Follow @Openwall on Twitter for new release announcements and other news
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Thu, 17 May 2012 04:24:49 +0400
From: Solar Designer <solar@...nwall.com>
To: john-dev@...ts.openwall.com
Subject: Re: Myrice: Weekly Report #5

myrice, Samuele -

On Wed, May 16, 2012 at 03:03:15PM +0800, myrice wrote:
> Accomplishments:
> 1. Unoptimized SHA512 on cuda

That's good.

> Priorities:
> 1. Optimize sha512-cuda and test SHA512 cuda code  on bull
> 2. SHA512 OpenCL code
> 3. Keep on making xsha512-opencl works on 7970

Yes, please.

What has happened to your plans to implement storage of loaded
hashes and/or salts on the GPU end, though?  And to support occasional
long passwords while not hurting performance for short passwords?
Also, you wanted to comment on "Password Generation on GPU", but I think
you never did (I exchanged a few messages with Frank, and that's all).
I understand that you can't work on too many sub-tasks at once; I am
merely asking, recalling that this stuff was previously among your
Priorities, and seeing that now it just disappeared.

> I am stuck at my xsha512-opencl on 7970. It seems that bitwise
> operation is strange on 64bit variable. This slows my progress,

That's nasty.  Do you have access to any other ATI/AMD GPU (no older
than 4xxx series)?  If not, would providing you with remote access to
one be of help (to see if the issue is with code generation for 7970
specifically or with AMD APP SDK or driver in general)?  Oh, I recall
that Samuele reported that your code failed on his 6970 as well, right?

Thanks,

Alexander

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.