From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([140.186.70.92]:42989) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1QMeAb-0000QH-PE for qemu-devel@nongnu.org; Wed, 18 May 2011 06:40:02 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1QMeAa-0005YR-Hf for qemu-devel@nongnu.org; Wed, 18 May 2011 06:40:01 -0400 Received: from mail-yx0-f173.google.com ([209.85.213.173]:37504) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1QMeAa-0005YN-CP for qemu-devel@nongnu.org; Wed, 18 May 2011 06:40:00 -0400 Received: by yxk8 with SMTP id 8so558394yxk.4 for ; Wed, 18 May 2011 03:39:59 -0700 (PDT) MIME-Version: 1.0 In-Reply-To: <1305661431-21289-1-git-send-email-jvrao@linux.vnet.ibm.com> References: <1305661431-21289-1-git-send-email-jvrao@linux.vnet.ibm.com> Date: Wed, 18 May 2011 11:39:59 +0100 Message-ID: From: Stefan Hajnoczi Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Subject: Re: [Qemu-devel] [V2 0/25] Async threading for VirtFS using glib threads & coroutines. List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: "Venkateswararao Jujjuri (JV)" Cc: aliguori@us.ibm.com, qemu-devel@nongnu.org, stefanha@linux.vnet.ibm.com On Tue, May 17, 2011 at 8:43 PM, Venkateswararao Jujjuri (JV) wrote: I sent comments on individual patches. This is turning out nice, many of the PDU handlers are now straight-line code that can be followed easily. > o Redesigned to use bh as per Stefan's suggestion. > =A0This made the code very simple but is little less performant compared = to V1. > =A0Anthony suggested to go-in with cleaner code and design (This version)= and > =A0deal with the performance later. > =A0Just to put in perspective: > =A0Sequential Writes of creating 1GB files using ffsb > =A0o Write size 8k > =A0With bh: 66.9MB/sec > =A0Without bh (marshalling routines): 74.9 MB/sec > > =A0o Write size 128k > =A0With bh: 117MB/sec > =A0Without bh (marshalling routines): 122MB/sec It would be interesting to hack in a post-yield callback which is invoked after yield transfers control back to the calling coroutine. This avoids setting up a BH and notifying the iothread to invoke it. I'm just suggesting this as an experiment to see if the BH mechanism causes the performance degradation, not as a real solution. Stefan