From owner-sc22wg5+sc22wg5-dom8=www.open-std.org@open-std.org Wed May 15 16:28:11 2013 Return-Path: X-Original-To: sc22wg5-dom8 Delivered-To: sc22wg5-dom8@www.open-std.org Received: by www.open-std.org (Postfix, from userid 521) id AE1AF356E91; Wed, 15 May 2013 16:28:11 +0200 (CEST) Delivered-To: sc22wg5@open-std.org Received: from ppsw-42.csi.cam.ac.uk (ppsw-mx-f.csi.cam.ac.uk [131.111.8.149]) by www.open-std.org (Postfix) with ESMTP id 6F442356E17 for ; Wed, 15 May 2013 16:27:56 +0200 (CEST) X-Cam-AntiVirus: no malware found X-Cam-ScannerInfo: http://www.cam.ac.uk/cs/email/scanner/ Received: from hermes-2.csi.cam.ac.uk ([131.111.8.54]:46289) by ppsw-42.csi.cam.ac.uk (smtp.hermes.cam.ac.uk [131.111.8.159]:25) with esmtpa (EXTERNAL:nmm1) id 1UccgM-0004rU-6g (Exim 4.80_167-5a66dd3) (return-path ); Wed, 15 May 2013 15:27:54 +0100 Received: from prayer by hermes-2.csi.cam.ac.uk (hermes.cam.ac.uk) with local (PRAYER:nmm1) id 1UccgM-0003mz-0h (Exim 4.72) (return-path ); Wed, 15 May 2013 15:27:54 +0100 Received: from [46.208.117.27] by webmail.hermes.cam.ac.uk with HTTP (Prayer-1.3.5); 15 May 2013 15:27:54 +0100 Date: 15 May 2013 15:27:54 +0100 From: "N.M. Maclaren" To: Tom Clune , "sc22wg5@open-std.org" Subject: Re: (j3.2006) (SC22WG5.4995) Existing support for uses of atomics in Fortran coarray codes Message-ID: In-Reply-To: <51938B16.6070903@cray.com> References: <20130514201252.6BDAE356E8A@www.open-std.org> <20130514212423.4DB6D356E8B@www.open-std.org> <51938B16.6070903@cray.com> X-Mailer: Prayer v1.3.5 Mime-Version: 1.0 Content-Type: text/plain; format=flowed; charset=ISO-8859-1 Sender: owner-sc22wg5@open-std.org Precedence: bulk List pruned heavily. On May 15 2013, Bill Long wrote: >On 5/14/13 4:01 PM, Tom Clune wrote: >> >> If the sendcount/recvcount are computable in advance, is there still a >> measurable performance to the use of co-arrays vs mpi? > >It is certainly possible. If you ship off the data from one image while >others might still be computing you can cut down on the overall >execution time, and reduce congestion in the network. ... Which can also be done in MPI. Whether coarrays is faster will depend on how the program is written and the implementations. As Bill says, it is certainly possible. >You do raise an interesting and important point. Coarrays are *not* >MPI. It is often not a good idea to think in terms of MPI and then try >to translate the MPI calls into similar coarray code. To get the most >out of coarrays, you often benefit from "think different". ... That is true even for fairly simple, serial code - say, between Fortran and C++. But I agree that it is much more true for parallelism; while coarrays and MPI have a lot in common, they are also fairly different. Not as different as MPI and OpenMP, of course. It is nothing to do with the standard, but those of us who teach such things would find it useful to see coding paradigms that have proven to be effective. This would also help when judging whether extensions are needed or not. Regards, Nick Maclaren.