Re: [PyCUDA] Trying to get test_driver.py to run on OSX

[Aaron Cohen]

Andreas Klöckner li...@... writes:

 
 Hi Aaron,
 
 On Mittwoch 16 September 2009, Aaron Cohen wrote:
  One potential problem is that I installed the latest CUDA package,
  2.3. Does PyCUDA work with 2.3? The documentation implies that it will.
  Should I downgrade to CUDA Driver 2.2? How do I uninstall the 2.3
  package?
 
 I'm using 2.3 with PyCUDA, and have been for a while (on Linux). There 
 shouldn't be any issues.
 
  After suffering though the linkage problems in the CUDA sdk and the
  MacPorts boost and fixing that stuff with help from this mailing list,
  I've run into another roadblock. When I run test_driver.py I get this
  error message:
 
  $ python2.5 test_driver.py
  Traceback (most recent call last):
 File test_driver.py, line 25, in module
   assert isinstance(pycuda.autoinit.device.get_attributes(), dict)
 File /opt/local/lib/python2.5/site-packages/pycuda-0.94beta-py2.5-
  macosx-10.5-i386.egg/pycuda/driver.py, line 51, in
  device_get_attributes
   for att in dir(device_attribute)
 File /opt/local/lib/python2.5/site-packages/pycuda-0.94beta-py2.5-
  macosx-10.5-i386.egg/pycuda/driver.py, line 52, in genexpr
   if att[0].isupper())
  pycuda._driver.LogicError: cuDeviceGetAttribute failed: not found
 
 By this point, the code has already gotten past a few CUDA calls, so you 
 should generally be fine--try some of the demos, I imagine they might work. 
 Can you trace which attribute is being queried (just insert a print statement 
 somewhere along this traceback)? One reason may be that you compiled against 
 a 
 CUDA header that doesn't match your installed GPU driver, in which case the 
 header would have told PyCUDA about device attributes that don't exist in an 
 older version of the driver, which would lead to this error. Or maybe there's 
 something else going on that we don't know about yet.
 
 HTH,
 Andreas
 
 
 ___
 PyCUDA mailing list
 pyc...@...
 http://tiker.net/mailman/listinfo/pycuda_tiker.net
 

Hi Andreas:
I'm just using the latest 2.3 CUDA distributions from NVIDIA. 
These are the first and only ones that I have installed. 
The programs that I compiled from the NVIDIA sdk seem to work.

test_math.py produces no output.
test_gpuarray.py produces a long string of failure messages. 
The last line is 23 failed, 1 passed in 6.15 seconds.

I can insert some print statements if you want, but the code 
that needs to be instrumented is in driver.py, 
and this is a pretty deep list comprehension statement, 
so I'll have to expand it 
out to insert print statements.

-Aaron




___
PyCUDA mailing list
PyCUDA@tiker.net
http://tiker.net/mailman/listinfo/pycuda_tiker.net

Re: [PyCUDA] problem with pycuda._driver.pyd

[Andreas Klöckner]

On Donnerstag 17 September 2009, mailboxalpha wrote:
  I looked for DLL files used in _driver.pyd and one of them is named
  nvcuda.dll.  There is no such file on my machine.  Perhaps that is the DLL
  file that could not be found. The required boost dlls have been copied to
  windows\system32 directory and the boost lib  directory has been added to
  the system path.

Can you please check which DLL the CUDA examples require? There will be one 
that has the runtime interface, which will likely be called something with 
cudart. You don't care about that one. Instead, that DLL in turn requires 
the driver interface, and that's the DLL name we need.

Andreas


signature.asc
Description: This is a digitally signed message part.
___
PyCUDA mailing list
PyCUDA@tiker.net
http://tiker.net/mailman/listinfo/pycuda_tiker.net