From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mga11.intel.com (mga11.intel.com [192.55.52.93]) by dpdk.org (Postfix) with ESMTP id 55C0C6A95 for ; Thu, 11 Dec 2014 02:45:01 +0100 (CET) Received: from fmsmga003.fm.intel.com ([10.253.24.29]) by fmsmga102.fm.intel.com with ESMTP; 10 Dec 2014 17:44:59 -0800 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="4.97,862,1389772800"; d="scan'208";a="427589377" Received: from kmsmsx151.gar.corp.intel.com ([172.21.73.86]) by FMSMGA003.fm.intel.com with ESMTP; 10 Dec 2014 17:34:14 -0800 Received: from shsmsx152.ccr.corp.intel.com (10.239.6.52) by KMSMSX151.gar.corp.intel.com (172.21.73.86) with Microsoft SMTP Server (TLS) id 14.3.195.1; Thu, 11 Dec 2014 09:44:57 +0800 Received: from shsmsx101.ccr.corp.intel.com ([169.254.1.110]) by SHSMSX152.ccr.corp.intel.com ([169.254.6.5]) with mapi id 14.03.0195.001; Thu, 11 Dec 2014 09:44:55 +0800 From: "Qiu, Michael" To: Jay Rolette , "Ananyev, Konstantin" Thread-Topic: [dpdk-dev] [PATCH] Avoid possible memory cpoy when sort hugepages Thread-Index: AQHQFMFwbDvf06mMf0CGakJFSjiBzQ== Date: Thu, 11 Dec 2014 01:44:55 +0000 Message-ID: <533710CFB86FA344BFBF2D6802E60286C9EA4B@SHSMSX101.ccr.corp.intel.com> References: <1418178341-4193-1-git-send-email-michael.qiu@intel.com> <20141210104110.GB10056@bricha3-MOBL3> <533710CFB86FA344BFBF2D6802E60286C9E768@SHSMSX101.ccr.corp.intel.com> <2601191342CEEE43887BDE71AB977258213BEA97@IRSMSX105.ger.corp.intel.com> <2601191342CEEE43887BDE71AB977258213BED0C@IRSMSX105.ger.corp.intel.com> <2601191342CEEE43887BDE71AB977258213BED20@IRSMSX105.ger.corp.intel.com> Accept-Language: en-US Content-Language: en-US X-MS-Has-Attach: X-MS-TNEF-Correlator: x-originating-ip: [10.239.127.40] Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: quoted-printable MIME-Version: 1.0 Cc: "dev@dpdk.org" Subject: Re: [dpdk-dev] [PATCH] Avoid possible memory cpoy when sort hugepages X-BeenThere: dev@dpdk.org X-Mailman-Version: 2.1.15 Precedence: list List-Id: patches and discussions about DPDK List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Thu, 11 Dec 2014 01:45:01 -0000 Archived-At: List-Archive: On 12/11/2014 5:37 AM, Jay Rolette wrote:=0A= > On Wed, Dec 10, 2014 at 12:39 PM, Ananyev, Konstantin <=0A= > konstantin.ananyev@intel.com> wrote:=0A= >=0A= >>> I just got through replacing that entire function in my repo with a cal= l=0A= >> to qsort() from the standard library last night myself. Faster=0A= >>> (although probably not material to most deployments) and less code.=0A= >> If you feel like it is worth it, why not to submit a patch? :)=0A= >=0A= > On Haswell and IvyBridge Xeons, with 128 1G huge pages, it doesn't make a= =0A= > user-noticeable difference in the time required for=0A= > rte_eal_hugepage_init(). The reason I went ahead and checked it in my rep= o=0A= > is because:=0A= >=0A= > a) it eats at my soul to see an O(n^2) case for something where qsort() i= s=0A= > trivial to use=0A= > b) we will increase that up to ~232 1G huge pages soon. Likely doesn't=0A= > matter at that point either, but since it was already written...=0A= >=0A= > What *does* chew up a lot of time in init is where the huge pages are bei= ng=0A= > explicitly zeroed in map_all_hugepages().=0A= >=0A= > Removing that memset() makes find_numasocket() blow up, but I was able to= =0A= > do a quick test where I only memset 1 byte on each page. That cut init ti= me=0A= > by 30% (~20 seconds in my test). Significant, but since I'm not entirely= =0A= > sure it is safe, I'm not making that change right now.=0A= >=0A= > On Linux, shared memory that isn't file-backed is automatically zeroed=0A= > before the app gets it. However, I haven't had a chance to chase down=0A= > whether that applies to huge pages or not, much less how hugetlbfs factor= s=0A= > into the equation.=0A= >=0A= > Back to the question about the patch, if you guys are interested in it,= =0A= > I'll have to figure out your patch submission process. Shouldn't be a hug= e=0A= > deal other than the fact that we are on DPDK 1.6 (r2).=0A= =0A= Go ahead and post it :)=0A= =0A= Thanks,=0A= Michael=0A= > Cheers,=0A= > Jay=0A= >=0A= =0A=