From owner-freebsd-stable Fri Apr 27 18: 6:52 2001 Delivered-To: freebsd-stable@freebsd.org Received: from femail9.sdc1.sfba.home.com (femail9.sdc1.sfba.home.com [24.0.95.89]) by hub.freebsd.org (Postfix) with ESMTP id C87E337B423 for ; Fri, 27 Apr 2001 18:06:48 -0700 (PDT) (envelope-from graywane@home.com) Received: from cg392862-a.adubn1.nj.home.com ([65.2.79.221]) by femail9.sdc1.sfba.home.com (InterMail vM.4.01.03.20 201-229-121-120-20010223) with ESMTP id <20010428010648.UNXX17191.femail9.sdc1.sfba.home.com@cg392862-a.adubn1.nj.home.com> for ; Fri, 27 Apr 2001 18:06:48 -0700 Received: (from graywane@localhost) by cg392862-a.adubn1.nj.home.com (8.11.3/8.11.3) id f3S16jR01255 for stable@FreeBSD.ORG; Fri, 27 Apr 2001 21:06:46 -0400 (EDT) (envelope-from graywane) Date: Fri, 27 Apr 2001 21:06:45 -0400 From: Graywane To: stable@FreeBSD.ORG Subject: Re: Trouble with 4.3-RELEASE compiler Message-ID: <20010427210644.A1197@home.com> References: <20010427155725.L18676@fw.wintelcom.net> <200104280035.UAA11427@ns1.rwwa.com> Mime-Version: 1.0 Content-Type: multipart/signed; micalg=pgp-md5; protocol="application/pgp-signature"; boundary="FL5UXtIhxfXey3p5" Content-Disposition: inline User-Agent: Mutt/1.2.5i In-Reply-To: <200104280035.UAA11427@ns1.rwwa.com>; from witr@rwwa.com on Fri, Apr 27, 2001 at 08:38:30PM -0400 Sender: owner-freebsd-stable@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG --FL5UXtIhxfXey3p5 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline Content-Transfer-Encoding: quoted-printable On Fri, Apr 27, 2001 at 08:38:30PM -0400, User Witr wrote: > Out of curiosity, how much potential performance is FreeBSD throwing > away by banning -O2 and -O3? To my naive eyes it would seem better > to light that candle (try fix the -O2 bug) than curse the darkness. > Assuming there is significant performance gain and there truly is only > one major -O2 bug. -O2 used to be considered "safe" didn't it? The strength reduce "bug" has been present in gcc as far back as the 2.7 series. Every time they "fix" this "bug" they only end up subtly changing the program that demonstrates the problem. The problem is that it is hard to code a strength reduce optimization on platforms with small register sets (such as the Intel platform). Having very few people truly understand the optimization code while trying to support numerous platforms only exacerbates the problem. Try browsing the gcc code and you will see what I mean. If it were easy to fix then it would have been fixed many years ago. I have been bitten by the bug several times myself both in code I wrote and in other people code. One ftp program I compiled about a year ago would only crash if compiled with -fstrength-reduce (or any optimization level which included it -- like -O2). --=20 Note: See http://www.members.home.net/graywane/ for PGP information. --FL5UXtIhxfXey3p5 Content-Type: application/pgp-signature Content-Disposition: inline -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.0.4 (FreeBSD) Comment: For info see http://www.gnupg.org iEYEARECAAYFAjrqF6QACgkQeHdFaBWUGN3eYACfdQyKcXHy/UG5CQ449iLBw/hQ pxEAn09g16vn1QaveRSwcWS0X+POb04Z =ADmb -----END PGP SIGNATURE----- --FL5UXtIhxfXey3p5-- To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-stable" in the body of the message