From owner-freebsd-questions@FreeBSD.ORG Fri Aug 20 13:24:19 2010 Return-Path: Delivered-To: freebsd-questions@FreeBSD.org Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:4f8:fff6::34]) by hub.freebsd.org (Postfix) with ESMTP id 5B01C10656A4 for ; Fri, 20 Aug 2010 13:24:19 +0000 (UTC) (envelope-from bf1783@googlemail.com) Received: from mail-ew0-f54.google.com (mail-ew0-f54.google.com [209.85.215.54]) by mx1.freebsd.org (Postfix) with ESMTP id E51248FC1F for ; Fri, 20 Aug 2010 13:24:18 +0000 (UTC) Received: by ewy26 with SMTP id 26so2381836ewy.13 for ; Fri, 20 Aug 2010 06:24:17 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=googlemail.com; s=gamma; h=domainkey-signature:mime-version:received:received:reply-to:date :message-id:subject:from:to:content-type; bh=DBq/155J22V3YC03ZMlRyBKXU1E9wjdIMRtW+tMXKJs=; b=D0igkEpeY9Tke7TyCskjMdqwEhBfkvgsK5gvKLb+gcUBGMQmUYkwq8VmtGo1iOkfaS XlHZowPQFAQ1aVhF9pnbzuAthmdbeZaXXUZK+8m2k8GYsKIb7+e8p6hoYzAqmojwQ4SW i8AJDEIfaglBzydK+q8AvyV3Tr3YnWGGgqNNU= DomainKey-Signature: a=rsa-sha1; c=nofws; d=googlemail.com; s=gamma; h=mime-version:reply-to:date:message-id:subject:from:to:content-type; b=YmCXP4mOiBJAZ2N7Gh2TCwvsSsehmCcV8R96XwzHaiCajR7uk+vxoGcNCZANgtAh8F z6hoUSOvyGG75fYcj7HHe6jJKiHBkfiyBSqXEhhDb/GCCwLbbHVrOyp4Fyz/gOzyOst7 STtYVfCnh/H1sM574tTonO5g0Z9jveSlzyqQs= MIME-Version: 1.0 Received: by 10.216.74.75 with SMTP id w53mr1170708wed.86.1282310656331; Fri, 20 Aug 2010 06:24:16 -0700 (PDT) Received: by 10.216.183.212 with HTTP; Fri, 20 Aug 2010 06:24:16 -0700 (PDT) Date: Fri, 20 Aug 2010 13:24:16 +0000 Message-ID: From: "b. f." To: freebsd-questions@FreeBSD.org Content-Type: text/plain; charset=ISO-8859-1 Cc: Subject: Re: FreeBSD, GPGPU and OpenCL/CUDA X-BeenThere: freebsd-questions@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list Reply-To: bf1783@gmail.com List-Id: User questions List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Fri, 20 Aug 2010 13:24:19 -0000 ... > MPI is typically dependent on the network not OpenMP. OpenMP 3.0 can > be made more scalable if there's tasks built-into the kernel that can > be cleanly exposed to userland. (Like OpenSolaris + libtask from > Moinak is a good example) If you have a specific set of modifications in mind, then you should bring them up on freebsd-hackers, for example. There are active users of OpenMPI around ( e.g., http://www.bsdcan.org/2010/schedule/events/169.en.html , and I'm sure they would be willing to discuss improvements. >> 2) HPC ready compiler.. (Sorry guys, but LLVM is just >> not production ready for this task and is missing Fortran) > Not ready now, perhaps -- but development there is fairly rapid, and the switch to llvm, if there is to be one, is still some time off. However, later versions of gcc are in ports, there are some discussions regarding the revival of the icc port, and there is work underway to allow users to more easily use alternative compilers and toolchains for the base system as well as for ports: http://www.freebsd.org/news/status/report-2010-01-2010-03.html#Out-of-Tree-Toolchain > >> 3) IB network drivers > >Don't know the status of Infiniband drivers, are there drivers? There is a port of of the Open Fabrics Enterprise Distribution underway: http://svn.freebsd.org/viewvc/base/projects/ofed/ And there is: http://svn.freebsd.org/viewvc/base/head/sys/contrib/rdma/ >> 4) Hardware vendor to deliver a complete solution + support >> (iXsystems?) Well, that's not up to us. All we can do is port more software and encourage people to use it. And "we" includes "you". b.