Path: utzoo!utgpu!jarvis.csri.toronto.edu!rutgers!shelby!apple!usc!samsung!uunet!mcsun!hp4nl!ruuinf!piet
From: piet@cs.ruu.nl (Piet van Oostrum)
Newsgroups: gnu.gcc.bug
Subject: Re: Better code generated with -fno-function-cse despite documentation
Message-ID: <1801@ruuinf.cs.ruu.nl>
Date: 15 Nov 89 11:47:06 GMT
References: <8911102301.AA03955@deng>
Sender: news@ruuinf.cs.ruu.nl
Reply-To: piet@cs.ruu.nl (Piet van Oostrum)
Distribution: gnu
Organization: Dept of Computer Science, University of Utrecht, Holland
Lines: 16
In-reply-to: eb@LUCID.COM (Eric Benson)

In article <8911102301.AA03955@deng>, eb@LUCID (Eric Benson) writes:
 `
 `However, I have found at least one example of a function that actually
 `runs faster with -fno-function-cse that without.  This is the famous
 `
 `It looks like it might be preferable for -fno-function-cse to be the
 `default, at least in the case when a PC-relative call can be used.
 `
This is also true on the Harris HCX computers (tahoe clones). On this
machine a call with a PC-rel address is about 4 times as fast as one with
an indirect register.
-- 
Piet* van Oostrum, Dept of Computer Science, University of Utrecht
Padualaan 14, P.O. Box 80.089, 3508 TB Utrecht,  The Netherlands.
Telephone: +31-30-531806   Uucp: uunet!mcsun!hp4nl!ruuinf!piet
Telefax:   +31-30-513791   Internet: piet@cs.ruu.nl      (*`Pete')