From owner-sc22wg5+sc22wg5-dom8=www.open-std.org@open-std.org Fri Jan 18 09:40:01 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 92643356CAB; Fri, 18 Jan 2013 09:40:01 +0100 (CET) Delivered-To: sc22wg5@open-std.org Received: from nag-j.co.jp (nag-j.co.jp [111.68.142.10]) by www.open-std.org (Postfix) with ESMTP id 9635B356BE0 for ; Fri, 18 Jan 2013 09:39:59 +0100 (CET) Received: from Maru6 (218-42-159-105.cust.bit-drive.ne.jp [218.42.159.105]) (authenticated bits=0) by nag-j.co.jp (8.14.5/8.14.5) with ESMTP id r0I6fLwp099739 for ; Fri, 18 Jan 2013 06:41:26 GMT (envelope-from malcolm@nag-j.co.jp) Message-ID: From: "Malcolm Cohen" To: "WG5" References: <20130108205438.50F4E356B56@www.open-std.org> <20130113210632.B1C67356BEA@www.open-std.org> <20130116050422.92E30356BE0@www.open-std.org><20130116191114.A6A75356A2B@www.open-std.org><003301cdf436$a0a80670$e1f81350$@chivers@chiversandbryan.co.uk> <20130117155349.AD211356957@www.open-std.org> In-Reply-To: <20130117155349.AD211356957@www.open-std.org> Subject: Re: [ukfortran] (SC22WG5.4906) (j3.2006) Comment on a comment on the WG5 letterballot on N1947 Date: Fri, 18 Jan 2013 15:41:22 +0900 Organization: =?utf-8?B?5pel5pysTkFH?= MIME-Version: 1.0 Content-Type: text/plain; format=flowed; charset="utf-8"; reply-type=response Content-Transfer-Encoding: 7bit X-Priority: 3 X-MSMail-Priority: Normal Importance: Normal X-Mailer: Microsoft Windows Live Mail 15.4.3555.308 X-MimeOLE: Produced By Microsoft MimeOLE V15.4.3555.308 Sender: owner-sc22wg5@open-std.org Precedence: bulk Ian Chivers wrote: > There have also been processors that have supported > integer overflow detection. Bill Long replied: >Yes, I recall the VAX had this. Interesting, though, that the feature seems to >have not caught on. Seeing as the capability is in the x86 family (which powers a number of Cray designs), it has caught on quite well it seems to me - even on HPC machines. That it is not used by most current compilers speaks, IMO, more of the programming culture promoted by C than of missing hardware; the x86 support does require some extra code to be generated, but then so does subscript error detection and most compilers offer that. >a serious performance hit, of course. But you would do this only for debugging Yes there is a performance hit, though not as big as doing it in software with no hardware support. But "having the capability to detect" is what would be asked, not requiring detection all the time. Whether this should be exposed at the programming level, perhaps by something not entirely unlike John's ENABLE block, is an interesting question; IMO that could be a good idea provided its effects did not propagate down or up i.e. was only within the lexical block. However, interesting though all these discussions are, doing anything about this in F2015 is certainly out of the question so we are getting a bit ahead of ourselves. It might be better considered as a part of a "High Reliability Fortran" package at some indeterminate time in the future though. Cheers, -- ................................Malcolm Cohen, Nihon NAG, Tokyo.