home *** CD-ROM | disk | FTP | other *** search
- Path: sundog.tiac.net!usenet
- From: dmorin@tiac.net (Duane Morin)
- Newsgroups: comp.lang.c
- Subject: BORLAND Floating Point emulation question
- Date: Fri, 26 Jan 1996 14:29:45 GMT
- Organization: The Internet Access Company
- Message-ID: <4eaoa7$3cf@sundog.tiac.net>
- NNTP-Posting-Host: 206.119.56.19
- X-Newsreader: Forte Free Agent 1.0.82
-
- Hi,
-
- Apologies for the compiler-specific question, but this is really the
- most appropriate newsgroup I could find. I'm compiling up some stuff
- using Borland 4.5 (pure C, no C++) to put onto an embedded system.
- Problem - the embedded system has no math co-processor, so we need to
- do floating point emulation.
-
- Microsoft's compiler solved this nicely with an 'alternate math'
- option that didn't ever bother checking for the existence of a math
- chip, it just called some functions to do the math. Perfecto. Good
- speed, too.
-
- But we may need to use Borland. Borland seems to only have one
- emulation option, and that generates software interrupts that I really
- can't use. (I step into the interrupt and my program goes off into
- nevernever land. I don't even want to get into writing my own
- interrupt handlers, etc...)
-
- Am I right in this statement about Borland? Does it have the
- equivalent of Microsoft's "alternate math" support? How impossible
- would it be to convince Borland to use somebody else's math package
- (say, Microsoft's?)
-
- Thanks for any help,
- Duane
-
- ========================================================
- Views and opinions expressed above are mine and do not
- necessarily reflect those of my employer.
- http://www.tiac.net/users/dmorin/ <-- A real "home" page
- ========================================================
-
-