Openwall GNU/*/Linux - a small security-enhanced Linux distro for servers
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Mon, 6 Apr 2015 14:40:45 +0200
From: Agnieszka Bielec <bielecagnieszka8@...il.com>
To: john-dev@...ts.openwall.com
Subject: Re: [GSoC] John the Ripper support for PHC finalists

2015-04-06 11:58 GMT+02:00 Frank Dittrich <frank.dittrich@...lbox.org>:
>you replied to Solar Designer, but quoted one of my messages. Please
>stop doing this.

Sorry I'm not sure what you mean.

>Somehow in your test output these two lines (or some similar lines) are
>missing:  [...]

I cut the output because I added some printfs

>What problems did you have? Did you mention them in any of your previous
>mails in a way that others could understand these problems?

I mentioned that I need to know the m_cost value before the kernel
compilation. I didn't mention about details because when I was trying with
--cost  there was a lot of problems with that and all the
time I was discovering a new one

>> I can give a list of arguments why I did it in this way if you want

>Please do. May be even better would have been discussing it before you
>went ahead implementing this.

1.first of all --cost only chooses "good" hashes, if I want to read --cost
before the kernel compilation I need to use this

2015-04-04 21:08 GMT+02:00 Agnieszka Bielec <bielecagnieszka8@...il.com>:
>I've discovered that --cost modifies also fmt_test structure in fmt_main
>(but only with the --test flag!). I can take advantage of this.

It's only a trick like moving init() to set_key()


2.
2015-04-04 21:08 GMT+02:00 Agnieszka Bielec <bielecagnieszka8@...il.com>:
>If I change one hash in fmt_test to a wrong one
>I have
>./john --test --format="django-scrypt" --cost="14,8,1"
>Speed for cost 1 (N) of 14, cost 2 (r) of 8, cost 3 (p) of 1
>Raw:    403 c/s real, 14.0 c/s virtual

>./john django --format="django-scrypt" --cost="14,8,1"
>Self test failed (cmp_all(1))

>It's a bug? (auto tune for all hashes)

2015-04-05 0:01 GMT+02:00 Frank Dittrich <frank.dittrich@...lbox.org>:
>This is not a bug, but expected behavior.

this is wery bad for me because I need to compile the kernel again for
various values or choose the bigger one

3. When I add tunable functions and when I initialize everything in
reset() but not in init (like in opencl_mysqlsha1_fmt_plug.c)
 - it crashes somewhere beyond my code.
If I have init normal - (not like in opencl_mysqlsha1_fmt_plug.c) it works


4.
2015-04-04 21:08 GMT+02:00 Agnieszka Bielec <bielecagnieszka8@...il.com>:
>I've discovered that --cost modifies also fmt_test structure in fmt_main
>(but only with the --test flag!). I can take advantage of this.

If I have the range of cost values here is a problem again
I need to
"compile the kernel again for
various values or choose the bigger one... or compile the kernel over
and over again"

this are most important reasons

_______________________________________________________________________
This list is only for my of pomelo-opencl using __private
memory. Already I've tested __global memory which is slow, __local which
is limited and __private memory also isn't good.
Big chunks on __private in reality are actually "virtual" __global and
__local memory. I read somewhere (I don't remember the link) that not
always graphics cards good cope with this and from here I have
OUT_OUT_RESOURCES in --dev=1 and in --dev=5 and --dev=6 for the bigger
values (but maybe I am wrong, maybe someone has an other idea).
I want now to experiment with __global and __local memory managed by
hand and we will see what happen.
I think that pomelo hash is really hard to implement using GPU

I uploaded my code (__private memory) to the branch "private_memory"
but it's rather a demo. I have some warnings but it's not important now

and we run
./john --test --format=pomelo-opencl --dev=3 --t_cost=4 --m_cost=4
john doesn't check if --t_cost and --m_cost is given now

_____________________________________________________________________

I'm including tests for __global memory and __private

I've added some printfs to know how many memory is used

http://pastebin.com/Rqe5yKsH

I'm wondering why on --dev=2 opencl using
global memory was fast, ~ 150k

Thanks

Content of type "text/html" skipped

Powered by blists - more mailing lists

Your e-mail address:

Powered by Openwall GNU/*/Linux - Powered by OpenVZ