From mboxrd@z Thu Jan  1 00:00:00 1970
Received: (from majordomo@localhost) by pauillac.inria.fr (8.7.6/8.7.3) id XAA23590; Sun, 20 Oct 2002 23:58:16 +0200 (MET DST)
X-Authentication-Warning: pauillac.inria.fr: majordomo set sender to owner-caml-list@pauillac.inria.fr using -f
Received: from concorde.inria.fr (concorde.inria.fr [192.93.2.39]) by pauillac.inria.fr (8.7.6/8.7.3) with ESMTP id XAA23806 for <caml-list@pauillac.inria.fr>; Sun, 20 Oct 2002 23:58:15 +0200 (MET DST)
Received: from pop9.ucdavis.edu (pop9.ucdavis.edu [169.237.105.19])
	by concorde.inria.fr (8.11.1/8.11.1) with ESMTP id g9KLwE509984
	for <caml-list@inria.fr>; Sun, 20 Oct 2002 23:58:14 +0200 (MET DST)
Received: from beech (mail@[128.120.141.217])
	by pop9.ucdavis.edu (8.11.6/8.11.0/IT4.6.2) with ESMTP id g9KLwCt03071
	for <caml-list@inria.fr>; Sun, 20 Oct 2002 14:58:12 -0700 (PDT)
Received: from ijtrotts by beech with local (Exim 3.35 #1 (Debian))
	id 183O5v-0002Ki-00
	for <caml-list@inria.fr>; Sun, 20 Oct 2002 14:58:47 -0700
Date: Sun, 20 Oct 2002 14:58:47 -0700
To: OCaml Mailing List <caml-list@inria.fr>
Subject: Re: [Caml-list] matrix-matrix multiply - O'Caml is 6 times slower than C
Message-ID: <20021020215847.GA8750@beech>
References: <r99u1jjh5k8.fsf@quimby-xp.cs.cornell.edu> <20021020114251.A7642@pauillac.inria.fr> <r99of9pgf2t.fsf@quimby-xp.cs.cornell.edu>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <r99of9pgf2t.fsf@quimby-xp.cs.cornell.edu>
User-Agent: Mutt/1.3.28i
From: Issac Trotts <ijtrotts@ucdavis.edu>
Sender: owner-caml-list@pauillac.inria.fr
Precedence: bulk

> I don't see how to do strength-reduction without introducing references,
> which appears to reduce performance more than is gained.

You might try converting your references to mutable fields.
Here's an example of the performance gain:

% cat ref.ml

let x = ref 1.0 in
let n = int_of_string Sys.argv.(1) in
for i = 1 to n do x := !x +. 1.0 done

% cat ref2.ml

type t = { mutable f:float };;
let x = { f = 1.0 } in
let n = int_of_string Sys.argv.(1) in
for i = 1 to n do x.f <- x.f +. 1.0 done

% ocamlopt -o ref{,.ml}

% ocamlopt -o ref2{,.ml}

% time ./ref 100000000
./ref 100000000  2.51s user 0.00s system 99% cpu 2.515 total

% time ./ref2 100000000
./ref2 100000000  1.54s user 0.01s system 100% cpu 1.542 total

The standard references take 1.6 times as much cpu time in this
case.

Issac
 
-------------------
To unsubscribe, mail caml-list-request@inria.fr Archives: http://caml.inria.fr
Bug reports: http://caml.inria.fr/bin/caml-bugs FAQ: http://caml.inria.fr/FAQ/
Beginner's list: http://groups.yahoo.com/group/ocaml_beginners