From mboxrd@z Thu Jan 1 00:00:00 1970 From: Jens Axboe Subject: Re: Latest vfs scalability patch Date: Tue, 6 Oct 2009 14:51:26 +0200 Message-ID: <20091006125125.GU5216@kernel.dk> References: <20091006064919.GB30316@wotan.suse.de> <20091006101414.GM5216@kernel.dk> <20091006102604.GN5216@kernel.dk> <1254827438.21044.251.camel@laptop> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: Nick Piggin , Linux Kernel Mailing List , linux-fsdevel@vger.kernel.org, Ravikiran G Thirumalai , Linus Torvalds To: Peter Zijlstra Return-path: Received: from brick.kernel.dk ([93.163.65.50]:34487 "EHLO kernel.dk" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754386AbZJFMwD (ORCPT ); Tue, 6 Oct 2009 08:52:03 -0400 Content-Disposition: inline In-Reply-To: <1254827438.21044.251.camel@laptop> Sender: linux-fsdevel-owner@vger.kernel.org List-ID: On Tue, Oct 06 2009, Peter Zijlstra wrote: > On Tue, 2009-10-06 at 12:26 +0200, Jens Axboe wrote: > > > I did a quick perf analysis on that, but only on 8 clients (64 clients > > basically causes perf to shit itself, it's just not functional). > > Even when used with -a so that you profile each cpu? I can imagine the > cacheline contention from the inherited counters to render a 64 thread > machine less than usable? No, that's without -a. I used -a for perf top, and that has some impact on performance while it's running, but millions of miles apart from what it does without it :-) -- Jens Axboe