home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!decwrl!contessa!mwm
- From: mwm@contessa.palo-alto.ca.us (Mike Meyer)
- Subject: Re: how to advocate new software/hardware features (Re: Hardware Support for Numeric Algorithms)
- Newsgroups: comp.arch
- Distribution: world
- References: <TMB.92Nov14145619@pollux.idiap.ch> <Bxq7y0.IJ@mentor.cc.purdue.edu> <mwm.2n6z@contessa.palo-alto.ca.us> <7903@charon.cwi.nl>
- X-NewsSoftware: Amiga Yarn 3.4, 1992/08/12 15:49:52
- Keywords:
- Summary:
- Message-ID: <mwm.2nup@contessa.palo-alto.ca.us>
- Date: 15 Nov 92 23:20:37 PST
- Organization: Missionaria Phonibalonica
- Lines: 25
-
- In <7903@charon.cwi.nl>, dik@cwi.nl (Dik T. Winter) wrote:
- > In article <mwm.2n6z@contessa.palo-alto.ca.us> mwm@contessa.palo-alto.ca.us (Mike Meyer) writes:
- > > In <Bxq7y0.IJ@mentor.cc.purdue.edu>, hrubin@mentor.cc.purdue.edu (Herman Rubin) wrote:
- > > > With instruction scheduling, one place for an optimizer would be in the
- > > > ASSEMBLER, and even interaction between the programmer and the compiler/
- > > > assembler. I do not know of any such.
- > >
- > > Herman, I've *never* seen you suggest something that hasn't been tried
- > > already, in some form or another. This is no exception. It's been
- > > tried. It didn't work very well.
- > >
- > Wrong. The MIPS assembler performs some optimizations (like delay slot
- > filling and instruction scheduling). It works quite well.
-
- My fault - I was referring to the "interaction between the programmer
- and the compiler/assembler" when I said "didn't work very well". I
- should have made things clearer.
-
- I do know that "optimizing" assemblers of various kinds have been
- around for a long time, and do work quite well. Given what I've seen
- of MIPS compiler technology (or should I call it SGI compiler
- technology now? :-), I'm sure they did wonderful things.
-
- <mike
-
-