From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <benedikt.meurer@googlemail.com>
X-Original-To: caml-list@yquem.inria.fr
Delivered-To: caml-list@yquem.inria.fr
Received: from mail1-relais-roc.national.inria.fr (mail1-relais-roc.national.inria.fr [192.134.164.82])
	by yquem.inria.fr (Postfix) with ESMTP id 31073BC57
	for <caml-list@yquem.inria.fr>; Wed, 15 Dec 2010 11:29:16 +0100 (CET)
X-IronPort-Anti-Spam-Filtered: true
X-IronPort-Anti-Spam-Result: AskAAAAnCE3RVdYtk2dsb2JhbACVeI4kCBYBAgkJCgkRBCCnOIwIAQWOFgEEhUo
X-IronPort-AV: E=Sophos;i="4.59,347,1288566000"; 
   d="scan'208";a="92414992"
Received: from mail-bw0-f45.google.com ([209.85.214.45])
  by mail1-smtp-roc.national.inria.fr with ESMTP; 15 Dec 2010 11:29:15 +0100
Received: by bwz16 with SMTP id 16so2039867bwz.18
        for <caml-list@yquem.inria.fr>; Wed, 15 Dec 2010 02:29:15 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed;
        d=googlemail.com; s=gamma;
        h=domainkey-signature:received:received:content-type:mime-version
         :subject:from:in-reply-to:date:content-transfer-encoding:message-id
         :references:to:x-mailer;
        bh=sI1ssySuJ5IWulQiEjTvY8zxhtujGF9ediafUt3d97M=;
        b=OxWAmhxvBBPAmKxCPCBhoZ2XkRM0bCgfKjfoux/l2yzbvXIBus9LKwk7sMCwQiRDWj
         kQ4ohD97Naso3J6X4zKCj+Da1LpN0STx8izLVbvGXoOWBLSM6abALRkxspzFq3ZpbkgX
         /2/fbY5yhjc8IAmQExaEZR4p6CgWm8p2VVlxY=
DomainKey-Signature: a=rsa-sha1; c=nofws;
        d=googlemail.com; s=gamma;
        h=content-type:mime-version:subject:from:in-reply-to:date
         :content-transfer-encoding:message-id:references:to:x-mailer;
        b=QuSTrctK5sCf5su22VglHGKoM4hcUYKg+i6DUkK9g8mZ3CToY/xt1L+V9JrbA4K9SM
         qu7ujW+uqHVkrD0YcTq5jP6Bp23A8n3Lj1+PlmKQJ3S+PhMDJYlX1RypIqLk5Tjumre9
         UNKIf5uV7WxKe37RI2W2b9zhptnUVgJP1LAig=
Received: by 10.204.63.2 with SMTP id z2mr6631054bkh.66.1292408953933;
        Wed, 15 Dec 2010 02:29:13 -0800 (PST)
Received: from bespin.informatik.uni-siegen.de (bespin.informatik.uni-siegen.de [141.99.131.92])
        by mx.google.com with ESMTPS id a27sm396195bka.10.2010.12.15.02.29.11
        (version=SSLv3 cipher=RC4-MD5);
        Wed, 15 Dec 2010 02:29:12 -0800 (PST)
Content-Type: text/plain; charset=us-ascii
Mime-Version: 1.0 (Apple Message framework v1081)
Subject: Re: Value types (Was: [Caml-list] ocamlopt LLVM support)
From: Benedikt Meurer <benedikt.meurer@googlemail.com>
In-Reply-To: <4D05DCB6.2060100@frisch.fr>
Date: Wed, 15 Dec 2010 11:29:10 +0100
Content-Transfer-Encoding: quoted-printable
Message-Id: <85AB623B-0278-41F0-9C70-8E8814C0A16F@googlemail.com>
References: <036001cb9a0c$725acef0$57106cd0$@com>	<20101212175524.73a8e285@deb0> <9264BEE6-DBAE-4523-93AC-4560615D2AC5@googlemail.com> <4D05DCB6.2060100@frisch.fr>
To: caml-list@yquem.inria.fr
X-Mailer: Apple Mail (2.1081)
X-Spam: no; 0.00; ocamlopt:01 frisch:01 compiler:01 conceptually:01 speedups:01 viewcvs:01 ocaml:01 bytecomp:01 wrote:01 wrote:01 heap:01 caml-list:01 alain:01 alain:01 diff:02 


On Dec 13, 2010, at 09:43 , Alain Frisch wrote:

> On 12/12/2010 08:09 PM, Benedikt Meurer wrote:
>> The boxing involved is relevant, but boxing in general is not the
>> issue. In this special case, the "let nlen, n =3D if..." code =
requires
>> heap allocation, because of the way the pattern is compiled. This =
could
>> be fixed by moving the condition out of the code and using two if's =
to
>> select n/nlen separately (doesn't speed up that much). Fixing the
>> pattern compiler to handle these cases might be interesting for =
general
>> benefit.
>=20
> Instead of duplicating the conditional, you could also push the =
assignments to bound variables down the expression. For instance:
>=20
> let (x, y) =3D if b then (u, v) else (v, u) in ...
>=20
> can be replaced, conceptually, by:
>=20
> let x =3D <dummy> in
> let y =3D <dummy> in
> if b then (x <- u; y <- v) else (x <- v; y <- u);
> ...
>=20
> and similarly when the bound expression is a pattern matching.
>=20
>=20
> I've played with this a few months ago and could observe important =
speedups (27%, 20%) on two micro-benchmarks.
>=20
> The diff is really small:
>=20
> =
http://caml.inria.fr/cgi-bin/viewcvs.cgi/ocaml/branches/inplace_let/byteco=
mp/matching.ml?rev=3D10475&sortby=3Ddate&r2=3D10475&r1=3D10474

Nice. But it would be even better to avoid the dummy, in your example

  let x =3D u in
  let y =3D v in
  if b then x <- v; y <- u

This does not only avoid the dummy, but would also allow lowering to =
"cmovcc" instructions in the backend selector (atleast for x86-32/64).=20=


> -- Alain

Benedikt=