Follow @Openwall on Twitter for new release announcements and other news
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Sun, 23 Jun 2013 23:36:35 +0400
From: Solar Designer <>
Subject: Re: Sayantan's weekly report #2

On Mon, Jun 24, 2013 at 12:51:39AM +0530, Sayantan Datta wrote:
> Accomplishment:
> 1. Accurate search for the set of ranges to be computed on GPU based on 
> requested no. of keys by the format.
> 2. Draft implementation of password generation on gpu for descrypt-opencl.
>             - Maximum no. of ranges that can be computed on GPU is 4 
> due to limited availability of local memory.
>             - Each range has maximum of 32 chars.

4 ranges is more than enough, but 32 chars is sometimes not enough.  Why
do they consume much local memory, can't we have this data shared?

Also, I think we should include an optimization in some/all kernels,
where a consecutive range of ASCII codes (and beyond, for 8-bit chars)
would be specified.  Thus, there would be two ways to pass a charset
spec for a char position into the kernel: either list of chars (need one
byte per char), or range of chars (two bytes to specify the entire range).

> Priorities:
> 1. More testing required.
> 2. Find out the bottleneck in descrypt-opencl 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.