Re: Stats Question correction

New Message Reply Date view Thread view Subject view Author view

Sam Chu (c00chu00++at++nchc.gov.tw)
Sat, 4 Apr 1998 14:56:04 +0800 (CST)

>Yesterday I posted a response to a question about stats for a
>multi-channel, single pipe configuration.
>
>The stats for the channel were reporting something like this:
>
>frame=50.7 app=3.5 cull=5.7 draw=9.2 isect=2.3
>
>The question I got wrong was the interpretation of the "frame" stat.
>This figure is in fact the channel latency, i.e., the length of time from when
>the frame first enters the APP stage to the time when the DRAW stage
>completes. I erroneously reported this as rather the period time of the draw
>stage. So the numbers are not supposed to add up at all.

>In more recent versions of Performer the term "frame" has been replaced with
>"latency".
>

  I still get some confused. I try 'perfly' in a single channel single pipe
configuration, and got the following result.
  (I run "perfly -m 4 -p 1 esprit.flt" (using APPCULL_DRAW mode))

  15.0/30.0 HZ LOCK APPCULL_DRAW
  frame = 96.3 app=0.4 cull=1.0 draw=31.2 isect=0.0
  
  I think the reason that app+cull+draw+isect(32.6) in much small than
fram(96.3) is that the process in wait for the Transformation and Fill
stage to finish. So May I say that app+cull+draw is just for the CPU stage?

(In "Performance Turning & Debugging" chapter said that the Graphics
 pipeline comprises the CPU stage, Transformation stage and Fill statge)
(In "Frame and Load Contril" - Review of Rendering Stage said that
 Draw Issue graphics library commands to a Geometry Pipiline in order to
 create an image for subsequent display)

  If the above is true, how can I measure the time spent in the
Transformation stage and Fill stage?

  Thank you for your answer.

Sam Chu
National Center for High-Performance Computing
Scientific Visualization Lab Email: c00chu00++at++nchc.gov.tw
Tel: (886)35-776085 Ext 248 Fax : (886)35-773538
=======================================================================
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 - 17:57:12 PDT

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