Texture Memory problems on iR

New Message Reply Date view Thread view Subject view Author view

WDea++at++swri.edu
Fri, 16 Aug 96 9:46:55 CDT


My program basically plays one movie after another for a long period of time.
This is accomplished by changing the texture on a polygon for each new
frame of the movie. It is supposed to run a 30 frames per second, and at the
beginning it does. Unfortunately, after a few of these movies have played
it slows down to 5 frames per second for no apparent reason. If I use an
internal format with 32 bit texels, this slowdown happens after
approximately 8 movies (including several frames/movie) have been loaded
and played. If I change it to use 16 bit texels it gets about twice as far
before slowing down. This implies to me that the problem has to do with
the available texture memory.

 I have tried everything that I can think of to free up the texture memory
after I am done with each movie. I am idling the textures and I have tried
every means I could think of for deleting them. The freeImage pfTexture
command only frees the image after it has no references (according to the
documentation). I don't see how this command can ever free the image
because as long as I can call it from the pfTexture, it is referenced by that
texture. pfDelete is equally ineffective for the same reason, and both pfFree
and pfMemory::free cause the system to crash. The program is not getting
anywhere near the 512 texture limit that I have heard of, yet. However,
that will become an issue that needs to be addressed at some point.

Currently, however, it appears that for some reason the texture memory
gets filled up, and even though most of the textures in it aren't being used
anymore they are still being paged in and out slowing things down.

The system I am using is an iR with 16 MB texture memory, 1 RM6 with 4 R10K
CPUs and 512 MB RAM. The software is Performer 2.1 on IRIX 6.2.

Has anyone heard or dealt with such a problem? Thanks for any help.

Wm. Frank Dea and Mark Lewis
wdea++at++swri.edu
Southwest Research Institute

=======================================================================
List Archives, FAQ, FTP: http://www.sgi.com/Technology/Performer/
            Submissions: info-performer++at++sgi.com
        Admin. requests: info-performer-request++at++sgi.com


New Message Reply Date view Thread view Subject view Author view

This archive was generated by hypermail 2.0b2 on Mon Aug 10 1998 - 17:53:22 PDT

This message has been cleansed for anti-spam protection. Replace '++at++' in any mail addresses with the '@' symbol.