, I'd hope.)

It's interesting that every control previously had its own dedicated tooltip component, instead of having all controls share a single system wide tooltip. I'm curious why they designed it that way.

Kinda annoying that the article doesn't really answer the core question, which is how much time was saved in the start up time. It does give a 0.05ms per tooltip figure, so I guess multiplied by 38000 gives ~2s saved, which is not too bad.

Don't have access to read the code, but I think ideally there should be only one instance created at startup, right?

I once made the mistake to buy some sound effects from Fab, I had to download the entire Unreal Engine and start it to create a project to then import the assets..

It took the whole afternoon

It's no wonder UE5 games have the reputation of being poorly optimized, you need an insane machine only just to run the editor..

State of the art graphics pipeline, but webdev level of bloat when it comes to software.. I'd even argue electron is a smoother experience tan Unreal Engine Editor

Insanity