Re: Framerate on iR...

New Message Reply Date view Thread view Subject view Author view

Allan Schaffer (allan)
Thu, 4 Sep 1997 18:34:28 -0700


On Sep 3, 3:12pm, Jan Barglowski wrote:
>
> I'm trying to optimize my framerate. At 30Hz PFPHASE_LOCK my app
> runs well, but about every 2 seconds the stats display "jumps"
> and I consistently get 2-4 misses shown in the stats window. For
> the app/cull/draw misses, there are zero misses shown.
>
> Now the fun part. I run the app at 60Hz PFPHASE_LOCK and I get
> the exact same symptoms: 2-4 misses and none shown for each of the
> processes. Thinking it was my program, I then ran "perfly esprit.flt"
> and got the same statistics.
>
> Setup: 12cpu iR, IRIX 6.2, Performer 2.1, running programs as root
> with non-degrading priorities and locked-down CPUs. My app can easily
> run 60Hz (well, I disabled preFrame() and postFrame() for these tests ;-)

It's most likely that something is waking up (on another CPU) and
drawing some graphics, which causes a graphics context switch to
occur in the pipeline. Do you have a clock, gr_osview, gr_top, top,
osview, or any other program running which might be periodically
updating its display? Does your application print debug info into a
shell window? Are messages coming in from this mailing list every
two seconds? :-)

Even if the culprit is covered by your simulation window, it will
still cause the graphics pipe to switch contexts (so it can determine
which window -is- on top) and the simulation will glitch.

Allan

-- 
Allan Schaffer                                                allan++at++sgi.com
Silicon Graphics                               http://reality.sgi.com/allan
=======================================================================
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 - 17:55:53 PDT

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