Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!sun-barr!olivea!uunet!ogicse!cvedc!gssc!timr
From: timr@gssc.UUCP (Tim Roberts)
Newsgroups: comp.lang.c
Subject: Re: micro-optimizing loops (was Help with casts)
Message-ID: <6597@gssc.UUCP>
Date: 26 Feb 91 20:44:05 GMT
References: <10191@dog.ee.lbl.gov> <344@smds.UUCP> <10250@dog.ee.lbl.gov>
Organization: Graphic Software Systems, Inc. Beaverton, Or.
Lines: 29

In article <10250@dog.ee.lbl.gov> torek@elf.ee.lbl.gov (Chris Torek) writes:
>(with any luck this will die its own death after this...)

Sorry.  I just had to comment.

In article <344@smds.UUCP> rh@smds.UUCP (Richard Harter) writes:
>For reasons that are not clear to me many optimizing compilers will not
>collapse the two machine instructions
>		dec r1
>		bge 1$
>into the available single instruction to do the same thing.  Perhaps
>some of our compiler writers can explain this to us.

From the Intel i486 Programmers Reference Manual and the Department of
Counterintuitive Advice:

  o The LOOP instruction takes two more clock cycles to execute than the
    equivalent decrement and conditional jump instructions.

Even more strange:

  o The JECXZ (jump on ECX=0) instruction takes one more clock cycle to 
    execute than the equivalent compare and conditional jump instructions.

-- 
timr@gssc.gss.com	Tim N Roberts, CCP	Graphic Software Systems
						Beaverton, OR

This is a very long palindrome. .emordnilap gnol yrev a si sihT