From mboxrd@z Thu Jan 1 00:00:00 1970 From: Ricardo Wurmus Subject: Re: R Date: Fri, 5 Feb 2016 13:37:10 +0100 Message-ID: References: <20160204193924.GA8852@debian> <20160205100040.GA4213@debian.eduroam.u-bordeaux.fr> 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]:35380) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1aRfdc-0004C9-CI for guix-devel@gnu.org; Fri, 05 Feb 2016 07:37:25 -0500 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1aRfdZ-0000Px-5g for guix-devel@gnu.org; Fri, 05 Feb 2016 07:37:24 -0500 Received: from sinope.bbbm.mdc-berlin.de ([141.80.25.23]:40466) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1aRfdY-0000OY-T1 for guix-devel@gnu.org; Fri, 05 Feb 2016 07:37:21 -0500 In-Reply-To: <20160205100040.GA4213@debian.eduroam.u-bordeaux.fr> 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: Andreas Enge Cc: guix-devel@gnu.org Andreas Enge writes: > I did get a reply, but not using R, I do not quite know what to make of= it: > https://bugs.r-project.org/bugzilla/show_bug.cgi?id=3D16697 > It would be nice if someone using R could help me out; if you send me a= n > ssh key, I could give you access to an arm machine. > > To start, a probably simple question: Why do we use "--with-blas=3Dopen= blas"? I did this when we switched from ATLAS to OpenBLAS (for substitutability reasons). It is possible I have misunderstood the meaning of this configure flag. Here=E2=80=99s the installation manual on building with a specific BLAS implementation: https://cran.r-project.org/doc/manuals/r-release/R-admin.html#BLAS According to this section, =E2=80=9CR can make use of enhanced BLAS (Basic Linear Algebra Subpro= grams, http://www.netlib.org/blas/faq.html) routines. However, these have to be explicitly requested at configure time: R provides an internal BLAS which is well-tested and will be adequate for most uses of R.=E2= =80=9D There=E2=80=99s a separate section for OpenBLAS, which also mentions that= using a =E2=80=9Cshared BLAS=E2=80=9D is a preferred alternative to linking wit= h a particular BLAS implementation. We already pass =E2=80=9C--enable-BLAS-shlib=E2=80=9D= , so we actually do use a shared BLAS, IIUC. I guess we could drop the =E2=80=9C--with-blas=E2=80=9D flag and see if t= hat helps. I remember struggling with this a lot in the past. I=E2=80=99ll try buildi= ng R again without that flag and see if it still works. I=E2=80=99m sorry if that=E2=80=99s what=E2=80=99s responsible for the fa= ilures, but it=E2=80=99s pretty hard for me to understand how linking with a high-performance BLAS implementation could have caused these build/test failures on other platforms. I=E2=80=99ll report back when I have a successful build or an interesting failure to share. ~~ Ricardo