[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <CABob6ios+6GLL2Z9T0CrNVcmnfwxquU58aKc9wZg+ctNBd0anw@mail.gmail.com>
Date: Thu, 9 Aug 2012 17:29:18 +0200
From: Lukas Odzioba <lukas.odzioba@...il.com>
To: john-dev@...ts.openwall.com
Subject: Re: Current -fixes GPU formats vs TS
2012/8/6 magnum <john.magnum@...hmail.com>:
> form=pwsafe-cuda guesses: 0 unk unk : Expected
> count(s) (1485) [!!!FAILED!!!]
> Self test failed (cmp_all(1))
>
> ...pwsafe-cuda sometimes crack 90% of the hashes though:
>
> form=pwsafe-cuda guesses: 1375 time: 0:00:13:34 :
> Expected count(s) (1500) [!!!FAILED!!!]
I think that the root cause is somewhere near that:
-when i do not use gpu for cuda for some time and then run -test
-fo=pwsafe-cuda format fails self tests - just once, next run 1s later
and I got pass on self test and later it passes too for some good time
-when i run -test, format fails always
I thought that it is caused by lack of cuda_init but even after adding
it problem still remains. I tried to clean buffers at the init, check
for allocation errors and similar often bugs but everything seems to
be ok.
I spent some time on it yesterday, I am able to reproduce this bug by
running -test but it is not so trivial to get the idea what is wrong.
Now I am trying to do step by step reproduction and probably this will
help find the problem.
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.
