Path: utzoo!utgpu!water!watmath!clyde!rutgers!husc6!hao!oddjob!gargoyle!ihnp4!ihlpe!res
From: res@ihlpe.ATT.COM (Rich Strebendt, AT&T-DSG @ Indian Hill West)
Newsgroups: comp.arch
Subject: Re: Approaches to computer system performance
Summary: Another vote for the high level approach
Keywords: Algorithms, Data structures, assembler, special purpose hardware
Message-ID: <2568@ihlpe.ATT.COM>
Date: 30 Jan 88 00:51:17 GMT
References: <839@ima.ISC.COM> <2158@geac.UUCP> <604@bnr-rsc.UUCP> <3081@watcgl.waterloo.edu>
Organization: AT&T Bell Laboratories - Naperville, Illinois
Lines: 25

In article <3081@watcgl.waterloo.edu>, tbray@watsol.waterloo.edu (Tim Bray) writes:
> The more years I build up in this
> business, the more I come to believe that technical sidesteps like
> assembler coding, special purpose hardware, and the like do not give
> anything like the payback you get from careful attention to algorithms.

I have been developing software for on the order of 20 years now.  I
agree completely with Tim.  By polishing and tuning and assemblerizing
and twiddling, you can get a fair bit of improvement.  In order to get
order-of-magnitude improvement you have to find a better algorithm, not
more tweeks and tunes.

As a case in point, I was asked to develop the on-board (burned in ROM)
software for a tape controller.  The goal was an order of magnitude
increase in performance.  The fundamental algorithm of the beast was
rethought and a completely different approach was used than that used
in the old controller.  We got better than the order of magnitude
improvement without one word of assembler.  The software was written in
C, and executes on a stock WE32100 microcomputer.  

So, if you need to get a big performance improvement in a program, do
not dink with the code -- pitch it and rethink the algorithms.

				Rich Strebendt
				...!ihnp4![iwsl6|ihlpe|ihaxa]!res