Date: Thu, 23 Sep 1999 13:57:47 -0500 From: Dan Nelson <dnelson@emsphone.com> To: bsd@picard.mandrakesoft.de Cc: Matthew Dillon <dillon@apollo.backplane.com>, The Hermit Hacker <scrappy@hub.org>, freebsd-current@FreeBSD.ORG Subject: Re: gcc optimizer in -current system ... Message-ID: <19990923135747.A24995@dan.emsphone.com> In-Reply-To: <Pine.LNX.4.10.9909232009480.3675-100000@picard.mandrakesoft.de> References: <199909231803.LAA28917@apollo.backplane.com> <Pine.LNX.4.10.9909232009480.3675-100000@picard.mandrakesoft.de>
next in thread | previous in thread | raw e-mail | index | archive | help
In the last episode (Sep 23), bsd@picard.mandrakesoft.de said: > On Thu, 23 Sep 1999, Matthew Dillon wrote: > > I tend not to like the higher optimization levels because they > > cause the compiler to attempt to turn static functions into inlines > > and, in my opinion, it doesn't do a very good job of selecting > > which functions to convert. The result is that I see bloated > > binaries with no performance gain to show for it. > > > > EGCS's -Os is my favorite. > > Have you tried specifying -O6 and -Os (With -Os following -O6 because > we want it to override the values set by -O6)? I haven't tried it for > a while, but at least in an older egcs snapshot (somewhere between > 1.1.2 and gcc 2.95), it worked (optimize as much as possible, but > value size over speed). Order doesn't matter, since -Os is more like an optimization flag then a level (should really have been called -foptimize-size IMHO). It doesn't disable/enable anything else apart from forcing the minimum O level to 2. -- Dan Nelson dnelson@emsphone.com To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-current" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?19990923135747.A24995>