Consistent lagspike with clicking

Hey guys, I’ve been having this odd lagspike occur in some of my games, which a friend also reports happening too. If you click and drag or click while moving your camera, the microprofiler picks up a lagspike and causes the game to freeze albeit just for a fraction of a second. This happens while you’re just flying around in studio, or in a “Run” playtest, but if you’re in a playtest controlling a character or in the live game this does not happen. It only happens in certain games, maybe having to do with part count? I have no idea what causes this and I’m just grasping at straws. I know this question is kind of vague but I’m really reaching out for any input if you have any. Thanks!

Here’s what it looks like:
https://cdn.discordapp.com/attachments/201831997784457216/911763137290723328/iveaGKWjAw.gif

(gif of me clicking while moving my camera and clicking & dragging, which causes the lag spike seen in the microprofiler up top

2 Likes

I believe this might just be a bug.

Change the micrprofiler to Detailed mode, and pause after clicking a few times, so you can see what is causing this issue.

The orange bar at the top of the microprofiler is not very useful.

I tried that, but there wasn’t anything particularly interesting (I think?) https://cdn.discordapp.com/attachments/201831997784457216/911764260927963156/unknown.png

Sorry, that’s not really enough context. Figure out where exactly the actual lag is during the click, and, take a few big screenshots of what’s going on, and, look around for things related to the click where the spike happens. You can narrow the view to just include the bar where the spike is, and, then you only need to look vertically.

The microprofiler can really help understand why lag is happening, but, you actually need to look at a lot of what’s going on, because it can be more than one thing.

You want to look for the sort of biggest groups, the stuff which is taking up the most horizontal space, because that’s the stuff taking up the most time, even if the bars are broken up in places.

1 Like

That’s pretty much what I did, though from the picture that might not have been clear. As far as I can tell, the only irregular pattern for each spike is the huge “Render” bar time-length. The processing before and after it seems relatively normal when compared to average frames. It’s more difficult than usual to scrub through the microprofiler right now because scrolling/zooming seems to be broken, but nothing really stands out at all.

I made a short video that might help contextualize the spike.