[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Sat, 11 Apr 2015 19:13:10 +0200
From: Frank Dittrich <frank.dittrich@...lbox.org>
To: john-dev@...ts.openwall.com
Subject: Re: [GSoC] John the Ripper support for PHC finalists
On 04/11/2015 07:03 PM, Agnieszka Bielec wrote:
> tests on well:
>
> without SSE2
>
> a@...l:~/nosse/run$ ./john --format=pomelo --test
> Will run 8 OpenMP threads
> Benchmarking: pomelo, Generic pomelo [Pomelo]... (8xOMP) DONE
> Many salts: 61440 c/s real, 7670 c/s virtual
> Only one salt: 61312 c/s real, 7683 c/s virtual
>
> a@...l:~/nosse/run$ OMP_SCALE=2 ./john --format=pomelo --test
> Will run 8 OpenMP threads
> Benchmarking: pomelo, Generic pomelo [Pomelo]... (8xOMP) DONE
> Many salts: 61440 c/s real, 7680 c/s virtual
> Only one salt: 60831 c/s real, 7670 c/s virtual
>
> SSE2
>
> a@...l:~/sse/run$ ./john --format=pomelo --test
> Will run 8 OpenMP threads
> Benchmarking: pomelo, Generic pomelo [Pomelo]... (8xOMP) DONE
> Speed for cost 1 (N) of 2, cost 2 (r) of 2
> Many salts: 89984 c/s real, 11262 c/s virtual
> Only one salt: 89856 c/s real, 11232 c/s virtual
>
> a@...l:~/sse/run$ OMP_SCALE=2 ./john --format=pomelo --test
> Will run 8 OpenMP threads
> Benchmarking: pomelo, Generic pomelo [Pomelo]... (8xOMP) DONE
> Speed for cost 1 (N) of 2, cost 2 (r) of 2
> Many salts: 90112 c/s real, 11264 c/s virtual
> Only one salt: 89856 c/s real, 11232 c/s virtual
On my system (Intel quad core i5-4570) I get 51648 c/s without SSE2 and
73216 c/s with SSE2.
While the memset at the end of POMELO() is necessary to clear the
memory, it is not necessary for us doing password "guessing" instead of
hashing real user passwords.
Just commenting out this memset in pomelo_plug.c and pomelo_sse2_plug.c
helps a little bit: I get 52160 c/s without SSE2 and 74240 c/s with SSE2.
Frank
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.
