From mboxrd@z Thu Jan 1 00:00:00 1970 From: Federico Beffa Subject: Re: forcing local build from a package definition Date: Sun, 26 Oct 2014 09:34:33 +0100 Message-ID: References: <87ppdhxpl7.fsf@gmail.com> <87r3xvpz1v.fsf@gnu.org> Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: quoted-printable Return-path: Received: from eggs.gnu.org ([2001:4830:134:3::10]:52777) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1XiJHY-0008LA-4U for guix-devel@gnu.org; Sun, 26 Oct 2014 04:34:37 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1XiJHX-0002YT-4j for guix-devel@gnu.org; Sun, 26 Oct 2014 04:34:36 -0400 In-Reply-To: <87r3xvpz1v.fsf@gnu.org> List-Id: "Development of GNU Guix and the GNU System distribution." List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: guix-devel-bounces+gcggd-guix-devel=m.gmane.org@gnu.org Sender: guix-devel-bounces+gcggd-guix-devel=m.gmane.org@gnu.org To: =?UTF-8?Q?Ludovic_Court=C3=A8s?= Cc: Guix-devel On Sun, Oct 26, 2014 at 12:25 AM, Ludovic Court=C3=A8s wrote= : > I think ATLAS should do like GMP, libc, etc.: build all the possible > variants, and then use IFUNC or a similar mechanism to select the right > variant at load time. I would like to highlight two points: 1. there are many variants (see below). 2. even for the same configure time configuration, the build phase does not necessarily produce identical libraries all the time. The build phase tries many variants of algorithms, times them and picks the best performing one. The result may be different depending on the overall performance of the system such as on memory access time, and several other factors. Here the list of configure time variants: $ ./xprint_enums Architectural enums (Config's enum MACHTYPE): 0 =3D 'UNKNOWN' 1 =3D 'POWER3' 2 =3D 'POWER4' 3 =3D 'POWER5' 4 =3D 'PPCG4' 5 =3D 'PPCG5' 6 =3D 'POWER6' 7 =3D 'POWER7' 8 =3D 'POWERe6500' 9 =3D 'IBMz9' 10 =3D 'IBMz10' 11 =3D 'IBMz196' 12 =3D 'x86x87' 13 =3D 'x86SSE1' 14 =3D 'x86SSE2' 15 =3D 'x86SSE3' 16 =3D 'P5' 17 =3D 'P5MMX' 18 =3D 'PPRO' 19 =3D 'PII' 20 =3D 'PIII' 21 =3D 'PM' 22 =3D 'CoreSolo' 23 =3D 'CoreDuo' 24 =3D 'Core2Solo' 25 =3D 'Core2' 26 =3D 'Corei1' 27 =3D 'Corei2' 28 =3D 'Corei3' 29 =3D 'Atom' 30 =3D 'P4' 31 =3D 'P4E' 32 =3D 'Efficeon' 33 =3D 'K7' 34 =3D 'HAMMER' 35 =3D 'AMD64K10h' 36 =3D 'AMDLLANO' 37 =3D 'AMDDOZER' 38 =3D 'AMDDRIVER' 39 =3D 'UNKNOWNx86' 40 =3D 'IA64Itan' 41 =3D 'IA64Itan2' 42 =3D 'USI' 43 =3D 'USII' 44 =3D 'USIII' 45 =3D 'USIV' 46 =3D 'UST1' 47 =3D 'UST2' 48 =3D 'UnknownUS' 49 =3D 'MIPSR1xK' 50 =3D 'MIPSICE9' 51 =3D 'ARMv7' Operating System enums (Config's enum OSTYPE): 0 =3D 'UNKNOWN' 1 =3D 'Linux' 2 =3D 'SunOS' 3 =3D 'SunOS4' 4 =3D 'OSF1' 5 =3D 'IRIX' 6 =3D 'AIX' 7 =3D 'Win9x' 8 =3D 'WinNT' 9 =3D 'Win64' 10 =3D 'HPUX' 11 =3D 'FreeBSD' 12 =3D 'OSX' Compiler integer defines: 0 =3D 'ICC' 1 =3D 'SMC' 2 =3D 'DMC' 3 =3D 'SKC' 4 =3D 'DKC' 5 =3D 'XCC' 6 =3D 'GCC' 7 =3D 'F77' ISA extensions are combined by adding their values together (bitvector): none: 1 VSX: 2 AltiVec: 4 AVXMAC: 8 AVXFMA4: 16 AVX: 32 SSE3: 64 SSE2: 128 SSE1: 256 3DNow: 512 NEON: 1024 > That doesn=E2=80=99t answer your initial question, though. For now, I th= ink > it=E2=80=99s OK to let it do its configure-time tuning and add a statemen= t in > the description about substitutes, unless other packages depend on it > (in the former case, people would be installing it explicitly, so they > would most likely know what they=E2=80=99re doing.) I'm preparing numpy/scipy and they depend on ATLAS. But initially it will be independent. Regards, Fede