Sample breakout game has poor performance #2304
Replies: 1 comment 10 replies
-
|
@QuentinLeCaignec That doesn't seem right at all! I don't think you are doing anything wrong, this should work as you expect. I'm not able to reproduce this on my hardware using your sample repo. (On my older Surface Book 2 I'm running 60fps pretty consistently)
Can you collect some information for me:
|
Beta Was this translation helpful? Give feedback.
-
|
@eonarheim I've tried enabling the Canvas fallback on Firefox and it seems more stable, less constant up/down but not as smooth as chrome. This is with or without hardware acceleration. On chrome, enabling Canvas fallback with hardware acceleration doesn't seem to make a difference, the performances are smooth either way. Without hardware acceleration however, the Canvas fallback is noticeably more stable but still not perfectly smooth. So if make a sort of ranking of stability from most stable to least stable :
In general, Chrome can have spikes and sudden drops but seems smoother, whereas even with HA and Canvas firefox it never entirely stable. I might have missed your answer, but do you know why Firefox seems to perform generally worse than chrome ? |
Beta Was this translation helpful? Give feedback.
-
|
@QuentinLeCaignec thanks for the thorough research! Indeed, Firefox generally performs worse than chrome at least when it comes to graphics things. |
Beta Was this translation helpful? Give feedback.
-
|
@QuentinLeCaignec I've spent some time this week on #2309 and have managed to improve the performance on chrome/firefox a good deal. The non-HA is still not as good as I want, but I'm able to get the breakout sample to around 30fps. I think our near term solution (still in progress) will be an automatic 2D canvas fallback and/or a message to the user around enabling HA. I also updated the sample to use more robust collision logic to prevent the sticking and oscillating problems you spotted https://github.com/excaliburjs/sample-breakout I'm planning on a new version (v0.26.0) of Excalibur very soon (probably this weekend no promises though), but if you want to play with some of the performance bits early npm packages >= |
Beta Was this translation helpful? Give feedback.
-
|
Wow! I wasn't expecting such rapid improvement. Well I'm glad if I helped a little with my feedback :) |
Beta Was this translation helpful? Give feedback.
-
|
We've merged this feature! #2310 |
Beta Was this translation helpful? Give feedback.
-
Hello !
I thought I'd try this out to see if I liked it, as I'm interested in making simple games with TS.
My plan was to use Vite, and Vue on top of the excalibur app to handle UI, possibly to add complex and highly interactive UI elements (I'd also want to try integrating an excalibur game inside of a more "normal" webpage).
I made a simple little test with a Vue 3 + Vite + TS template, added excalibur and pasted the code from the sample breakout game (https://github.com/excaliburjs/sample-breakout/blob/main/src/main.ts#L17) in my main.ts.
I don't initialize/start the Vue App, only testing excalibur for now, and I serve with Vite in dev mode.
It seems to have pretty poor performances on both chrome and firefox. Even with browser devtools and excalibur-devtools off, it will not be able to maintain a constant 60 fps, constantly dropping to 50, 40, and will have seemingly random drops to 30 or lower even when the ball is just moving with no collision happening. Regularly the frames will drop at the wrong time and the ball will phase or half-phase through the paddle or the screen walls. For some reason on firefox with debug toggled, the performances are abysmal, hovering around 30 FPS at best.
On both chrome and firefox, I open a private window with no extensions.
Here's the repo : https://gitlab.com/QuentinLeCaignec/some-text-adventure
This doesn't seem normal for the most basic example game, especially the tutorial game shown by the docs. Am I doing something wrong? Is this normal?
Beta Was this translation helpful? Give feedback.
All reactions