Openwall GNU/*/Linux - a small security-enhanced Linux distro for servers
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Sun, 24 Jun 2012 13:11:44 +0800
From: myrice <qqlddg@...il.com>
To: john-dev@...ts.openwall.com
Subject: Re: xsha512-opencl, raw-sha512-opencl on HD 7970 (was:
 Myrice: Weekly Report #3)

On Sun, Jun 24, 2012 at 5:08 AM, Solar Designer <solar@...nwall.com> wrote:
> myrice -
>
> This is a reminder that the problem mentioned below is still present.
> Yet raw-sha512-opencl passes self-test fine (but is slow).
> $ ./john -te -fo=xsha512-opencl -pla=1
> OpenCL platform 1: AMD Accelerated Parallel Processing, 2 device(s).
> Using device 0: Tahiti
> Compilation log: LOOP UNROLL: pragma unroll (line 163)
>    Unrolled as requested!
> LOOP UNROLL: pragma unroll (line 168)
>    Unrolled as requested!
> LOOP UNROLL: pragma unroll (line 183)
>    Unrolled as requested!
>
> Optimal Group work Size = 256
> Global work size = 2097152
> Benchmarking: Mac OS X 10.7+ salted SHA-512 [OpenCL]... FAILED (get_hash[0](0))
>

This is strange. I have already fixed it I think. I will test it later.

> $ ./john -te -fo=raw-sha512-opencl -pla=1
> OpenCL platform 1: AMD Accelerated Parallel Processing, 2 device(s).
> Using device 0: Tahiti
> Compilation log: LOOP UNROLL: pragma unroll (line 163)
>    Unrolled as requested!
> LOOP UNROLL: pragma unroll (line 168)
>    Unrolled as requested!
> LOOP UNROLL: pragma unroll (line 183)
>    Unrolled as requested!
>
> Optimal Group work Size = 128
> Global work size = 524288
> Benchmarking: Raw SHA-512 [OpenCL]... DONE
> Raw:    17854K c/s real, 29756K c/s virtual
>

The speed of raw SAH512-512 is the same with one salt in
xsha512-opencl format. The CPU takes too much time. The GPU password
generation and async copy you posted just now could improve the
performance. I will comment on that now.

Thanks
myrice

Powered by blists - more mailing lists

Your e-mail address:

Powered by Openwall GNU/*/Linux - Powered by OpenVZ