From owner-sc22wg5+sc22wg5-dom8=www.open-std.org@open-std.org Mon Mar 24 17:08:49 2014 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 04435356F43; Mon, 24 Mar 2014 17:08:48 +0100 (CET) Delivered-To: sc22wg5@open-std.org X-Greylist: delayed 444 seconds by postgrey-1.34 at www5.open-std.org; Mon, 24 Mar 2014 17:08:47 CET Received: from exprod6og118.obsmtp.com (exprod6og118.obsmtp.com [64.18.1.233]) by www.open-std.org (Postfix) with ESMTP id EF6103567ED for ; Mon, 24 Mar 2014 17:08:46 +0100 (CET) Received: from CFWEX01.americas.cray.com ([136.162.34.11]) (using TLSv1) by exprod6ob118.postini.com ([64.18.5.12]) with SMTP ID DSNKUzBYjuTwOtGZ9RVhgzrhvtHDIqY07r2x@postini.com; Mon, 24 Mar 2014 09:08:48 PDT Received: from CFWEX02.americas.cray.com (172.30.74.25) by CFWEX01.americas.cray.com (172.30.88.25) with Microsoft SMTP Server (TLS) id 14.2.347.0; Mon, 24 Mar 2014 10:58:14 -0500 Received: from CFWEX01.americas.cray.com ([169.254.1.43]) by cfwex02.americas.cray.com ([169.254.2.226]) with mapi id 14.02.0347.000; Mon, 24 Mar 2014 10:58:13 -0500 From: Bill Long To: "sc22wg5@open-std.org" Subject: Re: (j3.2006) (SC22WG5.5195) [ukfortran] Ballot on draft DTS Thread-Topic: (j3.2006) (SC22WG5.5195) [ukfortran] Ballot on draft DTS Thread-Index: AQHPR2VEBRNX880tfEuxwGLWT509v5rwuMmA Date: Mon, 24 Mar 2014 15:58:13 +0000 Message-ID: References: <20140312154430.1B7899EB083@www.open-std.org> <20140324133031.CC99D35856F@www.open-std.org> In-Reply-To: <20140324133031.CC99D35856F@www.open-std.org> Accept-Language: en-US Content-Language: en-US X-MS-Has-Attach: X-MS-TNEF-Correlator: x-originating-ip: [172.31.21.214] Content-Type: text/plain; charset="Windows-1252" Content-ID: <93F05B40AA0CF4468272C10A6831995C@cray.com> Content-Transfer-Encoding: quoted-printable MIME-Version: 1.0 Sender: owner-sc22wg5@open-std.org Precedence: bulk On Mar 24, 2014, at 8:30 AM, N.M. Maclaren wrote: > This is a WG5 letter ballot on N2007, the fourth draft DTS for TS 18508, > Additional Parallel Features in Fortran. >=20 > Please answer the following question "Is N2007 ready for forwarding to=20 > SC22 as the DTS?" in one of these ways.=20 >=20 >=20 > 3) No, for the following reasons. >=20 >=20 > Regards, > Nick Maclaren. >=20 >=20 >=20 > Because this is a long response, the most serious and difficult points > are: >=20 > Events.2) This TS introduces a facility that is arguably in conflict > with Fortran 2008 page 23:1-2, and is not described as processor 23:1 is a subclause title. 23:2 is the following text: In descriptive text, an equivalent English word is frequently used in place= of a syntactic term. Particular state- I=92m having trouble understanding how this is relevant.=20 > dependent. It is not clear how to fix it, and it should be removed. > At the very least, it should be explicitly specified to be processor- > dependent. >=20 > Collectives.1) This TS introduces a conflict with Fortran 2008's concept > and specification of variable definition context. This is definitely > fixable, but may need significant changes to the collectives. It would be simpler to fix this in 16.6.7.=20 >=20 > General.1 and General.2) This TS should be explicitly removed from > consideration for inclusion in the next Fortran standard (see N1979), > because it will probably be be infeasible to specify a data consistency > model by the deadline for the first CD ballot, and there is very > unlikely to be sufficient implementation and user experience. >=20 It today=92s environment, vendors are usually unwilling to implement langua= ge features until after the standard is passed. Witness how long has taken= for compilers to conform to F2008 and even F2003. There are exceptions, = of course. Cray has implemented both the collectives and atomics from the T= S, because of user pressure. The main experience so far has been bug filin= gs against MPI because the Fortran collectives were significantly faster th= an the MPI ones. (Resolved by replacing the guts of the MPI routines with = calls to the library used by Fortran.) Implementing EVENTS is straightforw= ard (arguably trivial) if atomics are already available, but vendors are wa= iting for the syntax and rules to settle down first. It is unhelpful for = code portability to release features with syntax that ends up changing. Tea= ms are a bigger effort to implement, and vendors are not that keen on spend= ing resources implementing a version before at least a draft of the TS gets= passed the first ballot.=20 =85.. > My understanding is that there are, as yet, only two full and released > implementations of Fortran 2008 coarrays (Cray's and Intel's), and very > little user experience with the latter; IBM's may be imminent, but I > know of no others. In all cases, the implementation works only with the > vendor's own MPI, as far as I know. It seems very unlikely indeed that > there will be significant experience with implementing and using the > facilities in this TS before February 2015.=20 >=20 I=92m a bit unclear what =93works with=94 means here. The Intel implementat= ion relies on MPI. The Cray one does not use any MPI, nor are there any su= ch plans. It would defeat the performance advantage of using coarrays. (Al= though as noted above, there are examples of push-back to improve MPI in th= e face of coarray performance.) The gfortran project plans, last time I = looked, do not include the use of MPI. They do appear to be making progre= ss, and could be the next significant implementation to be available.=20 It is reasonably expected that codes may use both coarrays and MPI. This is= outside the scope of WG5. I=92ve brought up the need for the MPI committe= e to properly define this. The rules that ensure maximum security are fair= ly straightforward, and have been successfully employed for years. (Note t= hat the OpenMP committee is similarly not keeping up with changes in Fortra= n. Members have been reminded of this.) Cheers, Bill =20 Bill Long = longb@cray.com Fortran Technical Suport & voice: 651-60= 5-9024 Bioinformatics Software Development fax: 651-605-9142 Cray Inc./ Cray Plaza, Suite 210/ 380 Jackson St./ St. Paul, MN 55101