[PyCUDA] PyCuda include path

[Vu Nguyen]

Bryan Tong Minh <bryan.tongminh at ...> writes:

> 
> I could not find a bugtracker, so I'm posting this patch on this mailing list.
> 
> PyCuda had problems finding its header files using Python 2.6 on Windows. This
> is because it is installed in Lib\site-packages\pycuda, while the header files
> reside in Include\pycuda. To solve this, I replaced 
_find_pycuda_include_path()
> in compiler.py with the attached code
> 
> Best regards,
> Bryan
> 
> def _find_pycuda_include_path():
>     from imp import find_module
>     file, pathname, descr = find_module("pycuda")
> 
>     from os.path import join, exists
>     installed_path = join(pathname, "..", "include", "pycuda")
>     installed_path2 = join(pathname, "..", '..', '..', "include", "pycuda")
> 
>     development_path = join(pathname, "..", "src", "cuda")
>     development_path2 = join(pathname, "..", "..", "..", "src", "cuda")
> 
>     import sys
>     usr_path = "/usr/include/pycuda"
>     usr_local_path = "/usr/local/include/pycuda"
>     prefix_path = join(sys.prefix, "include" , "pycuda")
> 
>     if exists(installed_path):
>         return installed_path
>     elif exists(installed_path2):
>         return installed_path2
>     elif exists(development_path):
>         return development_path
>     elif exists(development_path2):
>         return development_path2
>     else:
>         if sys.platform == "linux2":
>             if exists(prefix_path):
>                 return prefix_path
>             elif exists(usr_path):
>                 return usr_path
>             elif exists(usr_local_path):
>                 return usr_local_path
> 
>         raise RuntimeError("could not find path to PyCUDA's C header files")
> 
> 

Thank you, this is a much better fix (I returned the absolute path, which worked 
for me).

I have gotten pycuda to work with Enthought Python distribution. However, I get 
this warning:

Warning (from warnings module):
  File "C:\Enthought Python\lib\site-packages\pycuda\compiler.py", line 110
    +stdout+stderr)
UserWarning: The CUDA compiler suceeded, but said the following:
kernel.cu
tmpxft_0000158c_00000000-3_kernel.cudafe1.gpu
tmpxft_0000158c_00000000-8_kernel.cudafe2.gpu

Is this something I should be worried about or can I just remove this check from 
compiler? It gives correct results, but this warning always shows up.

When I try using pycuda with Python(x,y), I always get "ImportError: DLL invalid 
memory access" originated from driver.py 1st line: import _driver. But I'm fine 
with Enthought's.

Thank you again.
-VN