Date: Fri, 15 Jun 2012 23:06:21 +0200 From: Simon Marechal <simon@...quise.net> To: john-dev@...ts.openwall.com Subject: Re: Re: [patch] optional new raw sha1 implemetation On 06/15/2012 07:38 PM, Simon Marechal wrote: > On my i7-3770, this code give between 15.8M and 16.2M c/s with gcc > 4.6.3. The precompiled intrinsics code gives a consistent 16.5M. This is > really good. I am downloading ICC and will try to port some tricks. Finally managed to download and rebuilt everything with the latest icc (this was a pain). The build did had to be fixed : * it doesn't like the redefinition of _mm_testz_si128. I just commented it out but a good #ifdef is required * ASFLAGS needs to be overridden so that the compiler knows it should use AVX and uses CPU_detect should I do it and submit a patch or would magnum/jfoug do it himself ? I suppose I can even just push it ... The current SSE code cracks 19.8M c/s. Taviso's is faster at 21.3M c/s, and doesn't use the register scheduling trick that is in sse-intrinsics.c. This _might_ mean it could be faster. Some of the improvements can't be moved to the intrinsics code as it would break things, so it makes sense to keep it as is : * reversing the last steps * limiting the max password length The code seems to be full on fun stuff however, and to be a good read.
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.