Manta Interactive Ray Tracer Development Mailing List

Text archives Help


[MANTA] Manta benchmark: float vs. double


Chronological Thread 
  • From: Abe Stephens <abe@sci.utah.edu>
  • To: "'manta@sci.utah.edu'" <manta@sci.utah.edu>
  • Cc: Dan McLachlan <mclach@sgi.com>
  • Subject: [MANTA] Manta benchmark: float vs. double
  • Date: Thu, 13 Oct 2005 16:28:38 -0600
James corrected all of the manta code so that we can switch the Real typedef between float and double and still recompile.

I ran a few benchmarks comparing the code before all of his changes (r622), with the Real=float and Real=double (r623). There wasn't much of a difference in the Boeing777 code, however Real=double was faster for scene0.

Much of the performance critical code in the kdtree is hard coded float. We did a more limited experiment earlier and found a speedup from making that switch. (So the only difference in the path1 and path2 tests is in all of the core manta support code.

Boeing 777:
http://www.sci.utah.edu/~abe/manta_doc/mantatype-bench/path1.html
http://www.sci.utah.edu/~abe/manta_doc/mantatype-bench/path2.html
http://www.sci.utah.edu/~abe/manta_doc/mantatype-bench/path2-720x405.html

Scene 0:
http://www.sci.utah.edu/~abe/manta_doc/mantatype-bench/scene0-720x405.html

Abe




Archive powered by MHonArc 2.6.16.

Top of page