Xref: utzoo comp.lang.misc:3521 comp.arch:11528 Path: utzoo!attcan!utgpu!jarvis.csri.toronto.edu!mailrus!ncar!unmvax!bbx!bbxsda!scott From: scott@bbxsda.UUCP (Scott Amspoker) Newsgroups: comp.lang.misc,comp.arch Subject: Re: Fast conversions, another urban myth? Message-ID: <167@bbxsda.UUCP> Date: 26 Sep 89 15:55:52 GMT References: <832@dms.UUCP> <688@UALTAVM.BITNET> <9dAz02zs58y201@amdahl.uts.amdahl.com> <27935@winchester.mips.COM> <136@bbxsda.UUCP> <1989Sep22.201906.10618@utzoo.uucp> <164@bbxsda.UUCP> <1989Sep25.184425.20936@utzoo.uucp> Reply-To: scott@bbxsda.UUCP (Scott Amspoker) Organization: Basis International, Albuquerque, NM Lines: 25 In article <1989Sep25.184425.20936@utzoo.uucp> henry@utzoo.uucp (Henry Spencer) writes: >Decimal *floating* point? For money? Are you sure you don't mean >decimal *fixed* point? (That is, integers plus a bit of compile-time >bookkeeping for decimal-point positions.) If you're thinking of COBOL >arithmetic, that's definitely fixed-point. Forgive me for not being more descriptive. I live with this stuff every day and forget that what I am doing is *not* very typical. What I am talking about is a decimal floatingpoint representation of numbers that are normally used for business calculations which includes an arbitrary amount of precision. The user may specify a desired precision at any time for any calculation. Fixed-point is not as practical under these circumstances. Data items do not carry a precision attribute with them. Any data item must be prepared to take on any precision (up to 14 digits) at any time. I'm not trying to convert the whole world over to this manner of doing math. I merely suggested that a little help in the hardware (such as access to the half-carry) would be of use to us and perhaps many other developers who are in a similar situation. -- Scott Amspoker Basis International, Albuquerque, NM (505) 345-5232