Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.2 9/5/84; site scirtp.UUCP
Path: utzoo!watmath!clyde!bonnie!akgua!mcnc!rti-sel!scirtp!dfh
From: dfh@scirtp.UUCP (David F. Hinnant)
Newsgroups: net.lang.c
Subject: What does the optimiser optimise?
Message-ID: <455@scirtp.UUCP>
Date: Fri, 13-Sep-85 12:27:08 EDT
Article-I.D.: scirtp.455
Posted: Fri Sep 13 12:27:08 1985
Date-Received: Sat, 14-Sep-85 17:14:20 EDT
Distribution: net
Organization: SCI Systems, Research Triangle Park, NC
Lines: 48


  I am wondering (for the purpose of benchmarking) what kinds of C code
the optimiser (or bad-code-eliminator if you prefer) easily handles and
what kinds of constructs it may-or-may-not optimise.  I'm not so
interested in specifics, but rather what I sould expect from different
optimisers.  As a starting point, how are the following code fragments 
handled?


1)	x = (a+b);
	x = (a+b) +c;
	x = (a+b) +c / d;
	etc.

	Does the optimiser recognize that it's done (a+b) before and
	save this in a temporary location for future use?  How complicated
	can the (a+b) sub-expression become?

2)	for (i=1; i < x; i++)

	Is any optimisation done here?


3)	if (x) {
		.....
		x = 1;
		y = 1;
		z = 1;
	} else {
		.....
		x = 1;
		y = 1;
		z = 1;
	}

	Is a common section of code generated for the last 3 statements?
	I.e., does branch-tail merging occur?

4)	Is code that is never reached removed?


Replies I recieve by mail will be summarized if interest merits.

				As always, Thanks in advance.
-- 
				David Hinnant
				SCI Systems, Inc.
				{decvax, akgua}!mcnc!rti-sel!scirtp!dfh