tech-userlevel archive

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

Re: floating point woes on i386



At Tue, 30 Sep 2025 12:00:31 +0200, Jörg Sonnenberger <joerg%bec.de@localhost> wrote:
Subject: Re: floating point woes on i386
> 
> On 9/30/25 9:30 AM, Greg A. Woods wrote:
> 
> > At Sat, 27 Sep 2025 10:36:59 +0200, Joerg Sonnenberger <joerg%bec.de@localhost> wrote:
> > Subject: Re: floating point woes on i386
> >> 
> >> There is on newer i386 also the option of
> >> using -mfpmath=sse. That's strictly
> >> speaking the only way to get IEEE754
> >> semantics on i386.
> > 
> > On NetBSD/i386 10.0 with the system's GCC-10.5 the -mfpmath=sse option
> > doesn't actually do anything:
> 
> Well, of course. You also need to use a
> CPU that supports SSE2 and/or enable the
> feature with -msse2.

Well, of course indeed!  Thanks for the reminder!  GCC is so much fun!
It is all really well documented in NetBSD's gcc(1) manual page, if I
had dared to read enough of it carefully!

Yes, with "-msse2 -mfpmath=sse" paranoia completes with zero flaws.

Adding "+387" to the latter also requires "-fexcess-precision=standard",
i.e. "-std=c89", but then it also completes, possibly slightly faster,
with zero flaws.

-- 
					Greg A. Woods <gwoods%acm.org@localhost>

Kelowna, BC     +1 250 762-7675           RoboHack <woods%robohack.ca@localhost>
Planix, Inc. <woods%planix.com@localhost>     Avoncote Farms <woods%avoncote.ca@localhost>

Attachment: pgp88fP2Q9u9_.pgp
Description: OpenPGP Digital Signature



Home | Main Index | Thread Index | Old Index