Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!uunet!mcsun!ukc!dcl-cs!aber-cs!athene!pcg
From: pcg@cs.aber.ac.uk (Piercarlo Grandi)
Newsgroups: comp.arch
Subject: Re: loop unrolling (was:Re: Register Count)
Message-ID: <PCG.91Jan19211942@teacho.cs.aber.ac.uk>
Date: 19 Jan 91 21:19:42 GMT
References: <PCG.91Jan10162301@odin.cs.aber.ac.uk> <11566@pt.cs.cmu.edu>
	<PCG.91Jan16195742@teacho.cs.aber.ac.uk> <5869@labtam.labtam.oz>
	<PCG.91Jan19120851@odin.cs.aber.ac.uk>
Sender: cho@aber-cs.UUCP
Organization: Coleg Prifysgol Cymru
Lines: 44
Nntp-Posting-Host: teacho
In-reply-to: pcg@cs.aber.ac.uk's message of 19 Jan 91 12:08:51 GMT

On 19 Jan 91 12:08:51 GMT, pcg@cs.aber.ac.uk (Piercarlo Grandi) said:

pcg> 	for (i = 0; i < K; i++)		[ ... ]
pcg> 	    /* body 1 */;

pcg> These can be rewritten as (4-way-unrolling):

pcg> 	for (i = 0; (i+4) < K; i += 4)	[ ... ]
pcg> 	{
pcg> 	    /* body 1 */;
pcg> 	    /* body 1 */;
pcg> 	    /* body 1 */;
pcg> 	    /* body 1 */;
pcg> 	}
pcg> 	if (i < K) switch (K-i)
pcg> 	{
pcg> 	case 3: /* body 1 */;
pcg> 	case 2: /* body 1 */;
pcg> 	case 1: /* body 1 */;
pcg> 	}

OOPS: sorry, I forgot that 'i' might be used in 'body 1' or after the
loop end; the corrected (hope no more errors!) version is:

	for (i = 0; (i+4) < K; i++)
	{
	    /* body 1 */; i++;
	    /* body 1 */; i++;
	    /* body 1 */; i++;
	    /* body 1 */;
	}
	if (i < K) switch (K-i)
	{
	case 3: /* body 1 */; i++;
	case 2: /* body 1 */; i++;
	case 1: /* body 1 */; i++;
	}

This does not really change the argument in which this example was used,
though.
--
Piercarlo Grandi                   | ARPA: pcg%uk.ac.aber.cs@nsfnet-relay.ac.uk
Dept of CS, UCW Aberystwyth        | UUCP: ...!mcsun!ukc!aber-cs!pcg
Penglais, Aberystwyth SY23 3BZ, UK | INET: pcg@cs.aber.ac.uk