[<prev] [next>] [day] [month] [year] [list]
Date: Thu, 5 Jul 2012 00:28:55 +0800
From: myrice <qqlddg@...il.com>
To: john-dev@...ts.openwall.com
Subject: Myrice: Weekly Report #12
Accomplishment:
- Bug fixed for xsha512-cuda/opencl raw-sha512-cuda/opencl
Priorities:
- Loaded hashes on GPU
- Password Generation on GPU
In order to let you know the loaded hashes progress and make it
appropriate before I make something wrong. Here is what I have already
done and will do:
In order to do reset/clear cache after self test.
- I set a global variable(in_self_test) in format.c
- In the begin of fmt_self_test(), in_self_test set to 0. And in the
end of it, in_self_test set to 1.
- In my format, I recorded and checked in_self_test, if it changes, I
will call a reset
In order to place hashes/salts on constant memory on GPU, I recorded
4000 hashes and salts by now. In format.h, we record hashes in a link
list with salt. But copy a link list to GPU is complex and
inefficient.
- Allocate the same number of salts and hashes
- For every hash, record its salts. Here we may record same salt multiple times
The criteria for finishing loading all hashes is we see the first
loaded hash twice.
Move cmp_all to crypt_all
- Compare loaded salts with given salt, if the same, compare the hashes
(I think we have the hashes with same salt in consecutive index in
the array. Because in crack.c, we pass hashes with same salt to
cmp_all() )
- Record compare result in correspond array with hashes
In cmp_all()
- Check the array and return the result
Thanks
myrice
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.
