john-users - Re: OpenCL performance numbers on Radeon HD 5870

Openwall GNU/*/Linux 3.0 - a small security-enhanced Linux distro for servers

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date: Thu, 22 Sep 2011 21:18:10 -0400
From: Erik Winkler <ewinkler@...ls.com>
To: john-users@...ts.openwall.com
Subject: Re: OpenCL performance numbers on Radeon HD 5870

Alexander,

You are correct, this is different.

Here is the output from the other CRYPTMD5 and CRYPTSHA256.  The cryptsha256 fails for some reason.

/john -test -format:CRYPTMD5OPENCL
Benchmarking: CRYPTMD5OPENCL [MD5-based CRYPT]... end read_kernel_source

OpenCL Device: ATI Radeon HD 5670
setup context
build kernel
end build program
DONE
Raw:	24284 c/s real, 1024K c/s virtual

./john -test -format:CRYPTSHA256OPENCL
Benchmarking: CRYPTSHA256OPENCL [SHA256-based CRYPT]... end read_kernel_source

OpenCL Device: ATI Radeon HD 5670
setup context
build kernel
end build program
end init
crypt_all
FAILED (get_hash[0](0))

On Sep 22, 2011, at 8:27 PM, Solar Designer wrote:

> How is this relevant?  I think it is not.  Maybe you're confusing Lukas'
> built-in SHA-crypt support (in CUDA and OpenCL) with my implementation
> of "generic crypt(3)" in the official JtR?  These are distinct things.
> 
> Or am I missing something?
> 
> Thanks again,
> 
> Alexander