Skip site navigation (1)Skip section navigation (2)
Date:      Tue, 18 Apr 2006 13:40:02 -1000 (HST)
From:      Dave Cornejo <dave@dogwood.com>
To:        Bruce M Simpson <bms@spc.org>
Cc:        freebsd-net@freebsd.org
Subject:   Re: crypto accelerators
Message-ID:  <200604182340.k3INe2Ep085862@white.dogwood.com>
In-Reply-To: <20060418191015.GE28496@spc.org>

next in thread | previous in thread | raw e-mail | index | archive | help
> On Mon, Apr 17, 2006 at 04:44:38PM -1000, Dave Cornejo wrote:
> > So the question is whether these cards, regardless of their affect on
> > throughput, increase usable CPU cycles?  I have several Soekris 1401
> > cards and am wondering if there would be any point to putting them
> > into some machines that provide logins over ssh.  These machines are
> > generally pretty good spec, 2.4GHz+, 1GB RAM, Intel MBs, mostly
> > on-board peripherals.
> 
> Given that spec of machine, I don't see that a hardware cipher would
> offer much improvement -- and some of the available crypto accelerators
> don't perform Diffie-Helmann or AES, some do.
> 
> I myself have a ubsec(4) card, and even when I hacked OpenSSH to use
> OpenSSL engine support by default (with someone else's patch), I didn't
> see that much improvement (even when I forced the use of MD5, RSA and
> 3DES).
> 
> I could be wrong though - the above is qualitative not quantitative.
> 
> Regards,
> BMS

it sounds like you're thinking in terms of throughput and speed of the
encrypted connections, which i agree probably won't see much of an
improvement.

but it would seem to me that doing the heavy math off-CPU reduces the
amount of work the CPU does.  are these saved CPU cycles available to
someone who might be doing a compilation on this machine?

Doug Ambriskos answer (thanks!) implies that maybe they are.

thanks,
dave c



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200604182340.k3INe2Ep085862>