From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: weis
Received: (from weis@localhost) by pauillac.inria.fr (8.7.6/8.7.3) id OAA19902 for caml-redistribution@pauillac.inria.fr; Mon, 6 Mar 2000 14:09:55 +0100 (MET)
Resent-Message-Id: <200003061309.OAA19902@pauillac.inria.fr>
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 NAA12630 for <caml-list@pauillac.inria.fr>; Fri, 3 Mar 2000 13:45:55 +0100 (MET)
Received: from tobago.inria.fr (tobago.inria.fr [128.93.8.21])
	by concorde.inria.fr (8.8.7/8.8.7) with ESMTP id NAA13740;
	Fri, 3 Mar 2000 13:45:54 +0100 (MET)
Received: (from doligez@localhost) by tobago.inria.fr (8.6.10/8.6.6) id NAA12637; Fri, 3 Mar 2000 13:45:51 +0100
Date: Fri, 3 Mar 2000 13:45:51 +0100
From: Damien Doligez <Damien.Doligez@inria.fr>
Message-Id: <200003031245.NAA12637@tobago.inria.fr>
To: caml-list@inria.fr, debourse@email.enst.fr
Subject: Re:  Lazy evaluation & performance
Resent-From: weis@pauillac.inria.fr
Resent-Date: Mon, 6 Mar 2000 14:09:55 +0100
Resent-To: caml-redistribution@pauillac.inria.fr

>From: Benoit de Boursetty <debourse@email.enst.fr>

>Has anybody done benchmarks to eval the cost of lazy computation
>encapsulation, in terms of time, memory, garbage collection? I have no
>idea of how this is implemented...

In all the versions of O'Caml so far,
    lazy (some expression)
is exactly equivalent to
    Pervasives.ref (Lazy.Delayed (fun () -> (some expression)))
.

>Could anybody give me a hint about the order of magnitude of L?

Memory: two one-field blocks plus a closure (arbitrarily big,
depending on the free variables of your expression).

Time: whatever it takes to allocate the above, i.e. not much.
Certainly less than a dozen flops.  Maybe even less than one flop, but
that may depend on the compiler (bytecode or native) and on the
architecture.

-- Damien