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 SAA15121; Tue, 5 Nov 2002 18:09:04 +0100 (MET)
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 SAA14599 for <caml-list@pauillac.inria.fr>; Tue, 5 Nov 2002 18:09:03 +0100 (MET)
Received: from grisu.bik-gmbh.de (grisu.bik-gmbh.de [217.110.154.194])
	by concorde.inria.fr (8.11.1/8.11.1) with ESMTP id gA5H92510385
	for <caml-list@inria.fr>; Tue, 5 Nov 2002 18:09:02 +0100 (MET)
Received: from bik-gmbh.de (prony.bik-gmbh.de [194.233.237.133])
	by grisu.bik-gmbh.de (8.12.3/8.12.3) with ESMTP id gA5H8xU8056703;
	Tue, 5 Nov 2002 18:09:00 +0100 (CET)
	(envelope-from hars@bik-gmbh.de)
Message-ID: <3DC7FB24.9030209@bik-gmbh.de>
Date: Tue, 05 Nov 2002 18:08:52 +0100
From: Florian Hars <hars@bik-gmbh.de>
User-Agent: Mozilla/5.0 (X11; U; Linux i686; de-AT; rv:1.1) Gecko/20020826
X-Accept-Language: de-de, en-us, en
MIME-Version: 1.0
To: Magesh Kannan <magesh@ittc.ku.edu>
CC: caml-list@inria.fr
Subject: Re: [Caml-list] Currying vs Speed
References: <Pine.LNX.4.21.0211042019440.26138-100000@george.ittc.ku.edu>
Content-Type: multipart/mixed;
 boundary="------------050208080001020206080103"
Sender: owner-caml-list@pauillac.inria.fr
Precedence: bulk

This is a multi-part message in MIME format.
--------------050208080001020206080103
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit

Magesh Kannan wrote:
> Does the invocation (my_func_part 10 20) run any faster than
> (my_func_wrapper 5 10 20)?

It is comparable if you compile to bytecode, and much worse if you compile to 
native code. In that case, you may loose more than a factor of ten:

$ ./bench
my_func:         16.4 sec
my_func_wrapper: 20.2 sec
my_func_part:    17.1 sec
$ ocamlopt -inline 0 -o bench unix.cmxa bench.ml
$ ./bench
my_func:         0.6 sec
my_func_wrapper: 0.8 sec
my_func_part:    2.2 sec
$ ocamlopt -inline 10 -o bench unix.cmxa bench.ml
$ ./bench
my_func:         0.2 sec
my_func_wrapper: 0.2 sec
my_func_part:    2.3 sec

A full application of a function is optimized far better than a partial one.
Especially in the -inline 10 case, the function call is completely optimized 
away for the two fully applied versions.

Yours, Florian Hars.

--------------050208080001020206080103
Content-Type: text/plain;
 name="bench.ml"
Content-Transfer-Encoding: 7bit
Content-Disposition: inline;
 filename="bench.ml"

let my_func arg1 arg2 arg3 =
  let res = arg1 + arg2 + arg3 in
    res

let my_func_wrapper arg1 arg2 arg3 =
  my_func arg1 arg2 arg3

let my_func_part = my_func 5


let _ =
  let t0 = Unix.gettimeofday () in
  let five = 5 in
  let r = ref 0 in
  for i = 0 to 10000 do
    for j = 0 to 10000 do
      r := !r + my_func five i j 
    done
  done;
  Printf.printf "my_func:         %.1f sec\n" (Unix.gettimeofday () -. t0);
  let t0 = Unix.gettimeofday () in
  let five = 5 in
  let r = ref 0 in
  for i = 0 to 10000 do
    for j = 0 to 10000 do
      r := !r + my_func_wrapper five i j 
    done
  done;
  Printf.printf "my_func_wrapper: %.1f sec\n" (Unix.gettimeofday () -. t0);
  let t0 = Unix.gettimeofday () in
  let r = ref 0 in
  for i = 0 to 10000 do
    for j = 0 to 10000 do
      r := !r + my_func_part i j
    done
  done;
  Printf.printf "my_func_part:    %.1f sec\n" (Unix.gettimeofday () -. t0)

--------------050208080001020206080103--

-------------------
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