Openwall GNU/*/Linux - a small security-enhanced Linux distro for servers
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Mon, 30 Apr 2012 00:21:03 +0400
From: Solar Designer <solar@...nwall.com>
To: john-dev@...ts.openwall.com
Subject: Re: JtR compilation warnings on OS X 10.7.3 with Xcode 4.3.2

Dhiru -

Thank you for making this test!

On Sun, Apr 29, 2012 at 06:08:48PM +0530, Dhiru Kholia wrote:
> Hard to believe but Apple's CommonCrypto library is faster than OpenSSL on OS X.
...
> Both programs do 10000000 cycles of SHA1.

What was the input size - was the compression function invoked just once
per cycle or more than once?

Can you repeat this test for SHA-512?

> Should we start modifying code to use CommonCrypto on OS X?
> (modifications are very simple to make).

For SHA-1 in particular, we should use the SSE2+ intrinsics
implementation that we have in jumbo.  However, if we expect to fail to
move all SHA-1-using formats to that soon enough, then I am fine with us
moving to CommonCrypto temporarily (in builds for OS X only, indeed).
My understanding is that right now some of the SHA-1-using formats use
the intrinsics and some use OpenSSL.

(Indeed, we should also keep fallbacks for pre-SSE2 and non-x86 CPUs,
but recent Mac OS X implies x86 with SSE2+.)

Thanks again,

Alexander

Powered by blists - more mailing lists

Your e-mail address:

Powered by Openwall GNU/*/Linux - Powered by OpenVZ