Re: Draw Process

New Message Reply Date view Thread view Subject view Author view

Bernard Leclerc (bleclerc++at++poster.cae.ca)
Mon, 05 May 1997 22:40:54 -0400


James Barhorst wrote:

> Dear Performer Veterans,
>
> I am running a fairly simple 60 Hz performer application on a single
> processor of my 4-CPU Onyx (by using PFMP_APPCULLDRAW). When I
> display statistics, I notice that APP and CULL occur immediately at the
> beginning of the frame taking about 0.1 milliseconds each. Then nothing
> seems to happen for about 7 milliseconds until DRAW (which takes about
> 5-6 milliseconds) starts. DRAW always ends 3.3 milliseconds before the
> end of the frame. If DRAW takes longer, it seems to start sooner in the
> next frame. If it takes less time, it starts later next frame. I need more time
> at the end of each frame to do some additional processing on this CPU.
>
> Why is there this long delay before the DRAW? Is there any way to
> DRAW sooner?

James,

Could you provide us with the the pfPhase() used in your application? Also, remember
that in single process mode, the culling and the drawing are performed by the DRAW
stage. That explains why the statistics show almost no time spent in CULL. My guess
is the culling of your scene takes approximately 7 ms. Please, verify this by using
PFMP_APPCULL_DL_DRAW to force the pfCull() statistics to appear in the CULL stage.

Regarding your need for more time at the end of each frame, you don't have much
choice. In single process mode, at 60 Hz, you have only 16.6 ms to prepare the
frame, cull it, then render it. Tough luck ;-)

-- 
Bernard Leclerc, Technical Leader, 3-D Graphics Applications
CAE Electronics Ltd. (http://www.cae.ca)
8585 Cote De Liesse, Saint-Laurent, Quebec, Canada, H4L-4X4
tel: +1 514 341 2000 extension 2275, fax: +1 514 340 5496
=======================================================================
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 - 17:55:11 PDT

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