From mboxrd@z Thu Jan 1 00:00:00 1970 X-Spam-Checker-Version: SpamAssassin 3.4.4 (2020-01-24) on polar.synack.me X-Spam-Level: X-Spam-Status: No, score=-0.3 required=5.0 tests=BAYES_00, REPLYTO_WITHOUT_TO_CC autolearn=no autolearn_force=no version=3.4.4 Path: border1.nntp.dca3.giganews.com!backlog4.nntp.dca3.giganews.com!border4.nntp.dca.giganews.com!border2.nntp.dca.giganews.com!nntp.giganews.com!newspeer1.nac.net!feeder.erje.net!eu.feeder.erje.net!news.stack.nl!aioe.org!.POSTED!not-for-mail From: "Dmitry A. Kazakov" Newsgroups: comp.lang.ada Subject: Re: Very large scale parallelism Date: Sat, 8 Feb 2014 18:26:38 +0100 Organization: cbb software GmbH Message-ID: References: <9c70d590-7f72-4575-b8a5-4fdc496cc502@googlegroups.com> Reply-To: mailbox@dmitry-kazakov.de NNTP-Posting-Host: dRN93LcgZmpMwxQ2TpSF2g.user.speranza.aioe.org Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit X-Complaints-To: abuse@aioe.org User-Agent: 40tude_Dialog/2.0.15.1 X-Notice: Filtered by postfilter v. 0.8.2 X-Original-Bytes: 1840 Xref: number.nntp.dca.giganews.com comp.lang.ada:184726 Date: 2014-02-08T18:26:38+01:00 List-Id: On Sat, 8 Feb 2014 07:14:23 -0800 (PST), mockturtle wrote: > Some of the speakers talked about the need of tools to program computers > with a large number of cores (even millions). Are you sure they mean cores? Millions of cores would bring less than nothing as the memory will be the bottleneck. Speaking about the future of parallelism it IMO lies with transputer-like architectures rather than with multi-cores. Ada looks good with either paradigm, better with multi-cores, though. Considering transputers and massive parallelism, I think that Annex E is a non starter. In the hey day of transputers I fancied the idea of tasks running on separate transputers making entry calls transparently through the links. One problem with that is the model of shared memory Ada entry calls deploy. To make parallelism working on such architectures Ada must better abstract away by-reference/by-value semantics, allowing marshaling working for most objects. -- Regards, Dmitry A. Kazakov http://www.dmitry-kazakov.de