From mboxrd@z Thu Jan  1 00:00:00 1970
Received: from mail4-relais-sop.national.inria.fr (mail4-relais-sop.national.inria.fr [192.134.164.105])
	by walapai.inria.fr (8.13.6/8.13.6) with ESMTP id p3NJ3KHg015023
	for <caml-list@sympa-roc.inria.fr>; Sat, 23 Apr 2011 21:03:20 +0200
X-IronPort-Anti-Spam-Filtered: true
X-IronPort-Anti-Spam-Result: Ar4BAEghs03UnwdjkGdsb2JhbAClZxQBAQEBCQkNBxQEIcFRhXYEkj0
X-IronPort-AV: E=Sophos;i="4.64,259,1301868000"; 
   d="scan'208";a="93676560"
Received: from relay.pcl-ipout01.plus.net ([212.159.7.99])
  by mail4-smtp-sop.national.inria.fr with ESMTP/TLS/RC4-SHA; 23 Apr 2011 21:03:15 +0200
X-IronPort-Anti-Spam-Filtered: true
X-IronPort-Anti-Spam-Result: AvsEAM4hs01UXebj/2dsb2JhbAClZ3fBSoV2BJI9
Received: from outmx01.plus.net ([84.93.230.227])
  by relay.pcl-ipout01.plus.net with ESMTP; 23 Apr 2011 20:03:14 +0100
Received: from [80.229.123.248] (helo=WinEight)
	 by outmx01.plus.net with esmtp (Exim) id 1QDi6s-00070G-4J; Sat, 23 Apr 2011 20:03:14 +0100
From: "Jon Harrop" <jon@ffconsultancy.com>
To: "'Eray Ozkural'" <examachine@gmail.com>
Cc: <caml-list@inria.fr>
References: <76544177.594058.1303341821437.JavaMail.root@zmbs4.inria.fr>	<4DAFE141.7080003@inria.fr>	<4DAFF442.8000806@lexifi.com>	<799994864.610698.1303412613509.JavaMail.root@zmbs4.inria.fr>	<4DB136FB.6050302@inria.fr>	<1303463512.8429.1344.camel@thinkpad> <BANLkTi=1hkuu9XtMyH+c5RxpM4tvK_ro3A@mail.gmail.com>
In-Reply-To: <BANLkTi=1hkuu9XtMyH+c5RxpM4tvK_ro3A@mail.gmail.com>
Date: Sat, 23 Apr 2011 20:02:46 +0100
Message-ID: <00d801cc01e9$090a9d00$1b1fd700$@ffconsultancy.com>
MIME-Version: 1.0
Content-Type: text/plain;
	charset="US-ASCII"
Content-Transfer-Encoding: 7bit
X-Mailer: Microsoft Outlook 14.0
Thread-Index: AQGysqgJwKmI/txTCyK2eidtAsi+eQFULt0GAef07xcBKFzyqQFeqoamAdEGoYICDnGJ+ZRRVrdA
Content-Language: en-gb
Subject: RE: [Caml-list] Efficient OCaml multicore -- roadmap?

Eray wrote:
> I don't really care what others say, but to prove that this has any
performance value you should do the following:

There is certainly merit in your proposal but it is worth noting up front
that you cannot "prove" anything this way. We can only hope to test
falsifiable hypotheses. For example, if we observe better performance from
Cilk-style parallelism than from MPI then it might be because we suck at
writing MPI code rather than because the Cilk style is inherently more
efficient. A valuable result nonetheless but not technically a proof.

Cheers,
Jon.