Re: different results with different f77 opti

bliss@worm.convex.com (Brian Bliss)
Thu, 5 Oct 1995 22:21:25 GMT

          From comp.compilers

Related articles
different results with different f77 optimizers keyes@chem.bu.edu (Tom Keyes) (1995-09-24)
Re: different results with different f77 optimizers ah739@cleveland.Freenet.Edu (1995-09-29)
Re: different results with different f77 opti bliss@worm.convex.com (1995-10-05)
| List of all articles for this month |

Newsgroups: comp.compilers
From: bliss@worm.convex.com (Brian Bliss)
Keywords: Fortran, optimize, arithmetic
Organization: Engineering, Convex Computer Corporation, Richardson, Tx USA
References: 95-09-146 95-10-026
Date: Thu, 5 Oct 1995 22:21:25 GMT

>>'Challenger' with IRIX 5.2. I compile the Fortran source with f77
>>-O3 -non_shared;if I try -O3 alone I get a message that the system does
...
>>quantities was radically different. I have checked enough now to verify
>>that these two different optimizers really do lead to different results.
>[...]
>
>I would trust the unoptimized version more.


I would also, at first. However, if the difference is due to a mathematical
instability in the algorithm itself, the optmized version will typically
give a more accurate answer. Reduction functions can be performed on
specialized hardware which typically has extended precision, and if the
compiler used associative transformations to perform tree-height reduction
of expression trees, the resulting calculation have a better mathematical
stability, on the "average".


--


Post a followup to this message

Return to the comp.compilers page.
Search the comp.compilers archives again.