home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!auspex-gw!guy
- From: guy@Auspex.COM (Guy Harris)
- Newsgroups: comp.sys.sun.misc
- Subject: FPUless Sun-3's (was Re: Solaris 2 vs Windows NT...)
- Message-ID: <16623@auspex-gw.auspex.com>
- Date: 26 Jan 93 19:37:28 GMT
- References: <1jf1l0INN6ak@gap.caltech.edu> <1johm0INNp4e@darkstar.UCSC.EDU> <16584@auspex-gw.auspex.com> <C1FI1E.7CE@siemens.com>
- Sender: news@auspex-gw.auspex.com
- Organization: Auspex Systems, Santa Clara
- Lines: 10
- Nntp-Posting-Host: auspex.auspex.com
-
- >Curiously enough, -fswitch seems to generate code that was *slower* than
- >-fsoft code on a machine with a 68881, and slower than -f68881 on machines
- >that have one.
-
- Nothing particularly curious about that - "-fswitch" generated code that
- called library routines that checked what kind of FP hardware you had -
- on *every* call - and then called the appropriate routine, while
- "-fsoft" generated code that called library routines that did FP in
- software. "-f68881", of course, generated code that used the 68881
- directly....
-