From: Michal Ludvig <[EMAIL PROTECTED]> Subject: Re: VIA PadLock reloaded To: James Morris <[EMAIL PROTECTED]> Cc: CryptoAPI List <[EMAIL PROTECTED]> Date: Sun, 24 Oct 2004 01:55:03 +0200
From: Michal Ludvig <[EMAIL PROTECTED]> Date: Sun, 24 Oct 2004 01:55:03 +0200 To: James Morris <[EMAIL PROTECTED]> Cc: CryptoAPI List <[EMAIL PROTECTED]> Subject: Re: VIA PadLock reloaded User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.7) Gecko/20040616 Michal Ludvig wrote: > James Morris wrote: > >> On Sat, 23 Oct 2004, Michal Ludvig wrote: >> >>> I'm currently updating the driver for VIA PadLock cryptoengine and once >>> I'm on it I'd like to prepare it for kernel inclusion. >> >> Have you done any performance measurements with this? It would be >> interesting to see its effect on IPSec and disk encryption. > > Yes, some numbers are at http://www.logix.cz/michal/devel/padlock/bench.xp Just in case you have already looked there - few minutes ago I have added a new section with IPsec benchmark. Rough results: Plaintext throughput: 11.21 MB/s IPsec (ESP/transport) without HMAC: - PadLock AES: 11.00 MB/s (keylen independent) - AES-i586: 8.01 to 9.84 MB/s depending on the keylen - Generic C AES: 6.37 to 8.24 MB/s IPsec with HMAC-SHA256: - PadLock AES: 8.06 MB/s - Software AES slower by some 45% than without HMAC. As soon as I get VIA Esther that can do SHA1/SHA256 in hardware I'll update the padlock driver as well. Than I expect almost no slowdown even in HMAC mode (which is almost always used in ESP). Michal Ludvig _______________________________________________ Subscription: http://lists.logix.cz/mailman/listinfo/cryptoapi List archive: http://lists.logix.cz/pipermail/cryptoapi ---------- -- Eugen* Leitl <a href="http://leitl.org">leitl</a> ______________________________________________________________ ICBM: 48.07078, 11.61144 http://www.leitl.org 8B29F6BE: 099D 78BA 2FD3 B014 B08A 7779 75B0 2443 8B29 F6BE http://moleculardevices.org http://nanomachines.net
pgp2BDUwIHk2S.pgp
Description: PGP signature