Openwall GNU/*/Linux - a small security-enhanced Linux distro for servers
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Thu, 03 May 2012 15:00:22 -0300
From: Claudio André <claudioandre.br@...il.com>
To: john-dev@...ts.openwall.com
Subject: Re: cryptsha512-opencl (plans)

Em 03-05-2012 07:59, magnum escreveu:
> Claudio,
>
> Do you think you can get it working on AMD soon too? 

I'm not optimistic. Now i can segfault the compiler using my best code 
for NVIDIA. And I do compile only.
I checked on bull CPU, NVIDIA and 7970. It is the SAME source code, i 
even removed #defines, #ifdefs, etc. The source:

     opencl_build_kernel("$JOHN/cryptsha512_kernel_NVIDIA.cl", gpu_id);
     exit(0); // <==
---

The result is: (and i did nothing or am i crazy? Not that crazy at least)
OpenCL platform 1: AMD Accelerated Parallel Processing, 2 device(s).
Using device 0: Tahiti
Segmentation fault (core dumped)
---

So i will:
- try your PUTCHAR macro.
- do the final test and ajustment.
- create and submit a patch.
- and than come back to 7970. At this time, we might have a better 12.04 
driver.

If we have a new version:
- I can remove the support to AMD GPU (using nice warning message).
- I can remove the support to southern island AMD GPU (idem).

BTW: the NVIDIA code gives me more than 3500 c/s. It is not that bad.

Claudio


Powered by blists - more mailing lists

Your e-mail address:

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