Where's cutil64.dll?

Discussion and support for the CUDA Multiforcers (Windows and Linux)
  • Ads

Where's cutil64.dll?

Postby mac4N6 » Fri Aug 24, 2012 6:57 pm

bitweasel, et. al,

I'm trying to get set up on some Dell 6100 Cloud servers running Win 2008 Server. These are slick boxes with a Dell C410X--I've got access to 16 Tesla 2070's spread out over 4 CPU's. When I try to run GRTGen-CUDA.exe, it complains that I don't have cutil64.dll installed.

I've installed the NVidia CUDA Toolkit Version 4.0.17, NVidia CUDA Tools 4.0.17 and their gpucomputingsdk_4.0.19_win_64.exe program.

Not sure where the "cutil64.dll" is supposed to come from.

Can anyone point me in the right direction?

Thanks,

Mac

Okay, so I found the file on NVIDIA's site and downloaded all the other src/lib/dll's etc.

Now I have a new problem--I'm trying to generate tables using GRTGen-OpenCL.exe -h NTLM -c charsetall -l 7 -i 16 --chainlength=71000
I get
Error:<..\..\src\OpenGL_Common\GRTOpenCL.cpp:146>unknown.

Are there certain package dependencies for the windows version? What do I need to get running?

Thanks,

Mac
mac4N6
 
Posts: 3
Joined: Fri Aug 24, 2012 6:50 pm

Re: Where's cutil64.dll?

Postby Bitweasil » Sun Aug 26, 2012 10:12 pm

You probably need to run it from the directory it's installed in - it's got the OpenCL kernels referenced by relative path right now. I plan to fix this soon.

Any reason you don't just download the generated NTLM len7 tables from the website via a torrent?

https://cryptohaze.com/gpurainbowtables.php
Bitweasil
Site Admin
 
Posts: 912
Joined: Tue Jan 20, 2009 4:26 pm

Re: Where's cutil64.dll?

Postby mac4N6 » Mon Aug 27, 2012 12:42 pm

I was in the c:\Program Files\Cryptohaze-Windows directory. I dropped the = in the --chainlength=71000 argument and I stopped getting that error, but it's still dying almost immediately after starting. I don't get an error message, just a pop up box that says "GRTGen-OpenCL.exe has stopped working. A problem cause the program to stop working correctly. Please close the program."

I could down load the free tables from the bittorent site, and will do so. I'm currently generating additional rainbow crack tables, and thought that if I could use GPGPU enabled generators, such as your GRTGen-OpenCL generator, then I could greatly accelerate that process too.

Any thoughts?

Mac
mac4N6
 
Posts: 3
Joined: Fri Aug 24, 2012 6:50 pm

Re: Where's cutil64.dll?

Postby Bitweasil » Mon Aug 27, 2012 2:56 pm

My stuff is totally incompatible with RainbowCrack type tools.

I use a different reduction function and a different file format. If that's your goal, my stuff will not work for you, period.

I've not *actually* run my stuff on a Tesla - I don't have access to any.

Could you paste the output from the ShowConfig-OpenCL tool? You might have multiple OpenCL runtimes installed, which would confuse the GRT tools right now. And what version are you running? 1.31a?

I'd definitely like to find out what the cause is here, since it should, to the best of my knowledge, work on that platform.
Bitweasil
Site Admin
 
Posts: 912
Joined: Tue Jan 20, 2009 4:26 pm

Re: Where's cutil64.dll?

Postby mac4N6 » Tue Aug 28, 2012 11:58 pm

Here's the ShowConfig-OpenCL Results. I have four servers with this config in my attack cluster:

Platform ID 0:
CL_PLATFORM_NAME: NVIDIA CUDA
CL_PLATFORM_VENDOR: NVIDIA Corporation
CL_PLATFORM_VERSION: OpenCL 1.0 CUDA 4.0.1


Platform 0:
Device ID 0:
CL_DEVICE_NAME: Tesla M2070
CL_DEVICE_VENDOR: NVIDIA Corporation
CL_DEVICE_AVAILABLE: Yes
CL_DEVICE_MAX_COMPUTE_UNITS: 14
CL_DEVICE_TYPE: CL_DEVICE_TYPE_GPU
CL_DEVICE_MAX_CLOCK_FREQUENCY: 1147 MHz
CL_DEVICE_GLOBAL_MEM_SIZE: 1279 MB
CL_DEVICE_LOCAL_MEM_SIZE: 48 KB
CL_DEVICE_LOCAL_MEM_TYPE: CL_LOCAL
CL_DEVICE_MAX_CONSTANT_BUFFER_SIZE: 64 KB


Device ID 1:
CL_DEVICE_NAME: Tesla M2070
CL_DEVICE_VENDOR: NVIDIA Corporation
CL_DEVICE_AVAILABLE: Yes
CL_DEVICE_MAX_COMPUTE_UNITS: 14
CL_DEVICE_TYPE: CL_DEVICE_TYPE_GPU
CL_DEVICE_MAX_CLOCK_FREQUENCY: 1147 MHz
CL_DEVICE_GLOBAL_MEM_SIZE: 1279 MB
CL_DEVICE_LOCAL_MEM_SIZE: 48 KB
CL_DEVICE_LOCAL_MEM_TYPE: CL_LOCAL
CL_DEVICE_MAX_CONSTANT_BUFFER_SIZE: 64 KB


Device ID 2:
CL_DEVICE_NAME: Tesla M2070
CL_DEVICE_VENDOR: NVIDIA Corporation
CL_DEVICE_AVAILABLE: Yes
CL_DEVICE_MAX_COMPUTE_UNITS: 14
CL_DEVICE_TYPE: CL_DEVICE_TYPE_GPU
CL_DEVICE_MAX_CLOCK_FREQUENCY: 1147 MHz
CL_DEVICE_GLOBAL_MEM_SIZE: 1279 MB
CL_DEVICE_LOCAL_MEM_SIZE: 48 KB
CL_DEVICE_LOCAL_MEM_TYPE: CL_LOCAL
CL_DEVICE_MAX_CONSTANT_BUFFER_SIZE: 64 KB


Device ID 3:
CL_DEVICE_NAME: Tesla M2070
CL_DEVICE_VENDOR: NVIDIA Corporation
CL_DEVICE_AVAILABLE: Yes
CL_DEVICE_MAX_COMPUTE_UNITS: 14
CL_DEVICE_TYPE: CL_DEVICE_TYPE_GPU
CL_DEVICE_MAX_CLOCK_FREQUENCY: 1147 MHz
CL_DEVICE_GLOBAL_MEM_SIZE: 1279 MB
CL_DEVICE_LOCAL_MEM_SIZE: 48 KB
CL_DEVICE_LOCAL_MEM_TYPE: CL_LOCAL
CL_DEVICE_MAX_CONSTANT_BUFFER_SIZE: 64 KB


Why are the Tesla M2070's not recognized as CUDA by your code? If they were operational as OpenCL devices, will your apps be able to access their GPU's?
Is it my OpenCL Platform that needs to be updated to 1.31a?


Thanks,

Mac
mac4N6
 
Posts: 3
Joined: Fri Aug 24, 2012 6:50 pm

Re: Where's cutil64.dll?

Postby Bitweasil » Wed Aug 29, 2012 3:15 pm

mac4N6 wrote:Why are the Tesla M2070's not recognized as CUDA by your code? If they were operational as OpenCL devices, will your apps be able to access their GPU's?
Is it my OpenCL Platform that needs to be updated to 1.31a?


What driver revision are you running with them? I have run into issues with older drivers than I built against, and I vaguely recall that the Tesla drivers tend to be massively out of date.

Could you try putting the latest consumer drivers on one of the machines & see what happens? I do this with my script for EC2 - I put the latest consumer drivers on, since they are more up to date with CUDA.

Otherwise, an option is to build the tools for your environment. This should work as well, and will build against your current driver version.
Bitweasil
Site Admin
 
Posts: 912
Joined: Tue Jan 20, 2009 4:26 pm


Return to CUDA Multiforcers

Who is online

Users browsing this forum: No registered users and 1 guest

cron