CUDA_ERROR_INVALID_IMAGE with CUDA 3.2 on OS X

[GoogleCodeExporter]

What steps will reproduce the problem?
1.
Install Cuda Toolkit:
http://www.nvidia.com/object/thankyou.html?url=/compute/cuda/3_2_prod/toolkit/cu
datoolkit_3.2.17_macos.pkg

Install Cuda Drivers:
http://developer.download.nvidia.com/compute/cuda/3_2_prod/drivers/devdriver_3.2
.17_macos.dmg

2.
Compile and install pyrit core.

3.
Compile and install cuda core:
cd cpyrit-cuda-0.3.0
env LDFLAGS="-L/usr/local/cuda/lib" python setup.py build
sudo python setup.py install

4.
Try to list cores:
pyrit list_cores

What is the expected output? What do you see instead?

Expected: List of CPU and CUDA cores

Instead:
Traceback (most recent call last):
  File "/usr/local/bin/pyrit", line 6, in <module>
    pyrit_cli.Pyrit_CLI().initFromArgv()
  File "/Library/Python/2.6/site-packages/pyrit_cli.py", line 106, in initFromArgv
    func(self, **options)
  File "/Library/Python/2.6/site-packages/pyrit_cli.py", line 237, in list_cores
    with cpyrit.cpyrit.CPyrit() as cp:
  File "/Library/Python/2.6/site-packages/cpyrit/cpyrit.py", line 368, in __init__
    self.cores.append(CUDACore(queue=self, dev_idx=dev_idx))
  File "/Library/Python/2.6/site-packages/cpyrit/cpyrit.py", line 181, in __init__
    _cpyrit_cuda.CUDADevice.__init__(self, dev_idx)
SystemError: CUDA_ERROR_INVALID_IMAGE

What version of the product are you using? On what operating system?
Version 0.3.0
On OSX 10.6.6

Please provide any additional information below.

Cuda driver is up to date (screenshot)

Cuda pyrit core is linked against latest toolkit:
otool -L /Library/Python/2.6/site-packages/cpyrit/_cpyrit_cuda.so
/usr/lib/libssl.0.9.8.dylib (compatibility version 0.9.8, current version 0.9.8)
/usr/local/cuda/lib/libcuda.dylib (compatibility version 1.1.0, current version 
3.2.17)
/usr/lib/libz.1.dylib (compatibility version 1.0.0, current version 1.2.3)
/usr/lib/libSystem.B.dylib (compatibility version 1.0.0, current version 
125.2.1)

Original issue reported on code.google.com by alex.the...@gmail.com on 10 Jan 2011 at 11:03

Attachments:

• cuda.png

[GoogleCodeExporter]

which gpu are you using?

Original comment by lukas.l...@gmail.com on 11 Jan 2011 at 7:17

[GoogleCodeExporter]

Geforce GTX260

Original comment by alex.the...@gmail.com on 11 Jan 2011 at 8:06

[GoogleCodeExporter]

the GTX260 doesnt yet use Fermi, right? Please try 0.4-dev from svn anyway

Original comment by lukas.l...@gmail.com on 11 Jan 2011 at 6:54

[GoogleCodeExporter]

GTX2xx is not Fermi. GTX > 4xx is.
I tried the trunk and it works:

Pyrit 0.4.0-dev (svn r288) (C) 2008-2010 Lukas Lueg http://pyrit.googlecode.com
This code is distributed under the GNU General Public License v3+

The following cores seem available...
#1:  'CUDA-Device #1 'GeForce GTX 260''
#2:  'CPU-Core (SSE2)'
#3:  'Network-Clients'

But where is my 2nd CPU core? Is it occupied by utilizing cuda?

Original comment by alex.the...@gmail.com on 11 Jan 2011 at 8:33

[GoogleCodeExporter]

it is

Original comment by lukas.l...@gmail.com on 11 Jan 2011 at 8:34

• Changed state: WontFix