Follow @Openwall on Twitter for new release announcements and other news
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Tue, 17 Jul 2012 22:16:48 +0400
From: Solar Designer <>
Subject: Re: Keychain format OpenCL implementation, please review

On Tue, Jul 17, 2012 at 10:02:05PM +0530, Dhiru Kholia wrote:
> I have implemented Keychain format using OpenCL. Apply attached patch
> to magnum-jumbo branch for testing.

This is very nice.  Thank you!  (No, I haven't tested it yet.)

> Speed is 132K c/s on ATI 6970. It is roughly 300x faster than
> single-core CPU version (AMD X3 720). Hopefully, I haven't screwed
> things up ;)

I guess this means that your CPU code is a lot slower than it can/should
be. ;-)  And your 140K c/s figure for 7970 (mentioned elsewhere)
suggests that you might still be spending much time in CPU code, which
is why changing the GPU makes little difference.

IIRC, this is PBKDF2-HMAC-SHA-1 with 1000 iterations producing a 3DES
key, right?  If so, I think that's roughly the same as 2000 iterations
(since SHA-1's output size is not enough for a 168/192-bit 3DES key
as-is).  For MSCash2, we have 10240 iterations.  So you should have 5
times higher speed, or about 500k c/s on 7970.

Also, I notice that your code does not use bitselect() and rotate() yet -
see the kernel we use for MSCash2 and merge those changes.

Thanks again,


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.