From mboxrd@z Thu Jan 1 00:00:00 1970 From: Jeremy Fitzhardinge Subject: Re: AES Encryption information Date: Wed, 07 Jul 2010 09:42:33 -0700 Message-ID: <4C34AE79.4010108@goop.org> References: <4C337B5E.6050104@goop.org> <4C3464EF.9080105@invisiblethingslab.com> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <4C3464EF.9080105@invisiblethingslab.com> List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Sender: xen-devel-bounces@lists.xensource.com Errors-To: xen-devel-bounces@lists.xensource.com To: Joanna Rutkowska Cc: xen-devel@lists.xensource.com List-Id: xen-devel@lists.xenproject.org On 07/07/2010 04:28 AM, Joanna Rutkowska wrote: > I know this is really off-topic, but I'm curious whether you have a Core > i5/i7 processor with an AESNI instruction, and if you have, if you got > the aesni-intel module to work properly with your kernel? > > I noticed that using LUKS with a very fast SSD, that normally could have > a read throughput of around 200MB/s, significantly limits the > performance down to around 80-100 MB/s, with the bottleneck being the > kcryptd process easting 100% CPU (core). > No, this is a Core2 laptop. I don't do anything intensely IO bound on it (mostly seek-bound stuff), so I wouldn't notice a kcryptd performance regression too much. (Or, perhaps to be more accurate, when I switch to ssd I also added encryption, so the ssd still seems like marvel of speed compared to the hdd, even with the overhead.) J