Jay Gischer (gischer++at++puget.engr.sgi.com)
Thu, 2 Apr 1998 10:24:58 -0800 (PST)
The key realization here is that two channels are being drawn
sequentially on one pipe
I've taken the liberty of rephrasing the questions slightly for the
sake of clarity.
Q Why are the app+cull+draw+isect times much smaller than the frame
time? What does "frame=50.6" mean anyway?
A. The app, cull, draw, isect times are measured times from beginning
to end of phase for ONE channel. The Frame time is a cycle time,
namely the time from the finish of one frame to the finish of
the next frame of that ONE channel. So it includes the time to
draw the other channel as well. In your case, app+cull+draw+isect
add up to about 21.2, leaving 28.4 for the other frame and some
possible overhead, which doesn't seem completely unreasonable.
Q.Why is 50.6 MSec * 30 Hz larger than 1000 MSec?
A. I'm not positive about this one, but I believe that "30Hz" is the
target frame rate (which isn't being realized), rather.than the
realized frame rate. Remember that gathering, and drawing stats actually
takes a fair amount of time.
Q. What do "Misses" mean, and why does one channel have "Misses=0" and
the other have "Misses=61".
A. A "Miss" is counted every time a channel does not finish drawing in
time to make the target refresh cycle.
Your app was missing its target frame rate (see previous question),
But one of the channels was done in plenty of time to hit the frame
rate, so it doesn't have any misses. The other one didn't finish
in time, so it gets a "Miss" counted against it.
I hope that clarifies matters.
------------------------------------------------------------------------
Jay L Gischer + "I see great things in baseball. It's our game.
Advanced Graphics + It will repair our losses and be a blessing to
Software + us."
Silicon Graphics + -Walt Whitman
(415) 390-4277 +
gischer++at++sgi.com + "A life has no meaning except in the impact it
+ has on other lives"
+ -Jackie Robinson
=======================================================================
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 - 17:57:11 PDT