home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.arch
- Path: sparky!uunet!mcsun!Germany.EU.net!incom!orfeo!qb!vhs
- From: vhs@rhein-main.de (Volker Herminghaus-Shirai)
- Subject: Re: Hardware Support for Numeric Algorithms
- Message-ID: <1992Nov22.192342.10288@qb.rhein-main.de>
- Sender: vhs@qb.rhein-main.de (Volker Herminghaus-Shirai)
- Reply-To: vhs@rhein-main.de
- References: <1efg1gINNm6c@network.ucsd.edu>
- Date: Sun, 22 Nov 92 19:23:42 GMT
- Lines: 91
-
- In article <1efg1gINNm6c@network.ucsd.edu> mbk@lyapunov.ucsd.edu (Matt Kennel) writes:
- > akao@.com (Adam Kao) writes:
- > : In article <Bxv2t2.4FH@mentor.cc.purdue.edu> hrubin@pop.stat.purdue.edu (Herman
- Rubin) writes:
- [lament deleted]
- > Poor ideas or confusing notation in mathematics disappears with the next
- > journal issue, but computer software and languages tend to stay around
- > for a long time, and so even more so than math, they require substantial
- > thought put in their design.
-
- The poorest idea mathematicians had was to use only single characters as
- variable names and optimize the multiplication symbol away. This definitely
- did not disappear with the next journal issue (there were no journals at
- that time). Instead, it led to using all sorts of special characters like
- greek or ancient german ones (for matrices) due to name space overcrowding.
- BTW I read an article here some time ago by a statistician who complained
- about keyboards that couldn't enter these special characters :-)
-
- > Also remember that Usenet, and especially these newsgroups, have a very
- > CS-centric view of things, and I think that more people than you expect
- > "out there" have feeling similar to mine or Herman's.
-
- Well, usenet runs on real computers, not on mathematical theories.
-
- > I certainly believe us "o'dinry fo'ks" out here have things to learn from
- > computer scientists, especially w/r/t the social realities of software
- > engineering, and I hope the computer scientists ought to try to learn some
- > from us.
-
- Just as much as from any other professional group. The bankers would probably
- kick our butts if CS only created languages with mathematical notations,
- orphaning their Cobol superprogrammers ;-)
-
- > Incidentally, my favorite computer language, Sather, (a simplified
- > speedified Eiffel), was designed by Steve Omohundro, a computer whiz,
- > who got a PhD in high-energy physics. Somehow, it just feels "right".
-
- If it feels "right" for you, use it. I doubt I would like it, though.
- If Mathemathics can create demand high enough to support (also
- financially) the development of a special language, CS will eventually
- do it. Just as for other sciences and professions.
-
- > Back to prof. Rubin's point. I think it would be nice if language
- > designers would recognize that *hardware WILL change*, and that
- > they design it in a way such that it becomes natural or easy for
- > future implementors to put in new operations that don't interfere with
- > old ones. The idea is not to freeze into a single model of computing,
- > and have hardware extensions treated as pariahs, only for use by grungy
- > hackers, and so we shouldn't give much though to them, and as a result
- > everybody has a different ugly way of doing them. It's just a change of
- > attitude, mostly.
-
- Too bad we never know in advance just *how* hardware will change.
-
- > Scientists, especially in the high-performance world, have a more concrete
- > view of computers, that may be sometimes short-sighted, but in other cases
- > may be more refreshing.
- >
- > As an example, consider a Connection Machine---it's hardware is so fundamentally
- > different, and so to even have a chance of writing intelligent programs, it's
- > essential to be able to express these fundamentally new methods in software.
- > (And yes, they do it.) Herman's point is that more computers than you think
- > may need this, especially as supercomputer-research-type stuff starts coming
- > down to earth. We don't want 87 completely different dialects of C, or tricky
- > obscure code that compiles into something simple. We accept the necessity
- > of hardware dependence, but we want it tamed, rather than excommunicated.
-
- In the very moment you get hardware dependent you also lose portability. If
- you accept loss in portability, do it in assembler. If you need a portable
- assembler with elementary type-conversion and safe subroutine calls, symbolic
- names for variables and little overhead, use C. That's about as close to
- assembler (as hardware dependent) as you can get without resorting to assembler.
- I also can hardly imagine what a language would look like that had some special
- operators for e.g. "FP division by integer value with remainder saved" on one
- architecture but would compile and run on another one.
- Other than #define-ing some assembler macros for each machine and inlining the
- expansions.
-
- > : Adam
- > : "Those French people have a different word for _everything_!!!"
-
- Right, the word for "everything" is "tout" in french, which is indeed
- a different word from "everything" :-)
-
- --
- Volker Herminghaus-Shirai (vhs@rhein-main.de)
-
- Computer industry: Industry in which the number of units sold of any
- given product is inversely proportional to its technical excellence.
- See also: MS-DOS, Windows, IBM-PC, X, QWERTY,
- 80x86, TrueType, Survival of the shittest.
-