Xref: utzoo comp.lang.c:8550 comp.lang.misc:1329
Path: utzoo!mnetor!uunet!husc6!bbn!gatech!bloom-beacon!athena.mit.edu!tada
From: tada@athena.mit.edu (Michael Zehr)
Newsgroups: comp.lang.c,comp.lang.misc
Subject: Re: Languages vs. machines (was Re: The need for D-scussion)
Message-ID: <4054@bloom-beacon.MIT.EDU>
Date: 26 Mar 88 00:34:05 GMT
References: <12176@brl-adm.ARPA> <1988Mar11.215238.976@utzoo.uucp> <10763@mimsy.UUCP> <5308@hall.cray.com>
Sender: daemon@bloom-beacon.MIT.EDU
Reply-To: tada@athena.mit.edu (Michael Zehr)
Organization: Massachusetts Institute of Technology
Lines: 39
Summary: yes, ... but

In article <5308@hall.cray.com> pmk@hall.cray.com (Peter Klausler) writes:
>In article <10763@mimsy.UUCP>, chris@mimsy.UUCP (Chris Torek) writes:
>> As long as programmers are taught to think in terms of a machine,
>> rather than the language, it will be difficult to get portable code
>> that can be moved to that 1 teraflop computer that will come out
>> tomorrow, then to the 10 teraflop computer that will come out the day
>> after, and then to the 100 teraflop computer that will come out a week
>> from Monday.
>
>A routine written for optimal performance on one architecture may be coded
>so that it is portable to others, but it'll be optimal on only the one.
>Write for optimal performance on a VAX and your code will crawl on a Cray.
>Code for maximal portability and you'll be suboptimal everywhere, having
>reduced yourself to a lowest common denominator of machine.

Other people write:
[other stuff about portability vs. performance]

When we have 1, 10, or 100 TFLOP machines, it's likely that no one will care
about the software for it being portable to 1-5 MIP machines, cuz no one
wants to wait 100,000 times as long for the program to finish.  Also, if
we port software from current 1-5 MIP machines, who's going to care that
it's sub-optimal on a 20 TFLOP machine? If it needs to be optimal on one of
those in order to run fast enough, then no one has spent time trying to make
it optimal for a 5 MIP machine, cuz there wouldn't be any use.  It's much
more important that code written for one 5 MIP machine is portable to (and
fairly easily optimized on) another 5 MIP machine, and likewise for TFLOP
machines, but who cares about going between the different categories?

Hopefully by the time we have machines of that speed, we'll have much
better languages to work with, and software that will convert C code
into a Language that is portable-but-has-fantastically-optimizing-
compilers-for-lots-of-different-architectures.  At least I hope so! :-)



-------
michael j zehr
"My opinions are my own ... as is my spelling."