lkml.org 
[lkml]   [2003]   [Aug]   [26]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
From
SubjectRe: [RFC] Re: Blender profiling-1 O16.2int
Date
On Tue, 26 Aug 2003 12:54, Kester Maddock wrote:
> On Sun, Aug 17, 2003 at 11:36:42PM +1000, Con Kolivas wrote:
> > Now normally, blender should just sleep and wait till X comes
> > alive again before it does anything. However here it shows clearly that
> > it is spinning madly looking for something from X, and poor X can't do
> > anything. This is the busy on wait I've described.
> >
> > Second, any applications that exhibit this should be fixed since it is a
> > bug.
>
> Say if blender is polling the mouse in the view rotate loop? (If so, you
> should be able to just hold down the middle mouse to starve, without moving
> the mouse.)
>
> Is this really a bug in the application? Blender is interactive while
> rotating the view. More CPU means more frames per second which gives the
> user a better experience. The CPU usage will drop down when the user
> releases the middle mouse button.
>
> (OK, in this specific case blender could update the screen on mouse move
> events, but what about the general case eg a 3d game, where the screen
> is updated by eg monster ai?)
>
> And how do you fix it? Would sleep(0) in these loops do, or do you need to
> select(...) on X?
>
> CC me please on replys.
>
> Thanks,
>
> Kester Maddock.
> ^ sends occaisional patches to blender.

Thanks for you attention. Yes this is a scheduler issue that makes it prone to
the effect of priority inversion (heck even the mars rover module suffered
it), brought out by the quirk in the coding. I wouldn't even know where to
begin looking at the source of blender but the culprits we've tracked down so
far that cause this use select with a very short timeout (15ms in one case).
Basically it repeatedly times out, and X never gets adequate scheduling time
to respond because the applications themselves are the ones preempting X.

Con

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

\
 
 \ /
  Last update: 2005-03-22 13:48    [W:0.114 / U:1.300 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site