From owner-freebsd-ports@freebsd.org Sat Jul 2 18:33:42 2016 Return-Path: Delivered-To: freebsd-ports@mailman.ysv.freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:1900:2254:206a::19:1]) by mailman.ysv.freebsd.org (Postfix) with ESMTP id 8FDE3B8FCBC for ; Sat, 2 Jul 2016 18:33:42 +0000 (UTC) (envelope-from korekontakt@gmail.com) Received: from mailman.ysv.freebsd.org (unknown [127.0.1.3]) by mx1.freebsd.org (Postfix) with ESMTP id 69DD12283 for ; Sat, 2 Jul 2016 18:33:42 +0000 (UTC) (envelope-from korekontakt@gmail.com) Received: by mailman.ysv.freebsd.org (Postfix) id 69205B8FCBB; Sat, 2 Jul 2016 18:33:42 +0000 (UTC) Delivered-To: ports@mailman.ysv.freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:1900:2254:206a::19:1]) by mailman.ysv.freebsd.org (Postfix) with ESMTP id 68C26B8FCBA for ; Sat, 2 Jul 2016 18:33:42 +0000 (UTC) (envelope-from korekontakt@gmail.com) Received: from mail-qt0-x22e.google.com (mail-qt0-x22e.google.com [IPv6:2607:f8b0:400d:c0d::22e]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (Client CN "smtp.gmail.com", Issuer "Google Internet Authority G2" (verified OK)) by mx1.freebsd.org (Postfix) with ESMTPS id 1E8DA2282; Sat, 2 Jul 2016 18:33:42 +0000 (UTC) (envelope-from korekontakt@gmail.com) Received: by mail-qt0-x22e.google.com with SMTP id m2so72239407qtd.1; Sat, 02 Jul 2016 11:33:42 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20120113; h=mime-version:in-reply-to:references:from:date:message-id:subject:to :cc; bh=JvJBps+ELPrJIrSofI0dZkXif5d10lE8QXuQeKKlWR4=; b=N6gFkRHPQrFQDQakSLpFYdly/8kPpwbH1qcO0iFtTSNu3h/234at2PwANX+zw71zoZ oiYF/yCf/4WUEXaY7z/AzpM0UCIVUUk6qYO2fwsqMJ676yWpc8VZCE8WtHacNLYuS/QE sPSy4DK+JqSGR88mQde4/VLogAleX4pXRXPdlaCf7tOhaFtZffped76NMpqcuCMDrriO hF23c5Jpzdfbtal65RbG74ZPvZwCBkih+2a7um/SYv5fdDqtJo6nSCJx+2XAG2fC6Jgv 8ny4WoKIGs00uTMVquMwzzjGl++bBGJ+5E7PUjaJGnGsdnFSr+FyJ0z5/JtRL1jQGArZ Y18w== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:in-reply-to:references:from:date :message-id:subject:to:cc; bh=JvJBps+ELPrJIrSofI0dZkXif5d10lE8QXuQeKKlWR4=; b=h0/FBNcNah/1G9NNqlECfyH/HNLPSzSxMY+0Ejzg8dFemTRkkNodsL9fwV89b8uQbE psA8rqBW9j+lJAKNeeaj8DPsK9CWR16acQqSHpfiNWjxBUplcs+syyCXHdY9XRm1QdV7 eTYMdD2dbZSNOWvp31LHhN1s1wb3vBRGvjEoXKIBXU5zmTuqBZl9x3h38YTLGxPhlfND 7Ai8gTv4kZslWE5wij4mTK/OZjnSyXDgkCuEW0OJ1AFnuqa9w8uFWejf6h27JgXtPpaz yDhtJfZVgDBtQwYlJ9vDTVQNVOpN+cQEoigkKv0BJi2gQiKFWGEtdSkcnj+CjqrnnUzi 2W2Q== X-Gm-Message-State: ALyK8tLEQGzY1WWBryY9pQK2qlaRPojdub6MBLvUPYjS+eNpylOrpk9WLR08cq3FJv5aAapEK75xjWyZpCF1hg== X-Received: by 10.237.55.168 with SMTP id j37mr6969514qtb.102.1467484421267; Sat, 02 Jul 2016 11:33:41 -0700 (PDT) MIME-Version: 1.0 Received: by 10.237.52.229 with HTTP; Sat, 2 Jul 2016 11:33:40 -0700 (PDT) In-Reply-To: References: <20160609190710.GT41922@home.opsec.eu> <9f300278-31ab-bc50-1fc7-8e7e6dc44215@toco-domains.de> <8txv-34vw-wny_-_@vfemail.net> From: Raymond Cheung Date: Sun, 3 Jul 2016 02:33:40 +0800 Message-ID: Subject: Re: Torch7 ports (Was: Wxlua / Zbstudio) To: Jan Beich Cc: ports@freebsd.org, Kurt Jaeger , Torsten Zuehlsdorff Content-Type: text/plain; charset=UTF-8 X-Content-Filtered-By: Mailman/MimeDel 2.1.22 X-BeenThere: freebsd-ports@freebsd.org X-Mailman-Version: 2.1.22 Precedence: list List-Id: Porting software to FreeBSD List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Sat, 02 Jul 2016 18:33:42 -0000 I'm using FreeBSD 11.0-ALPHA5 to test. If I use clang/clang++ with the official distro, then nn.test() are passed but torch.test() are error/failed. If I use gcc/g++ for pkg/torch and others with clang/clang++ (and run exec '/usr/local/bin/lua51' -e ...), then two tests (max and min) of torch.test() are error/failed: max error in torch.max (value) - NaNs BOOL violation condition=false min error in torch.min - NaNs BOOL violation condition=false But I can't require 'nn': install/lib/lua/5.1/ffi.so: Undefined symbol "cpow" On Thu, Jun 30, 2016 at 9:19 AM, Raymond Cheung wrote: > Hi all, > > I tried the Jan's git ports. However, I got 21 errors out of 127 torch > tests. It said FFI can't point to some structures. Also, I can require nn > even it was installed via luarocks. It said the tester suite is missing. > > Are the blas finding codes located at math/TH? Thanks. > > Raymond > On Jun 24, 2016 17:12, "Raymond Cheung" wrote: > >> Hi Jan and Torsten, >> >> Thanks a lot for your help. >> >> I will try it later after taking a break. As I lack knowledge on C/C++, I >> spent a month to retry many ways on FreeBSD 10/11. I feel tried. >> Fortunately, I got some clues. >> >> During this month, I also learnt a lot on FreeBSD. As least I can build >> and install new world/kernel from GhostBSD 10.3 to 11 Alpha 4. >> >> Thanks again for your help. >> >> Raymond >> On Jun 24, 2016 06:41, "Jan Beich" wrote: >> >>> Torsten Zuehlsdorff writes: >>> >>> > Hello Raymond, >>> > >>> >> OpenBlas (make config; # with OpenMP option), OpenMP, Lapack & ++, >>> GotoBlas >>> >> are installed. Header files of OpenBlas is also included to >>> >> $CMAKE_LIBRARY_PATH. However, I still got the same error message, >>> missing >>> >> lapack. >>> > >>> > There are various variables to set to specify where to look vor the >>> > libs, like lapack. >>> > >>> > Sadly i'm out of time. Tomorrow i'm heading into vacation. When back i >>> > will come back to your request and try to create a port for >>> > this. Maybe we could success together (with more time). >>> >>> I did some work in the past on Torch7 ports before losing interest[1]. >>> Check math/TH if you want to see how it detects OpenBLAS (default). >>> >>> $ git clone https://github.com/jbeich/freebsd-ports torch-ports >>> $ export PORTSDIR=$PWD/torch-ports >>> $ cd $PORTSDIR/devel/lua-trepl >>> $ make install >>> $ th51 >>> >>> -- >>> [1] Many Torch pkgs rely on luarocks to build and resolve dependencies >>> and some have a hard dependency on luajit. My approach didn't scale >>> as writing such ports often required translating *.rockspec files >>> which can quickly grow into maintenance nightmare, so an infra work >>> had to be done beforehand. >>> >>