From: john terragon Subject: Re: aesni-intel slower than aes-x86_64 Date: Mon, 12 Apr 2010 10:09:36 -0700 (PDT) Message-ID: <263254.5283.qm@web45116.mail.sp1.yahoo.com> References: <4BC3150A.1080905@redhat.com> Mime-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: QUOTED-PRINTABLE Cc: linux-crypto@vger.kernel.org To: Milan Broz Return-path: Received: from web45116.mail.sp1.yahoo.com ([68.180.197.184]:24866 "HELO web45116.mail.sp1.yahoo.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with SMTP id S1752441Ab0DLRJg convert rfc822-to-8bit (ORCPT ); Mon, 12 Apr 2010 13:09:36 -0400 In-Reply-To: <4BC3150A.1080905@redhat.com> Sender: linux-crypto-owner@vger.kernel.org List-ID: --- On Mon, 4/12/10, Milan Broz wrote: > From: Milan Broz > Subject: Re: aesni-intel slower than aes-x86_64 > To: "john terragon" > Cc: linux-crypto@vger.kernel.org > Date: Monday, April 12, 2010, 5:41 AM > On 04/12/2010 12:52 AM, john terragon > wrote: > > My system has a core i5 520M and supports AES-NI. I > wanted to do a > > rude performance test and so I ran these commands on a > small (4GB) partition and on the dm-crypt device backed by > it: > >=20 > > 1) using the aesni-intel module: dd > if=3D/dev/dev/mapper/vol of=3D/dev/null bs=3D4k > > 2) using only the generic aes-x86_64 module: dd > if=3D/dev/dev/mapper/vol > > of=3D/dev/null bs=3D4k > > 3) dd if=3D/dev/sda4 of=3D/dev/null bs=3D4k=20 > >=20 > > What I got kind of surprised me: > > 1) ~ 67 MB/sec, with a low cpu load by kcryptd > > 2) ~ 79 MB/sec, with a higher cpu load by kcryptd > > 3) ~ 81 MB/sec >=20 > Just curious=A0 - is it the same if you add > iflag=3Ddirect? Also try larger block size bs=3D1M etc. >=20 With iflag=3Ddirect I get about 1.2 MB/sec more or less in all the thre= e cases. With bs=3D1M I get slightly better results in all the three ca= ses however the increase for aesni is proportionally less than the other tw= o cases. So the performance differences are even slightly larger with b= s=3D1M. I don't get it. As it is, it seems that the aes new instructio= ns are practically useless (with my core i5 520M, at least). John =20 =20 -- To unsubscribe from this list: send the line "unsubscribe linux-crypto"= in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html