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=-1.9 required=5.0 tests=BAYES_00,FREEMAIL_FROM
	autolearn=unavailable autolearn_force=no version=3.4.4
X-Received: by 10.36.221.144 with SMTP id t138mr8203584itf.37.1519628331104;
        Sun, 25 Feb 2018 22:58:51 -0800 (PST)
X-Received: by 10.157.41.184 with SMTP id n53mr436345otb.11.1519628330993;
 Sun, 25 Feb 2018 22:58:50 -0800 (PST)
Path: 
 eternal-september.org!reader01.eternal-september.org!reader02.eternal-september.org!feeder.eternal-september.org!border1.nntp.ams1.giganews.com!nntp.giganews.com!peer03.ams1!peer.ams1.xlned.com!news.xlned.com!peer02.am4!peer.am4.highwinds-media.com!peer03.iad!feed-me.highwinds-media.com!news.highwinds-media.com!w142no1751510ita.0!news-out.google.com!10ni2625ite.0!nntp.google.com!w142no1751507ita.0!postnews.google.com!glegroupsg2000goo.googlegroups.com!not-for-mail
Newsgroups: comp.lang.ada
Date: Sun, 25 Feb 2018 22:58:50 -0800 (PST)
In-Reply-To: <06efbe02-cdae-4fac-a17d-6d0c1be7848c@googlegroups.com>
Complaints-To: groups-abuse@google.com
Injection-Info: glegroupsg2000goo.googlegroups.com;
 posting-host=2601:191:8303:2100:7466:f44c:da21:40b1;
 posting-account=fdRd8woAAADTIlxCu9FgvDrUK4wPzvy3
NNTP-Posting-Host: 2601:191:8303:2100:7466:f44c:da21:40b1
References: <ef267893-03f7-48d1-8178-d69035e7c5fb@googlegroups.com>
 <a7b3cac4-b108-4941-a498-25d402b3b409@googlegroups.com>
 <p6bku6$s7j$1@dont-email.me>
 <83493d20-7001-405b-8658-8a3f5d6c90fa@googlegroups.com>
 <p6bv74$ac$1@dont-email.me>
 <b819f146-3c07-49ea-bb1a-cfe497e2df3b@googlegroups.com>
 <p6csbl$ls6$1@gioia.aioe.org>
 <a4ccd7ab-59a7-486b-afd9-41737dbdb706@googlegroups.com>
 <06efbe02-cdae-4fac-a17d-6d0c1be7848c@googlegroups.com>
User-Agent: G2/1.0
MIME-Version: 1.0
Message-ID: <b0722120-ea76-4bcc-beab-29bb1c408d22@googlegroups.com>
Subject: Re: GNAT can't vectorize Real_Matrix multiplication from
 Ada.Numerics.Real_Arrays. What a surprise!
From: Robert Eachus <rieachus@comcast.net>
Injection-Date: Mon, 26 Feb 2018 06:58:51 +0000
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
X-Received-Body-CRC: 4233185018
X-Received-Bytes: 2809
Xref: reader02.eternal-september.org comp.lang.ada:50647
Date: 2018-02-25T22:58:50-08:00
List-Id: <comp.lang.ada>

On Monday, February 19, 2018 at 9:31:26 PM UTC-5, Bojan Bozovic=20

> I have looked at ATLAS, however it can't spawn more threads than specifie=
d at compile time, so there's lots of possibility to optimize there, by spa=
wning as many threads as supported at run-time.

Aarrg!  Yes, there is a lot of work that needs to be done.  The intent is t=
hat you run ATLAS on your target environment, then use the best result as y=
our blas library.

But the problem I am fighting with right now is that on the most recent (hi=
gh-end) processors from Intel and AMD, you never want to use as many thread=
s as the hardware tells you are available at run-time.  In fact, it is comm=
on that if you have a processor which supports 8-threads, you want to run f=
our threads on all even or odd numbered threads.  The recent Threadripper a=
nd EPYC CPUs from AMD make it even more complex, as do any multisocket syst=
ems.  Usually you want to split the problem up completely and duplicate the=
 data on each hardware CPU chip.