From mboxrd@z Thu Jan  1 00:00:00 1970
Received: from eggs.gnu.org ([140.186.70.92]:40703)
	by lists.gnu.org with esmtp (Exim 4.71)
	(envelope-from <laurent.desnogues@gmail.com>) id 1QOrbC-0000NB-1i
	for qemu-devel@nongnu.org; Tue, 24 May 2011 09:24:39 -0400
Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71)
	(envelope-from <laurent.desnogues@gmail.com>) id 1QOrb8-0005b2-3E
	for qemu-devel@nongnu.org; Tue, 24 May 2011 09:24:38 -0400
Received: from mail-qw0-f45.google.com ([209.85.216.45]:40864)
	by eggs.gnu.org with esmtp (Exim 4.71)
	(envelope-from <laurent.desnogues@gmail.com>) id 1QOrb8-0005ax-0y
	for qemu-devel@nongnu.org; Tue, 24 May 2011 09:24:34 -0400
Received: by qwj8 with SMTP id 8so4022831qwj.4
	for <qemu-devel@nongnu.org>; Tue, 24 May 2011 06:24:33 -0700 (PDT)
MIME-Version: 1.0
In-Reply-To: <alpine.DEB.2.02.1105241423250.2267@bulbul>
References: <1306161654-4388-1-git-send-email-batuzovk@ispras.ru>
	<20110523212241.GB5112@hall.aurel32.net>
	<alpine.DEB.2.02.1105241423250.2267@bulbul>
Date: Tue, 24 May 2011 15:24:33 +0200
Message-ID: <BANLkTimhSQFNRq96XYTCWmGQtp_2vg3rzQ@mail.gmail.com>
From: Laurent Desnogues <laurent.desnogues@gmail.com>
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable
Subject: Re: [Qemu-devel] [RFC][PATCH v0 0/8] Improve register allocator
List-Id: <qemu-devel.nongnu.org>
List-Unsubscribe: <https://lists.nongnu.org/mailman/options/qemu-devel>,
	<mailto:qemu-devel-request@nongnu.org?subject=unsubscribe>
List-Archive: <http://lists.nongnu.org/archive/html/qemu-devel>
List-Post: <mailto:qemu-devel@nongnu.org>
List-Help: <mailto:qemu-devel-request@nongnu.org?subject=help>
List-Subscribe: <https://lists.nongnu.org/mailman/listinfo/qemu-devel>,
	<mailto:qemu-devel-request@nongnu.org?subject=subscribe>
To: Kirill Batuzov <batuzovk@ispras.ru>
Cc: zhur@ispras.ru, qemu-devel@nongnu.org, Aurelien Jarno <aurelien@aurel32.net>

On Tue, May 24, 2011 at 1:31 PM, Kirill Batuzov <batuzovk@ispras.ru> wrote:
[...]
> Gathered statistics shows some interesting things too. I've run matrix
> multiplication benchmark (guest - ARM, host - x86, linux-user mode, with
> my patches applied) and here are the results:
>
> spill count =A0 =A0 =A0 =A0 3916
> =A0real spills =A0 =A0 =A0 32
> =A0spills at bb end =A01023
> =A0spills at call:
> =A0 =A0globals =A0 =A0 =A0 =A0 2755
> =A0 =A0iarg passing =A0 =A00
> =A0 =A0call cloobers =A0 106
>
> Real spills are spills generated by register allocator when it runs out
> of registers. =A0They are less than 1% of all spills. =A0Other tests show
> similar behavior.

When you write "host x86", do you mean IA32 or x86_64?
That might change the number of real spills a lot if you meant
x86_64.

> I think any further improvements to register allocator without leveling
> conventions about saving globals at calls and BB ends somehow is
> useless.
>
> Currently we are looking if we can pass some globals on registers
> through basic block boundaries (inside one TB of course).

If by "basic block", you mean BB as implied by TCG br for
instance, I'm not sure all guests will benefit a lot.  If you
mean that you intend on putting several guests BB in a
single TB then I guess you'll have to first collect dynamic
statistics before dynamically switching to grouping BB.


Laurent