From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <gava@univ-paris12.fr>
X-Spam-Checker-Version: SpamAssassin 3.1.3 (2006-06-01) on yquem.inria.fr
X-Spam-Level: *
X-Spam-Status: No, score=1.0 required=5.0 tests=AWL,SPF_FAIL 
	autolearn=disabled version=3.1.3
X-Original-To: caml-list@yquem.inria.fr
Delivered-To: caml-list@yquem.inria.fr
Received: from discorde.inria.fr (discorde.inria.fr [192.93.2.38])
	by yquem.inria.fr (Postfix) with ESMTP id ABDBEBC6A
	for <caml-list@yquem.inria.fr>; Thu,  8 Feb 2007 00:42:47 +0100 (CET)
Received: from smtp19.orange.fr (smtp19.orange.fr [80.12.242.1])
	by discorde.inria.fr (8.13.6/8.13.6) with ESMTP id l17Ngkv4030574
	for <caml-list@yquem.inria.fr>; Thu, 8 Feb 2007 00:42:47 +0100
Received: from me-wanadoo.net (localhost [127.0.0.1])
	by mwinf1906.orange.fr (SMTP Server) with ESMTP id DDF0D1C000A8
	for <caml-list@yquem.inria.fr>; Thu,  8 Feb 2007 00:42:46 +0100 (CET)
Received: from [192.168.0.3] (ARouen-205-1-1-140.w80-11.abo.wanadoo.fr [80.11.112.140])
	by mwinf1906.orange.fr (SMTP Server) with ESMTP id 8242F1C0008C
	for <caml-list@yquem.inria.fr>; Thu,  8 Feb 2007 00:42:46 +0100 (CET)
X-ME-UUID: 20070207234246533.8242F1C0008C@mwinf1906.orange.fr
Message-ID: <45CA63F5.6020301@univ-paris12.fr>
Date: Thu, 08 Feb 2007 00:42:45 +0100
From: =?ISO-8859-1?Q?Fr=E9d=E9ric_Gava?= <gava@univ-paris12.fr>
Reply-To: gava@univ-paris12.fr
Organization: =?ISO-8859-1?Q?Universit=E9_de_Paris_12=2C_LACL?=
User-Agent: Thunderbird 1.5.0.9 (X11/20070104)
MIME-Version: 1.0
To: caml-list@yquem.inria.fr
Subject: Multiplication of matrix in C and OCaml
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 8bit
X-Miltered: at discorde with ID 45CA63F7.000 by Joe's j-chkmail (http://j-chkmail . ensmp . fr)!
X-Spam: no; 0.00; gava:01 gava:01 ocaml:01 ocaml:01 -unsafe:01 typedef:01 struct:01 mult:01 mult:01 12.:98 img:98 img:98 rand:98 rand:98 inline:01 

Dear All,

I compared multiplication of matrix in C and OCaml and I was a little 
surprise to see that the following C code (using -O2) is 8 time faster 
than the OCaml one (even with -unsafe).

Anybody have an idea to optimize my OCaml code or know why is there this 
"big" difference ?

many thanks

Frédéric Gava

ps: the difference is the same even when I use a non-polymorphic 
multiplication

*************** C Code ******************

typedef struct Complexe {double re; double img;} complexe;

__inline__ void* add(void *x, void *y)
{
complexe a=((complexe*) x)[0];
complexe b=((complexe*) y)[0];
return (void *) &((complexe){a.re+b.re, a.img+b.img});
}

__inline__ void* mult(void *x, void *y)
{
complexe a=((complexe*) x)[0];
complexe b=((complexe*) y)[0];
return (void *) &((complexe){a.re*b.re-a.img*b.img, a.re*b.img+a.img*b.re});
}

void multiply_generic(int n, void *zero, void* (*add)(void *a, void *b), 
void* (*mult)(void *a, void *b), void *a, void *b, void *c)
{
  register int i,j,k;
  void *tmp;
  void *cc;

   for (i=0; i<n; i++)
    {
     for (j=0;j<n;j++)
      {
       tmp=zero;
       for (k=0;k<n;k++) tmp=add(tmp,(mult(b+(k*n+j),a+(i*n+k))));
       cc=c+(i*n+j);
       cc=tmp;
      }
   }
}

complexe random_complexe(void)
{
  complexe c = {c.re=(double)(rand()%1000), c.img=(double)(rand()%1000)};
  return c;
}


Call this with (and with random matrix).

complexe zero;
zero.re=(double)0;
zero.img=(double)0;
multiply_generic(i,&zero,add,mult,a,b,c);


********************* OCaml Code *******************************

let random_complex n = {Complex.re=Random.float n; im=Random.float n}

let matrixRandom n = Array.init (n*n) (fun _ -> random_complex 1000.)

let multiplication n zero add mul a b c =
   let tmp=ref zero in
   for i=0 to n-1 do
    for j=0 to n-1 do
     tmp:=zero;
     for k=0 to n-1 do
      tmp:= add (!tmp) (mul b.(k*n+j) a.(i*n+k))
     done;
     c.(i*n+j)<-(!tmp)
    done
   done