It looks like I'm not the only one to notice this strange framerate issue.

http://www.pixelthismobile.com/blog/1/10/2010/flash-iphone-wildly-fluctuating-framerate-or-frame-skip

I'll investigate further tonight.

Kevin N.


On 6/17/11 11:19 AM, Kevin Newman wrote:
Hi guys,

I am running some benchmarks on iOS, to run AIR 2.7 through it's paces.

I used BunnyMark (and the blitting counter part) for context:
http://www.iainlobb.com/bunnies/BitmapTest.html

iPhone 3GS results:
Display List CPU: ~18 fps
Display List GPU: 24 fps (constant frame rate)
Blitting CPU: ~19 fps
Blitting GPU: I have to retest, my build was screwed, and it got late. :-)

I'll test iPhone 4 next.

There is a problem though - I don't trust the numbers on the CPU. It looks choppier than the frame rate would suggest, with some noticeable hiccups that aren't reflected in the frame rate number.

I suspect the problem is that Flash is skipping frames during rendering, but the ENTER_FRAME event is still ticking the FPS meter.

Is there a way to ensure an FPS measure only ticks if there is a RENDER event? Would moving it from ENTER_FRAME to RENDER actually do that?

I've only ever used RENDER with manual stage.invalidate() calls in ENTER_FRAME (and display list access all in RENDER, which might just be incorrect usage), and I suspect I'm actually disabling built in frame skipping by calling invalidate on every ENTER_FRAME event (because if the frame rate slows, so does the game logic) - I'm not really sure what the right way to use RENDER is.

So I guess that's the question, if I don't bother with stage.invalidate() and continue to do my display list stuff in ENTER_FRAME, and only measure FPS in RENDER, will I get a more accurate framerate value?

Thanks,

Kevin N.

_______________________________________________
Flashcoders mailing list
Flashcoders@chattyfig.figleaf.com
http://chattyfig.figleaf.com/mailman/listinfo/flashcoders

_______________________________________________
Flashcoders mailing list
Flashcoders@chattyfig.figleaf.com
http://chattyfig.figleaf.com/mailman/listinfo/flashcoders

Reply via email to