From: Evgeniy Polyakov Subject: Re: [PATCH 0/8] Intel I/O Acceleration Technology (I/OAT) Date: Sun, 5 Mar 2006 04:43:25 +0300 Message-ID: <20060305014324.GA20026@2ka.mipt.ru> References: <20060303214036.11908.10499.stgit@gitlost.site> <20060304.134144.122314124.davem@davemloft.net> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: jengelh@linux01.gwdg.de, christopher.leech@intel.com, linux-kernel@vger.kernel.org, netdev@vger.kernel.org Return-path: To: "David S. Miller" Content-Disposition: inline In-Reply-To: <20060304.134144.122314124.davem@davemloft.net> Sender: linux-kernel-owner@vger.kernel.org List-Id: netdev.vger.kernel.org On Sat, Mar 04, 2006 at 01:41:44PM -0800, David S. Miller (davem@davemloft.net) wrote: > From: Jan Engelhardt > Date: Sat, 4 Mar 2006 19:46:22 +0100 (MET) > > > Does this buy the normal standard desktop user anything? > > Absolutely, it optimizes end-node performance. It really depends on how it is used. According to investigation made for kevent based FS AIO reading, get_user_pages() performange graph looks like sqrt() function with plato starting on about 64-80 pages on Xeon 2.4Ghz with 1Gb of ram, while memcopy() is linear, so it can be noticebly slower than copy_to_user() if get_user_pages() is used aggressively, so userspace application must reuse the same, already grabbed buffer for maximum performance, but Intel folks did not provide theirs usage case and any benchmarks as far as I know. -- Evgeniy Polyakov