Date: Mon, 16 Jan 2012 21:40:34 +0400 From: Solar Designer <solar@...nwall.com> To: john-dev@...ts.openwall.com Subject: Re: SHA-crypt On Mon, Jan 16, 2012 at 04:48:46PM +0100, magnum wrote: > Pushed to GitHub now, both sha-crypt256 and sha-crypt512. I didn't find > much to optimise, it's spending over 95% in the OpenSSL functions. But > OMP should scale well (I only tested two cores). Thank you! Here's what I am getting on 2xE5420: Benchmarking: SHA-crypt-256 [OpenSSL 32/64]... (8xOMP) DONE Many salts: 1756 c/s real, 219 c/s virtual Only one salt: 1756 c/s real, 219 c/s virtual Benchmarking: SHA-crypt-512 [OpenSSL 64/64]... (8xOMP) DONE Many salts: 2112 c/s real, 262 c/s virtual Only one salt: 2092 c/s real, 262 c/s virtual I guess we should set BENCHMARK_LENGTH to -1 such that we only get one "Raw" line. Key setup takes a negligible portion of time for slow hashes like this. BTW, I notice that you use #include "omp.h" in your new source files, whereas the rest of JtR uses #include <omp.h> - I suggest that you use the latter (only look for omp.h in system directories). Alexander
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.