From mboxrd@z Thu Jan 1 00:00:00 1970 From: Eric Dumazet Subject: Re: [PATCH] RPS: Sparse connection optimizations Date: Wed, 02 May 2012 06:12:48 +0200 Message-ID: <1335931968.22133.47.camel@edumazet-glaptop> References: <1335607805-735-1-git-send-email-dczhu@mips.com> <1335614156.2900.66.camel@edumazet-glaptop> <4FA0B337.1070401@mips.com> Mime-Version: 1.0 Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: 7bit Cc: davem@davemloft.net, therbert@google.com, netdev@vger.kernel.org To: Deng-Cheng Zhu Return-path: Received: from mail-wg0-f44.google.com ([74.125.82.44]:58290 "EHLO mail-wg0-f44.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1750999Ab2EBEMx (ORCPT ); Wed, 2 May 2012 00:12:53 -0400 Received: by wgbdr13 with SMTP id dr13so196809wgb.1 for ; Tue, 01 May 2012 21:12:52 -0700 (PDT) In-Reply-To: <4FA0B337.1070401@mips.com> Sender: netdev-owner@vger.kernel.org List-ID: On Wed, 2012-05-02 at 12:08 +0800, Deng-Cheng Zhu wrote: > Unfortunately, on the Malta platform, NIC irqs are not suitable for SMP > IRQ affinity -- they are based on XT-PIC. However, I did do some RPS > tests where different CPU masks were assigned to the 2 NICs. And the > throughput *WAS* better than that of assigning the same mask to NICs. > But the problem addressed in this patch *STILL* exists -- hash indexing > causes imbalance across CPUs in the case of sparse connections. You mean your two NIC irqs are handled by CPU0 and this cant be changed ? That really is bad.