[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <4FF69DD7.9000406@linuxasylum.net>
Date: Fri, 06 Jul 2012 10:12:07 +0200
From: Samuele Giovanni Tonon <samu@...uxasylum.net>
To: john-users@...ts.openwall.com
Subject: Re: Re: 1.7.9-jumbo6 opencl on nvidia
On 07/05/2012 10:34 PM, Rick "Zero_Chaos" Farina wrote:
> On 07/05/2012 03:59 PM, Rick "Zero_Chaos" Farina wrote:
>> I notice that when I build for just op encl support or both opencl and
>> cuda support that "john --test" fails:
>
>> OpenCL platform 0: NVIDIA CUDA, 1 device(s).
>> Using device 0: NVS 4200M
>> OpenCL error (UNKNOWN ERROR :() in file (common-opencl.c) at line (39) -
>> (Source kernel not found!)
>
> Okay so I'm new at this.... it helps a lot when you actually install the
> opencl kernels... *doh*
>
>> I am currently packaging this up for gentoo and it could be my fault but
>> when I build straight from source for opencl I get this:
>
>
> Below, however, is still a valid concern. This happens with my package
> as well as with straight from source builds.
>
> Thanks!
>
>> OpenCL platform 0: NVIDIA CUDA, 1 device(s).
>> Using device 0: NVS 4200M
>> Local work size (LWS) 64, Global work size (GWS) 4194304
>> Benchmarking: Netscape LDAP salted SHA-1 [OpenCL]... DONE
>> Many salts: 26214K c/s real, 25982K c/s virtual
>> Only one salt: 22469K c/s real, 22469K c/s virtual
>
>> OpenCL platform 0: NVIDIA CUDA, 1 device(s).
>> Using device 0: NVS 4200M
>> Max local work size 1024 Error -4
>> Error -4
>> Error -4
>> Error -4
>> Error -4
>> Error -4
>> Error -4
>> Error -4
>> Error -4
>> Error -4
>> Error -4
>> Optimal local work size 1
>> (to avoid this test on next run do export LWS=1)
>> Local work size (LWS) 1, Global work size (GWS) 2097152
>> Benchmarking: Raw MD5 [OpenCL]... OpenCL error
>> (CL_MEM_OBJECT_ALLOCATION_FAILURE) in file (opencl_rawmd5_fmt.c) at line
>> (338) - (failed in clEnqueueWriteBuffer buffer_keys)
>> OpenCL error (CL_INVALID_CONTEXT) in file (rar_fmt.c) at line (356) -
>> (Error Unmapping aes_key)
>
hello,
this happens only when you do a -test .
this is well known and it'is due to some uninitialization problem during
the test cycle.
if you do
./john -fo:ssha-opencl -test
and
./john -fo:raw-md5-opencl -test
separately it should work with no problem while we are investigating the
issue.
cheers
Samuele
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.
