Manta Interactive Ray Tracer Development Mailing List

Text archives Help


[Manta] Re: Measuring packet tracing


Chronological Thread 
  • From: Abe Stephens < >
  • To:
  • Subject: [Manta] Re: Measuring packet tracing
  • Date: Fri, 26 Sep 2008 10:23:49 -0600
  • Domainkey-signature: a=rsa-sha1; c=nofws; d=gmail.com; s=gamma; h=message-id:from:to:in-reply-to:content-type:mime-version:subject :date:references:x-mailer; b=W0RxI3lFoprd/8FSfpNrCmCegWQC0GB7TZ2DMxP3bt205bSVcHNBnh7K3USxFEjtw2 f7+U0SAehAbhIEwGY5nRaaXGdthIvZACMmYGRAxICF/mzEW1lW3LIVzTI7b8GEqPXkCF 6PEw9UGkl/IsXtuTquzQH+1uCXj7DvjJzX+fg=
You might have found this already, but there is a time view mode in the viewer: 

t                   time profile display
Control-t           time profile scale increase
Control-T           time profile scale decrease

If the normalized time measurement for the fragment is greater than one, the color saturates to the largest color map entry. By changing the scaling factor you can get an idea of how the cost varies across the image (you might add a printf to output the new scale when you change it). This isn't exactly what you asked for, but it might be a reasonable starting point. Also, depending on build configuration in cmake there might be more than one ray packet per fragment.

For very high resolution timing certain platforms support some intel timing extensions. See: Core/Thread/Timer_altix.cc this or something similar might be available on other platforms, I've never checked.  Also I'm not sure if the timers in Core/Util match the timers in Core/Thread/Timer*.


Abe


On Sep 26, 2008, at 9:46 AM, Biagio Cosenza wrote:

Hi!
I would like to know if it's possible to measure how much time does it take to trace a packet?

I am looking for  single-packet timing  (not just on average) spent in tracing.

I tried with both Manta/Utils classes and clock (measuring clock cycles) function, without success. The problem is that this time is too small.

Did you have any experience with this kind of problem?

Thanks, Biagio



---
Biagio Cosenza
ISISLab, Dip. di Informatica ed Applicazioni "Renato M. Capocelli"
Universita' degli Studi di Salerno
http://www.dia.unisa.it/~cosenza



Archive powered by MHonArc 2.6.16.

Top of page