home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!gatech!darwin.sura.net!mips!apple!cambridge.apple.com!bill@cambridge.apple.com
- From: bill@cambridge.apple.com (Bill St. Clair)
- Newsgroups: comp.lang.lisp.mcl
- Subject: Re: Short float arrays
- Message-ID: <9208172200.AA09365@cambridge.apple.com>
- Date: 17 Aug 92 23:03:27 GMT
- Sender: info-mcl-request@cambridge.apple.com
- Lines: 14
- Approved: comp.lang.lisp.mcl@Cambridge.Apple.C0M
- Full-Name: Bill St. Clair
- Original-To: Jim Hurd <71501.1347@CompuServe.COM>
- Original-Cc: info-mcl
-
- >>Not necessary. Since SHORT-FLOAT's are immediate objects, storing
- >>them in an array with an element-type of T is as good as anything.
- >
- >Wouldn't arrays of short-floats eliminate some masking of the tag bits?
-
- If MCL's compiler did floating point optimizations (which it doesn't), then
- a SHORT-FLOAT array element type would be useful. Your declaration of the
- array's type would sometimes allow the compiler to generate better code.
-
- Since the compiler doesn't optimize floating point computations,
- a SHORT-FLOAT array element type would actually slow things down.
- It would require AREF to translate from array storage format to
- immediate short-float format rather than just returning the bits
- as they're stored in the array.
-