Follow @Openwall on Twitter for new release announcements and other news
[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Date: Fri, 4 May 2012 17:32:37 +0800
From: myrice <>
Subject: Re: SHA-512 on GPU

Samuele, magnum -

On Fri, May 4, 2012 at 3:35 PM, Samuele Giovanni Tonon
<> wrote:
> could you point me to the commit is causing your segfault so i can test
> it on other gpu cards ?

I am still debugging it. Now it is on
bull(/home/myrice/GSoC/myrice-JtR). You can copy it from bull.
In file, line 217. There is a printf. When I run
with ./john -te=1 -fo=xsha512-opencl -pla=1. There is segment fault. I
suspect that the address store causes the wrong result on 7970. Since
in my cuda code, I have experienced with memory align problem. That
is, when I convert a struct* within a uint64_t varible to char*, the
behavior is strange. So I will first eliminate all address store code.


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.