Follow @Openwall on Twitter for new release announcements and other news
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Tue, 10 Jul 2012 07:37:24 +0400
From: Solar Designer <>
Subject: Re: raw-sha256-cuda

On Tue, Jul 10, 2012 at 05:29:11AM +0200, Lukas Odzioba wrote:
> 2012/7/10 Solar Designer <>:
> > On Tue, Jul 10, 2012 at 05:11:15AM +0200, Lukas Odzioba wrote:
> >> I limited PLAINTEXT_LENGTH to 19 chars so on TS now it is getting 1476/1500.
> >
> > Now that this passes TS, I guess we should remove the "unreliable" comment.
> Should I send another patch for that?

Please feel free to.  I don't know what magnum prefers for trivial
changes like this.

> >> Of course these changes affects sha224 too, which is slightly faster.
> >
> > Why is it slightly faster - are you transferring the entire computed
> > hashes back to host?
> Yes, I suppose that I should do something similar to myrice's code?

In the long run, yes - but what myrice is doing will keep changing as we
figure out how to do it best.  So it is unimportant for you to do it
now, and your time is in fact better spent on optimizing your slow
hashes, etc. than on raw SHA-256 currently, now that you got it working
to the same extent that other fast hashes on GPU do (at least passing
test and reaching raw SHA-512's speed).



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.