Date: Sat, 25 Apr 2015 23:02:02 +0200 From: Agnieszka Bielec <bielecagnieszka8@...il.com> To: john-dev@...ts.openwall.com Subject: Re: [GSoC] John the Ripper support for PHC finalists 2015-04-25 20:52 GMT+02:00 Solar Designer <solar@...nwall.com>: > On Sat, Apr 25, 2015 at 08:41:54PM +0200, Agnieszka Bielec wrote: >> on well with AVX2 I receive results: >> >> a@...l:~/jtr2/run$ ./john --test --format=pomelo --cost=0:0,0:0 >> Will run 8 OpenMP threads >> Benchmarking: POMELO, Generic pomelo [AVX2]... (8xOMP) DONE >> Speed for cost 1 (N) of 0, cost 2 (r) of 0 >> Many salts: 1704K c/s real, 213274 c/s virtual >> Only one salt: 1055K c/s real, 132053 c/s virtual >> >> the difference is biggest for 0,0 costs also for 2,0 and 0,2 is a >> difference of 24% and 17% >> at this moment I don't know why this happens > > This suggests a bug in the benchmark. Where do the salts for the "many > salts" benchmark come from? > > set_key() in pomelo_fmt_plug.c doesn't look like it'd be that slow. > > Please investigate. yes, there is a bug but in my code, I modified bench.c to generate random keys and it seems that it takes so much time, for the CPU tests I must remove this modification
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.