|
Message-ID: <20080621160042.GA31771@openwall.com> Date: Sat, 21 Jun 2008 20:00:42 +0400 From: Solar Designer <solar@...nwall.com> To: john-users@...ts.openwall.com Subject: macosx-x86-64 Hi, Mac OS X 10.5+ (Leopard) and Xcode 3.0+ support 64-bit applications not only on PowerPC, but also on Intel (x86) CPUs - and I've just added the support into JtR as well. If you want to try this out and/or help test and benchmark it, please download the latest revisions of Makefile and x86-64.S (as of this writing) from here: http://cvsweb.openwall.com/cgi/cvsweb.cgi/Owl/packages/john/john/src/ These work as drop-in replacements for the files in JtR 1.7.2. The new make target is "macosx-x86-64". On my MacBook (Core 2 Duo 2.0 GHz), this achieves the expected speedup for DES-based crypt(3) hashes (+11%) and for MD5-based crypt(3) hashes (+47%), does not significantly affect performance of LM and Kerberos AFS hashes, but slows Blowfish-based crypt(3) hashes down (-17%). All of this was in comparison to the older 32-bit "macosx-x86-sse2" target. While at it, I've also converted the code in x86-64.S to use instruction pointer relative addressing - this was required for Mac OS X, but I applied the change for all operating systems. This helps reduce code size, and performance should either remain the same or maybe become slightly better (because of the reduced code size and thus having more space in the L1 instruction cache available for other needs). Testing on Athlon64 and Core 2 confirms this so far (no easily measurable change), but I would not mind more benchmarks/reports confirming that there is indeed no slowdown because of this change on any common CPU. Thanks, Alexander -- To unsubscribe, e-mail john-users-unsubscribe@...ts.openwall.com and reply to the automated confirmation request that will be sent to you.
Powered by blists - more mailing lists
Confused about mailing lists and their use? Read about mailing lists on Wikipedia and check out these guidelines on proper formatting of your messages.