From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mx1.redhat.com (mx1.redhat.com [209.132.183.28]) by dpdk.org (Postfix) with ESMTP id EEBDA5F16 for ; Fri, 7 Dec 2018 17:15:48 +0100 (CET) Received: from smtp.corp.redhat.com (int-mx06.intmail.prod.int.phx2.redhat.com [10.5.11.16]) (using TLSv1.2 with cipher AECDH-AES256-SHA (256/256 bits)) (No client certificate requested) by mx1.redhat.com (Postfix) with ESMTPS id 2B350C049AFF; Fri, 7 Dec 2018 16:15:48 +0000 (UTC) Received: from [10.36.112.52] (ovpn-112-52.ams2.redhat.com [10.36.112.52]) by smtp.corp.redhat.com (Postfix) with ESMTPS id EE7775C228; Fri, 7 Dec 2018 16:15:41 +0000 (UTC) To: dev@dpdk.org, jfreimann@redhat.com, tiwei.bie@intel.com, zhihong.wang@intel.com References: <20181203151036.11293-1-maxime.coquelin@redhat.com> <20181203151036.11293-3-maxime.coquelin@redhat.com> From: Maxime Coquelin Message-ID: <7b39a9cb-9799-e1d7-f3f9-4b29fe1627a2@redhat.com> Date: Fri, 7 Dec 2018 17:15:39 +0100 User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:60.0) Gecko/20100101 Thunderbird/60.2.1 MIME-Version: 1.0 In-Reply-To: <20181203151036.11293-3-maxime.coquelin@redhat.com> Content-Type: text/plain; charset=utf-8; format=flowed Content-Language: en-US Content-Transfer-Encoding: 7bit X-Scanned-By: MIMEDefang 2.79 on 10.5.11.16 X-Greylist: Sender IP whitelisted, not delayed by milter-greylist-4.5.16 (mx1.redhat.com [10.5.110.32]); Fri, 07 Dec 2018 16:15:48 +0000 (UTC) Subject: Re: [dpdk-dev] [PATCH 2/3] net/virtio: merge Rx mergeable and non-mergeable paths X-BeenThere: dev@dpdk.org X-Mailman-Version: 2.1.15 Precedence: list List-Id: DPDK patches and discussions List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Fri, 07 Dec 2018 16:15:49 -0000 On 12/3/18 4:10 PM, Maxime Coquelin wrote: > Signed-off-by: Maxime Coquelin > --- > drivers/net/virtio/virtio_ethdev.c | 5 -- > drivers/net/virtio/virtio_ethdev.h | 3 - > drivers/net/virtio/virtio_rxtx.c | 115 ++--------------------------- > 3 files changed, 7 insertions(+), 116 deletions(-) Intel STV team ran more tests and found a performance regression. I managed to reproduce it, and worked on optimizing the Rx path. I managed to go from -25% to -6% for this use case, but this is still a significant regression so I'll drop this patch. The good news is that the optimization is valid for mergeable buffers case, and it gains 5% so I'll post this optimization instead. Thanks to the STV team for the testing, Maxime