Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.2 9/17/84; site twitch.UUCP
Path: utzoo!watmath!clyde!cbosgd!ihnp4!houxm!hocad!twitch!grt
From: grt@twitch.UUCP ( G.R.Tomasevich)
Newsgroups: net.unix-wizards
Subject: C Optimization
Message-ID: <158@twitch.UUCP>
Date: Thu, 27-Dec-84 16:56:15 EST
Article-I.D.: twitch.158
Posted: Thu Dec 27 16:56:15 1984
Date-Received: Sat, 29-Dec-84 02:46:12 EST
References: <227@harvard.ARPA> <402@ecsvax.UUCP>
Organization: AT&T Bell Labs, Holmdel
Lines: 14

It seems to me that C was designed so that optimization would not really
be necessary.  One ought to be able to write code that cannot be optimized
any better by the compiler, aside from possible register usage details.
One can use register variables efficiently.  The current C optimization,
which eliminates extraneous machine jumps, ought to be enough.  Why break a
compiler which is straightforward now and is relatively close to the machine?
If you really want speed, use the assembler.

My past experience with FORTRAN H was that programs did not optimize much
if the code was written cleanly in the first place.

-- 
	George Tomasevich, ihnp4!{twitch|spastic}!grt
	AT&T Bell Laboratories, Holmdel, NJ