The Flash 8 player has a flag on MoveClip to tell them to cache their bitmaps, which should result in much better performance (I tried it with the drawing API, and it had a pretty dramatic effect in the refresh rate for complex drawings). I tried turning it on on all movieclips but bad things happened to LzText instances among other things. But as an optimization, I would like to look at where and under what conditions it makes sense to enable/disable that flag in the view system.
On 3/14/06, Neil Mix <[EMAIL PROTECTED]> wrote:
I think this is unreleated, but I'll throw it out anyway to see if it
rings a bell:
I recently spent a lot of time looking into something similar to
this. I discovered that my application was using 25-30% CPU when in
a steady state doing *nothing*. After extensive simplification, I
discovered that CPU usage of a Flash app in a steady state is
correlated to the total number of movie clips in the app (whether or
not they contain or do anything) -- the more movie clips, the higher
the CPU usage. It is also correlated to the frames-per-second of the
application -- the higher the fps, the more CPU that is used up.
If I were to make a guess about what is happening under-the-covers, I
suspect that Flash visits every movie clip instance when switching
from frame to frame. The greater the number of movie clips and the
higher the fps, the more visits it has to make and thus processor
usage goes up. Interesting, although not necessarily a startling
revelation.
Given the relative ease with which movie clips are automatically
generated in a Laszlo app, it does seem like it's fairly easy to
bloat processor usage unintentionally. I'm not familiar enough with
the LFC (nor Flash programming in general) to know whether such movie
clip creation is absolutely necessary. It's understandable why works
this way, but it would also be nice if there were some simple magic
way to reduce the number of movie clips in my app; as it stands, I'd
have to do extensive refactoring. :(
-Neil
On Mar 13, 2006, at 6:50 PM, Henry Minsky wrote:
> I was running a test app which prints some info to the debugger
> every few seconds, and I noticed when I left it
> for a while, the CPU usage kept spiking when it prints. I think it
> is due to the progressively higher cost of the debugger appending
> text to it's buffer and computing the new scroll position. It's odd
> because I start seeing CPU loads spiking when the print routine is
> called on Windows of around 15% when there is as little as 30
> kbytes of text in the window.
>
> Anyway, something is really expensive here in Flash string
> handling, either the text appending or some other operation
> associated with the scrolling.
>
>
>
>
>
> --
> Henry Minsky
> Software Architect
> [EMAIL PROTECTED]
>
> _______________________________________________
> Laszlo-dev mailing list
> [email protected]
> http://www.openlaszlo.org/mailman/listinfo/laszlo-dev
--
Henry Minsky
Software Architect
[EMAIL PROTECTED]
_______________________________________________ Laszlo-dev mailing list [email protected] http://www.openlaszlo.org/mailman/listinfo/laszlo-dev
