[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <CA+TsHUASQATz-mDKbt22zeY09jAT6WQCo2fgyvAiwb3bMeZb7w@mail.gmail.com>
Date: Thu, 10 Jan 2013 00:52:28 +0530
From: Sayantan Datta <std2048@...il.com>
To: john-dev@...ts.openwall.com
Subject: Re: des-opencl
On Tue, Jan 8, 2013 at 2:43 AM, Solar Designer <solar@...nwall.com> wrote:
> On Tue, Jan 08, 2013 at 01:11:23AM +0530, Sayantan Datta wrote:
> > So you mean to say something like this would do:
> >
> > for(i=1;i<=25;i++) {
> > hardcode k=0;
> > hardcode k=96;
> > hardcode k=192;
> > hardcode k=288;
> > .
> > .
> > .
> > hardcode k=672;
> > }
> >
> > Although this ain't anyway equivalent to the current loop structure, but
> > maybe they are same mathematically. Is this what you mean?
>
> Yes, except that you need to remove or undo the final swapping of B[]
> halves after the last DES round (16th).
>
> The complicated branching in the current loop is to save on B[] halves
> swapping overhead.
>
> > Use the binary patch whenever possible and manually switch between the
> > other two fallback modes.
>
> OK.
>
> Alexander
>
Hi Alexander,
Fully hardcoding the kernel really improved performance from 112M to 138M.
Benchmarking: Traditional DES [OpenCL]... DONE
Raw: 1379K c/s real, 209715K c/s virtual
Using zero salt and 2501 iterations.
Regards,
Sayantan
Content of type "text/html" skipped
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.
