Rob Jenkins (robj++at++quid)
Wed, 9 Apr 1997 15:07:26 -0700
In a double buffered application swapbuffers can only occur at a vertical
retrace. If you monitor is running at 60 hz then as long as you finish your
draw ( including clears ) in 1/60 seconds ( so approx 16mS ) then you'll
achieve 60 hz. If you then add something so that your draw takes 17mS, you have
to wait for every other retrace and your frame rate will drop to 30 hz. To
truly measure the time taken to do N clears, run single buffer. The time taken
should be sort of linear against the numbers of windows but there may also be a
point were at some number of windows the time taken go dramatically up. This is
because somewhere the contexts are kept in some table that has a finite length
( N windows ) and if you try to shuffle through more than that you get into
heavy gfx context switching. N would be different on different platforms I
think. I think the thing you're seeing is the double buffer thing I mentioned
forst though.
You could experiment with running the monitor at 72 hz, then you only get 1/72
secs to draw but at least if you miss that time you can still achieve 36hz if
you can draw in 1/36 secs ( rather than 30hz and then sit spinning for the next
retrace maybe )
Cheers
Rob
On Apr 9, 7:00am, Bryan Wasileski wrote:
> Subject: screen clear
> performer:
> Recently, I ran numerous performance tests on an Onyx2 IR. One of the
> things I tested, which is cause for great concern, was how fast I could
> clear the screen as a function of the number of channels. From the
> results, it appears that I can clear up to 22 channels at 60Hz and beyond
> that it drops to 30Hz. This test was performed with a 2cpu R10k, 1 RM-7.
>
> The the only thing the test app did was clear channels. Nothing else was
> being done, drawn, culled, etc.....
>
> Can someone confirm or dispute these results. Where's the
> over? Thanks.
>
> - Bryan Wasileski
> McDonnell Douglas Training Systems
> St. Louis, MO
> (314) 232-8496
> bwasileski++at++mdc.com
>
> =======================================================================
> List Archives, FAQ, FTP: http://www.sgi.com/Technology/Performer/
> Submissions: info-performer++at++sgi.com
> Admin. requests: info-performer-request++at++sgi.com
>-- End of excerpt from Bryan Wasileski
--
________________________________________________________________
Rob Jenkins mailto:robj++at++csd.sgi.com
Silicon Graphics, Mtn View, California, USA
=======================================================================
List Archives, FAQ, FTP: http://www.sgi.com/Technology/Performer/
Submissions: info-performer++at++sgi.com
Admin. requests: info-performer-request++at++sgi.com
This archive was generated by hypermail 2.0b2 on Mon Aug 10 1998 - 17:55:02 PDT