From mboxrd@z Thu Jan  1 00:00:00 1970
Received: from eggs.gnu.org ([2001:4830:134:3::10]:54864)
	by lists.gnu.org with esmtp (Exim 4.71)
	(envelope-from <agraf@suse.de>) id 1Xvaeb-00087r-UF
	for qemu-devel@nongnu.org; Mon, 01 Dec 2014 18:45:25 -0500
Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71)
	(envelope-from <agraf@suse.de>) id 1XvaeQ-0000bP-1i
	for qemu-devel@nongnu.org; Mon, 01 Dec 2014 18:45:17 -0500
Received: from cantor2.suse.de ([195.135.220.15]:48638 helo=mx2.suse.de)
	by eggs.gnu.org with esmtp (Exim 4.71)
	(envelope-from <agraf@suse.de>) id 1XvaeP-0000WL-ST
	for qemu-devel@nongnu.org; Mon, 01 Dec 2014 18:45:06 -0500
Message-ID: <547CFD7E.1080409@suse.de>
Date: Tue, 02 Dec 2014 00:45:02 +0100
From: Alexander Graf <agraf@suse.de>
MIME-Version: 1.0
References: <8E7B9828-1AB9-45A2-BDB5-40518B768BC1@greensocs.com>
	<87tx1faxvk.fsf@fimbulvetr.bsc.es>
In-Reply-To: <87tx1faxvk.fsf@fimbulvetr.bsc.es>
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: quoted-printable
Subject: Re: [Qemu-devel] Update on TCG Multithreading
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: Mark Burton <mark.burton@greensocs.com>, qemu-devel <qemu-devel@nongnu.org>, Bastian Koppelmann <kbastian@mail.uni-paderborn.de>, pavel Dovgaluk <Pavel.Dovgaluk@ispras.ru>, =?UTF-8?B?S09OUkFEIEZyw6lkw6lyaWM=?= <fred.konrad@greensocs.com>, Peter Maydell <peter.maydell@linaro.org>, "Dr. David Alan Gilbert" <dgilbert@redhat.com>, =?UTF-8?B?QWxleCBCZW5uw6k=?= =?UTF-8?B?ZQ==?= <alex.bennee@linaro.org>, Paolo Bonzini <pbonzini@redhat.com>



On 01.12.14 22:00, Llu=C3=ADs Vilanova wrote:
> Mark Burton writes:
>=20
>> All - first a huge thanks for those who have contributed, and those wh=
o have
>> expressed an interest in helping out.
>=20
>> One issue I=E2=80=99d like to see more opinions on is the question of =
a cache per core,
>> or a shared cache.
>> I have heard anecdotal evidence that a shared cache gives a major perf=
ormance
>> benefit=E2=80=A6.
>> Does anybody have anything more concrete?
>> (of course we will get numbers in the end if we implement the hybrid s=
cheme as
>> suggested in the wiki - but I=E2=80=99d still appreciate any feedback)=
.
>=20
> I think it makes sense to have a per-core pointer to a qom TCGCacheClas=
s. That
> can then have its own methods for working with updates, making it much =
simpler
> to work with different implementations, like completely avoiding locks =
(per-cpu
> cache) or a hybrid approach like the one described in the wiki.

I don't think you want to have indirect function calls in the fast path ;=
).


Alex