From mboxrd@z Thu Jan  1 00:00:00 1970
Received: from [61.129.64.66] (helo=mail.og-int.com)
	by linuxtogo.org with smtp (Exim 4.67)
	(envelope-from <yangtf@envision.com.cn>) id 1IPVaC-0000aP-Bc
	for openembedded-devel@lists.openembedded.org;
	Mon, 27 Aug 2007 05:48:08 +0200
Received: from [211.144.100.19]; Mon, 27 Aug 2007 11:46:25 +0800
From: =?UTF-8?Q?=E6=9D=A8=E5=90=8C=E5=B3=B0?= <yangtf@envision.com.cn>
To: openembedded-devel@lists.openembedded.org
Date: Mon, 27 Aug 2007 11:48:27 +0800
Message-Id: <1188186507.2137.17.camel@localhost.localdomain>
Mime-Version: 1.0
X-Mailer: Evolution 2.8.3 (2.8.3-2.fc6) 
X-Content-Filtered-By: Mailman/MimeDel 2.1.9
Subject: I hava an idea to Make OpenEmbedded Run Faster
X-BeenThere: openembedded-devel@lists.openembedded.org
X-Mailman-Version: 2.1.9
Precedence: list
Reply-To: openembedded-devel@lists.openembedded.org
List-Id: Using the OpenEmbedded metadata to build Distributions
	<openembedded-devel.lists.openembedded.org>
List-Unsubscribe: <http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-devel>,
	<mailto:openembedded-devel-request@lists.openembedded.org?subject=unsubscribe>
List-Archive: <http://lists.linuxtogo.org/pipermail/openembedded-devel>
List-Post: <mailto:openembedded-devel@lists.openembedded.org>
List-Help: <mailto:openembedded-devel-request@lists.openembedded.org?subject=help>
List-Subscribe: <http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-devel>,
	<mailto:openembedded-devel-request@lists.openembedded.org?subject=subscribe>
X-List-Received-Date: Mon, 27 Aug 2007 03:48:08 -0000
Content-Type: text/plain
Content-Transfer-Encoding: 7bit

EveryOne: 
      By two week  of using OE , I hava notice that task was topology
sort to a queue .and  Run Task One By One. 
draw on the experience of Design of CPU . we can make a multiply thread
run of it ! 

      I notice that the biggest bottleneck of OE is do_fetch . then we
can use a mutiply thread downloader instead of wget 
      And about task schedule ,use method like this:

     bb_No1 do_fetch   do_patch do_configure do_build do_install
     bb_No2                  do_fetch  do_patch       do_configure
do_build do_install
     bb_No3                                 do_fetch      do_patch
do_configure do_build do_install
     bb_No4
do_fetch  .......

     Idea 1 :    focus on download one file one time ,and make it more
fast use mutiply thread downloader 
     Idea 2 :    make 0ther task of bb files  to run  asynchronously 
     Idea 3 :   deal with task  topology more carefully to make sure
that  the bb file that was dependend was finish before
                    this one start to do_patch (Notice : not before
do_fetch ) 
     Idea 4 :   If topology can't make sure Idea 3 ,we must tell OE how
and when  to wait 

     If the Idea can come true ,The speed can go 3 times faster then
now 

Tongfeng Yang
2007-8-27