From owner-sc22wg5+sc22wg5-dom8=www.open-std.org@open-std.org Thu May 16 22:09:31 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 BEF0A356E40; Thu, 16 May 2013 22:09:31 +0200 (CEST) Delivered-To: sc22wg5@open-std.org Received: from exprod6og123.obsmtp.com (exprod6og123.obsmtp.com [64.18.1.241]) by www.open-std.org (Postfix) with ESMTP id 40FBE35694E for ; Thu, 16 May 2013 22:09:14 +0200 (CEST) Received: from CFWEX01.americas.cray.com ([136.162.34.11]) (using TLSv1) by exprod6ob123.postini.com ([64.18.5.12]) with SMTP ID DSNKUZU84Z+NJ4vUj/4QhyVvfBhqzc+he1m5@postini.com; Thu, 16 May 2013 13:09:30 PDT Received: from fortran.us.cray.com (172.31.19.200) by CFWEX01.americas.cray.com (172.30.88.25) with Microsoft SMTP Server id 14.2.342.3; Thu, 16 May 2013 15:09:05 -0500 Message-ID: <51953DAA.6020203@cray.com> Date: Thu, 16 May 2013 15:12:26 -0500 From: Bill Long Reply-To: Organization: Cray Inc. User-Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.6; rv:17.0) Gecko/20130509 Thunderbird/17.0.6 MIME-Version: 1.0 To: "N.M. Maclaren" CC: , Mark Batty , "Lionel, Steve" , Lorri Menard , Daniel C Chen Subject: Re: Existing support for uses of atomics in Fortran coarray codes References: <51929A9A.1090901@cray.com> In-Reply-To: Content-Type: text/plain; charset="ISO-8859-1"; format=flowed Content-Transfer-Encoding: 7bit Sender: owner-sc22wg5@open-std.org Precedence: bulk On 5/15/13 9:17 AM, N.M. Maclaren wrote: > Thank you very much. We have a lot of latitude in what we can specify, > but we do need to ensure that both programmers and implementors know > what is actually required. And we may as well be guided by existing > implementations. One other area of implementation might be relevant. We currently have ATOMIC_DEFINE and ATOMIC_REF intrinsics. How those are implemented can interact with the new atomics. For the Cray implementation, we use the following: ATOMIC_REF --> atomic_fetch_and_add (target, 0) An equally viable alternative would be atomic_fetch_and_or (target, 0). ATOMIC_DEFINE --> atomic_and_xor (target, 0, newval) I had proposed the and_xor form as an additional user-level intrinsic. But it did not get the votes needed. However, for things like this it is very valuable. And, of course, the implementation can still use it internally without it being available to users. By using atomic operations to implement these intrinsics, we ensure that they are operating in the same way as other atomics, and can be lumped together in the same memory model description. Cheers, Bill > > Regards, > Nick Maclaren. > > > > > -- Bill Long longb@cray.com Fortran Technical Support & voice: 651-605-9024 Bioinformatics Software Development fax: 651-605-9142 Cray Inc./Cray Plaza, Suite 210/380 Jackson St./St. Paul, MN 55101