I don't think I have hyperthreading on that computer.  So, the single proc
computer was not locking up on the first press of 'm'.  
 
I'd debug it, but I don't have enough hard-drive space for 3 debug versions
of osg :(  Compiling the debug binaries of osg takes about 5GB of space with
VisualStudio 7, and a long time to compile.
 

Zach


 

  _____  

From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of Chuck Cole
Sent: Wednesday, February 07, 2007 16:58
To: 'osg users'
Subject: RE: [osg-users] CVS performance test



I tried the updated osgviewer.cpp and I get the same results - locks up when
pressing 'm'.  Again, I've been testing .txp databases although it doesn't
seem to make a difference what type of model is being displayed.  What's a
little interesting is that Zach is getting a slightly different result in a
similar configuration as mine (see items 6 and 7 in his list).  With my
Pentium 4 3.0GHz, hyperthreading on, Windows XP configuration, the viewer
locks up on the first press of 'm'.

 

I'll try and do some more debugging to see what I can find out.

 

chuck

 

  _____  

From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of Robert Osfield
Sent: Wednesday, February 07, 2007 4:17 PM
To: osg users
Subject: Re: [osg-users] CVS performance test

 

Hi Zach,

There is chance that its Windows threading issue, but more likely a general
threading issue.  Unfortuantely, I don't have enough cores to test
CullThreadPerCameraDrawThreadPerContext, but even without enough cores I've
seen a few instabilities with this model, its very new and not yet debug
well.  I've disabled the toggling to this threading model in osgviewer.cpp
and checked it in.

Could you try a cvs update and see how you get on.

Cheers,
Robert.



On 2/7/07, Zach Deedler < [EMAIL PROTECTED] <mailto:[EMAIL PROTECTED]> > wrote:

I tested on more computers.

INVESTIGATION continued...
1) Lock up with Pentium D 3.2GHz dual core with hyperthreading, Windows XP
(this looks like
4 procs in the task manager).
2) All tests done with 'osgviewer cow.osg', then pressing 'm'
3) see below
4) see below
5) Lock up with Pentium D 2.8GHz dual core without hyperthreading, Windows
XP
6) Does not up with Pentium 4 2.4GHz, Windows XP with the first press of
'm'.

7) Does lock up with Pentium 4 2.4GHz, Windows XP with the second press of
'm'.
8) All tests with OSG CVS today's head revision.
9) All tests with OpenThreads CVS today's head revision.
10) All tests with Producer CVS today's head revision.
11) Setting the processor affinity with the task manager to 1 CPU still
resulted in lockup after pressing 'm'



Looks like a problem with the windows implementation of the threading
models?

Zach

-----Original Message-----
From: [EMAIL PROTECTED] [
<mailto:[EMAIL PROTECTED]>
mailto:[EMAIL PROTECTED] On Behalf Of Zach Deedler

Sent: Wednesday, February 07, 2007 15:29
To: 'osg users'
Subject: RE: [osg-users] CVS performance test

INVESTIGATION continued
3) This is the DEBUG_INFO output till where it get stuck:
....
increment SwapBuffers
Doing op SwapBuffers 00F8E418
get op 0 00F8E418
increment RunOperation
Doing op RunOperation 00F8E418
Doing op Render 00F346C0
Viewer::stopThreading() - stopping threading
ViewerDoubleBufferedRenderingOperation::release()
Cancelling OperationsThred 00F8E418 isRunning()=1
   Doing cancel 00F8E418
   Waiting for OperationsThread to cancel 00F8E418
   Waiting for OperationsThread to cancel 00F8E418
   Waiting for OperationsThread to cancel 00F8E418
   Waiting for OperationsThread to cancel 00F8E418
   Waiting for OperationsThread to cancel 00F8E418
   Waiting for OperationsThread to cancel 00F8E418
   Waiting for OperationsThread to cancel 00F8E418exit loop
1   Waiting for OperationsThread to cancel
00F8E418
  OperationsThread::cancel() thread cancelled 00F8E418 isRunning()=0
Viewer::stopThreading() - stopped threading.
Viewer::startThreading() - starting threading Making scene thread safe
_numThreadsOnBarrier = 1

4) At this point osgviewer is using 0% CPU, and not rendering.


Zach

-----Original Message-----
From: [EMAIL PROTECTED]
[  <mailto:[EMAIL PROTECTED]>
mailto:[EMAIL PROTECTED] On Behalf Of Zach Deedler
Sent: Wednesday, February 07, 2007 15:21
To: 'osg users'
Subject: [osg-users] CVS performance test

Hello,

PROBLEM
With the CVS version of osgviewer, as soon as I hit 'm', the viewer locks up
every time.

INVESTIGATION
1) I have a Pentium D 3.2GHz dual core with hyperthreading  (this looks like
4 procs in the task manager).
2) This was just with cow.osg


Zach

_______________________________________________
osg-users mailing list
[email protected]
http://openscenegraph.net/mailman/listinfo/osg-users
http://www.openscenegraph.org/

_______________________________________________ 
osg-users mailing list
[email protected]
http://openscenegraph.net/mailman/listinfo/osg-users
http://www.openscenegraph.org/




_______________________________________________
osg-users mailing list
[email protected]
http://openscenegraph.net/mailman/listinfo/osg-users
http://www.openscenegraph.org/

 

<<attachment: image001.jpg>>

_______________________________________________
osg-users mailing list
[email protected]
http://openscenegraph.net/mailman/listinfo/osg-users
http://www.openscenegraph.org/

Reply via email to