[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: SIGFPE and -mieee



I guess I should probably say why I thought a GCC patch seemed like the best 
solution to me.

I think not having -mieee on by default (and a machine option to disable it) 
was an oversight in GCC philosophy on someones part.

Why?  Because looking at the rest of the GCC options, the attitude seems to be 
that default means be very conservative.  Faster, but possibly 
incorrect/non-standard conforming code code, is only generated when the user 
specifically requests it.

I mean, -O0 is the default optimization level!  The entry under -ffast-math 
says that not even any of the -O options will break exact IEEE math 
compliance.

I certainly know that not having IEEE compliance by default was a nasty shock 
(that involved me spending a lot of time trying to track down a non-existant 
bug in my code) when I first started using the Alpha.

Just think.  Works for everyone, and can be made faster for those who want it 
(and know what they are doing).  : )

Later  -T

PS:  Ideally, we could maybe take the -mieee options right out.  Have it 
turned on by default and a -no-mieee option to disable it.  That, however, 
would probably cause several moments of brief anoyance amoungst the package 
maintainers. *grin*

-- 
 Tyson Whitehead  (-twhitehe@uwo.ca -- WSC-)
 Computer Engineer                        Dept. of Applied Mathematics,
 Graduate Student- Applied Mathematics    University of Western Ontario,
 GnuPG Key ID# 0x8A2AB5D8                 London, Ontario, Canada



Reply to: