Manta Interactive Ray Tracer Development Mailing List

Text archives Help


Re: [MANTA] Rendering to buffer for benchmarking Manta


Chronological Thread 
  • From: Abe Stephens <abe@sci.utah.edu>
  • To: Shreekanth Pavar <shreekanth.pavar@ucl.ac.uk>
  • Cc: MANTA <manta@sci.utah.edu>
  • Subject: Re: [MANTA] Rendering to buffer for benchmarking Manta
  • Date: Tue, 27 Feb 2007 08:07:45 -0700

Shreekanth Pavar wrote:
We are benchmarking Manta on an SGI Prism and would like to be able to render
to buffer but not display the image. According to the Manta Wiki you can do
this using "-imagedisplay null". However, this creates a window and then 
appears
to 'hang' when we run it.
It's actually rendering, just not displaying the image. Add a "-ui null" to disable the user interface as well.
We were also considering using the "-imagedisplay file" option to output the
frame buffers to .tga file format but the Wiki documentation says "This is
extremely expensive."
My guess is that comment suggests that it's extremely expensive relative to displaying the image on the local display.
Would you be able to advise us which of the above to use when measuring render
times with different polygonal datasets ?
There are two basic ways to measure performance the -bench command line option for a static view, or using animated camera paths. The wiki contains several examples of -bench and there is a page on camera paths here: http://code.sci.utah.edu/Manta/index.php/Camera_paths I've used camera paths in my own work.

There are some example command lines with benchmark results here:

http://code.sci.utah.edu/Manta/index.php/Commandlines

Abe





Archive powered by MHonArc 2.6.16.

Top of page