From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from pigeon.gentoo.org ([208.92.234.80] helo=lists.gentoo.org) by finch.gentoo.org with esmtp (Exim 4.60) (envelope-from ) id 1RTI77-0006HB-VM for garchives@archives.gentoo.org; Wed, 23 Nov 2011 19:04:13 +0000 Received: from pigeon.gentoo.org (localhost [127.0.0.1]) by pigeon.gentoo.org (Postfix) with SMTP id C0C8721C0CB; Wed, 23 Nov 2011 19:03:56 +0000 (UTC) Received: from mail-bw0-f53.google.com (mail-bw0-f53.google.com [209.85.214.53]) by pigeon.gentoo.org (Postfix) with ESMTP id 2E2CD21C062 for ; Wed, 23 Nov 2011 19:02:16 +0000 (UTC) Received: by bkaq10 with SMTP id q10so2349345bka.40 for ; Wed, 23 Nov 2011 11:02:16 -0800 (PST) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=gamma; h=mime-version:in-reply-to:references:date:message-id:subject:from:to :content-type; bh=aQSGuFgejOTV7z4WN5Y/v/Y9wGFnqlFRfPscmLJ4kDo=; b=OaVgNsbTIT3nFsZStGVEgS05HeJm9Zkl5VY36CHwvtetugFQFKu6hZTrAcMbBhpo0O ntRoSZ1DgozugJ9BvFbaACzxOH9dRziaWE2PsvDflyxLrnIBEq7usZVqY2Vl9mYy1F1F yxwaHjJWPFF5YOVrwnNlFo8xx+BrLiKqmpaiE= Precedence: bulk List-Post: List-Help: List-Unsubscribe: List-Subscribe: List-Id: Gentoo Linux mail X-BeenThere: gentoo-user@lists.gentoo.org Reply-to: gentoo-user@lists.gentoo.org MIME-Version: 1.0 Received: by 10.204.155.152 with SMTP id s24mr25754910bkw.5.1322074936166; Wed, 23 Nov 2011 11:02:16 -0800 (PST) Received: by 10.204.114.75 with HTTP; Wed, 23 Nov 2011 11:02:16 -0800 (PST) Received: by 10.204.114.75 with HTTP; Wed, 23 Nov 2011 11:02:16 -0800 (PST) In-Reply-To: <4ECD4009.7030307@xunil.at> References: <20111112005129.003a54d3@digimed.co.uk> <4EBF2185.7050801@gmail.com> <20111113115637.3766e813@digimed.co.uk> <4EC2FB10.6040106@xunil.at> <20111116002014.70c4fad4@zaphod.digimed.co.uk> <4EC388D7.6040907@xunil.at> <20111116100933.16d6c90f@digimed.co.uk> <4EC39A5D.2090801@xunil.at> <20111116112510.500a7f06@zaphod.digimed.co.uk> <4EC3A7C0.7010404@xunil.at> <4EC3DAE1.1050006@xunil.at> <20111116192327.5499e87b@digimed.co.uk> <4EC42C24.1090007@xunil.at> <20111117012727.050497db@digimed.co.uk> <4EC4E59D.4050407@xunil.at> <4EC5870B.4000301@xunil.at> <20111118000448.315087f7@digimed.co.uk> <4EC62982.1080709@xunil.at> <4ECA0F9A.4030509@xunil.at> <4ECA521E.4000500@xunil.at> <4ECD2583.2000709@xunil.at> <4ECD4009.7030307@xunil.at> Date: Wed, 23 Nov 2011 14:02:16 -0500 Message-ID: Subject: Re: [gentoo-user] Re: Re: Re: Another hardware thread From: Michael Mol To: gentoo-user@lists.gentoo.org Content-Type: multipart/alternative; boundary=000e0ce02e26efc9d104b26b8ef8 X-Archives-Salt: 43ce1660-f076-4e76-a712-7bdb6101a1bc X-Archives-Hash: 2079574ebb0f6bc5f71e296ea133ca07 --000e0ce02e26efc9d104b26b8ef8 Content-Type: text/plain; charset=UTF-8 On Nov 23, 2011 1:50 PM, "Stefan G. Weichinger" wrote: > > Am 23.11.2011 18:05, schrieb Michael Mol: > > On Wed, Nov 23, 2011 at 11:55 AM, Stefan G. Weichinger wrote: > >> > >> Ad CFLAGS for i7-2600: > >> > >> Is that too much ricer-style? --> > >> > >> ### gcc -march=native -E -v - &1 | sed -n 's/.* -v - //p' > >> > >> CFLAGS="-O2 -D_FORTIFY_SOURCE=2 -march=core2 -mcx16 -msahf -maes > >> -mpclmul -mpopcnt -mavx --param l1-cache-size=32 --param > >> l1-cache-line-size=64 --param l2-cache-size=8192 -mtune=generic" > >> > >> gcc-4.5.3-r1 ... > > > > That's equivalent to, what, "-O2 -D_FORTIFY_SOURCE=2 -march=native" ? > > I don't see anything to complain about. > > Ok, thanks. > > compiled gcc-4.6.2 (hey, it only takes ~16min now!): > > # /usr/bin/gcc-4.6.2 -march=native -E -v - &1 | sed -n > 's/.* -v - //p' > -D_FORTIFY_SOURCE=2 -march=corei7-avx -mcx16 -msahf -mno-movbe -maes > -mpclmul -mpopcnt -mno-abm -mno-lwp -mno-fma -mno-fma4 -mno-xop -mno-bmi > -mno-tbm -mavx -msse4.2 -msse4.1 --param l1-cache-size=32 --param > l1-cache-line-size=64 --param l2-cache-size=8192 -mtune=corei7-avx > > If I am "feeling adventurous" (Neil) I could now try my luck by choosing > this nice new compiler w/ options. Maybe *after* a nice backup. > > > Actually, it's pretty interesting seeing what that processor comes > > down to for -march=native. I wish there were a database of processors > > and their decomposed compiler tuning flags for comparison. That would > > be *very* interesting, from the standpoint of proc shopping and > > looking at the evolution of CPUs. > > Wouldn't a wiki-page do the trick? gentoo-wiki ... > > S > ISTR gcc's i7 optimizations giving someone in here trouble within the last couple weeks. As I recall, they dropped back to march and mtune=core2. --000e0ce02e26efc9d104b26b8ef8 Content-Type: text/html; charset=UTF-8 Content-Transfer-Encoding: quoted-printable


On Nov 23, 2011 1:50 PM, "Stefan G. Weichinger" <lists@xunil.at> wrote:
>
> Am 23.11.2011 18:05, schrieb Michael Mol:
> > On Wed, Nov 23, 2011 at 11:55 AM, Stefan G. Weichinger <lists@xunil.at> wrote:
> >>
> >> Ad CFLAGS for i7-2600:
> >>
> >> Is that too much ricer-style? =C2=A0-->
> >>
> >> ### gcc -march=3Dnative -E -v - </dev/null 2>&1 | s= ed -n 's/.* -v - //p'
> >>
> >> CFLAGS=3D"-O2 -D_FORTIFY_SOURCE=3D2 -march=3Dcore2 -mcx1= 6 -msahf -maes
> >> -mpclmul -mpopcnt -mavx --param l1-cache-size=3D32 --param > >> l1-cache-line-size=3D64 --param l2-cache-size=3D8192 -mtune= =3Dgeneric"
> >>
> >> gcc-4.5.3-r1 ...
> >
> > That's equivalent to, what, "-O2 -D_FORTIFY_SOURCE=3D2 -= march=3Dnative" ?
> > I don't see anything to complain about.
>
> Ok, thanks.
>
> compiled gcc-4.6.2 (hey, it only takes ~16min now!):
>
> # /usr/bin/gcc-4.6.2 =C2=A0-march=3Dnative -E -v - </dev/null 2>= &1 | sed -n
> 's/.* -v - //p'
> -D_FORTIFY_SOURCE=3D2 -march=3Dcorei7-avx -mcx16 -msahf -mno-movbe -ma= es
> -mpclmul -mpopcnt -mno-abm -mno-lwp -mno-fma -mno-fma4 -mno-xop -mno-b= mi
> -mno-tbm -mavx -msse4.2 -msse4.1 --param l1-cache-size=3D32 --param > l1-cache-line-size=3D64 --param l2-cache-size=3D8192 -mtune=3Dcorei7-a= vx
>
> If I am "feeling adventurous" (Neil) I could now try my luck= by choosing
> this nice new compiler w/ options. Maybe *after* a nice backup.
>
> > Actually, it's pretty interesting seeing what that processor = comes
> > down to for -march=3Dnative. I wish there were a database of proc= essors
> > and their decomposed compiler tuning flags for comparison. That w= ould
> > be *very* interesting, from the standpoint of proc shopping and > > looking at the evolution of CPUs.
>
> Wouldn't a wiki-page do the trick? gentoo-wiki ...
>
> S
>

ISTR gcc's i7 optimizations giving someone in here trouble within th= e last couple weeks. As I recall, they dropped back to march and mtune=3Dco= re2.

--000e0ce02e26efc9d104b26b8ef8--