Thanks Bogdan, that got it working. I figured it must be something to do with the path, but I had already added the relevant directories to the "Paths" tab in the Project Interpreter settings page, apparently those don't get included in the environment. For the record, I had to add /usr/local/cuda/bin (for nvcc), then /usr/local/bin (for clang compiler) and finally /usr/bin (for rm).
Ben ________________________________________ From: Bogdan Opanchuk [manti...@gmail.com] Sent: 16 January 2014 11:16 To: Rowland Ben Cc: pycuda@tiker.net Subject: Re: [PyCUDA] mvcc Errno 2 problem using PyCharm IDE Hi Ben, In Terminal, you have /usr/local/cuda/bin in .bash_profile (or analogous config file if you are using some other shell), so you can access nvcc without specifying the path to it. PyCharm does not use your .bash_profile, and requires you to set the environment by other means. See http://stackoverflow.com/questions/8732119/pycharm-path-on-os-x , it has two ways to do it (one that will work for other applications too, and one PyCharm-specific). On Thu, Jan 16, 2014 at 9:00 PM, Rowland Ben <rowland....@claudiusregaud.fr> wrote: > Dear list, > > I have been using PyCuda for a couple of months now and really enjoying > myself (albeit with a few frustrating moments along the way). However as I am > moving from test programs to a more complicated real case I want to use my > customary Python IDE (PyCharm by JetBrains) rather than just using text > editors to write the code. Unfortunately I haven’t been able to get the two > playing nicely together so far, and am now turning to you guys for help. > > I can import the PyCuda modules into my project with no problem, this part of > the code runs fine. Unfortunately when I get to the point where I actually > try and compile using a SourceModule, I get the following error: > > Traceback (most recent call last): > File "/Users/ben/MRSI/python/mrsi/cuda_flowchart.py", line 847, in > recompile_cuda_code > no_extern_c=True) > File > "/Users/ben/Library/Enthought/Canopy_64bit/User/lib/python2.7/site-packages/pycuda-2013.1.1-py2.7-macosx-10.6-x86_64.egg/pycuda/compiler.py", > line 262, in __init__ > arch, code, cache_dir, include_dirs) > File > "/Users/ben/Library/Enthought/Canopy_64bit/User/lib/python2.7/site-packages/pycuda-2013.1.1-py2.7-macosx-10.6-x86_64.egg/pycuda/compiler.py", > line 252, in compile > return compile_plain(source, options, keep, nvcc, cache_dir) > File > "/Users/ben/Library/Enthought/Canopy_64bit/User/lib/python2.7/site-packages/pycuda-2013.1.1-py2.7-macosx-10.6-x86_64.egg/pycuda/compiler.py", > line 118, in compile_plain > result, stdout, stderr = call_capture_output(cmdline, cwd=file_dir, > error_on_nonzero=False) > File > "/Users/ben/Library/Enthought/Canopy_64bit/User/lib/python2.7/site-packages/pytools-2013.5.6-py2.7.egg/pytools/prefork.py", > line 196, in call_capture_output > return forker[0].call_capture_output(cmdline, cwd, error_on_nonzero) > File > "/Users/ben/Library/Enthought/Canopy_64bit/User/lib/python2.7/site-packages/pytools-2013.5.6-py2.7.egg/pytools/prefork.py", > line 53, in call_capture_output > % ( " ".join(cmdline), e)) > pytools.prefork.ExecError: error invoking 'nvcc --cubin -arch sm_30 -m64 > -I/Users/ben/MRSI/pycuda > -I/Users/ben/Library/Enthought/Canopy_64bit/User/lib/python2.7/site-packages/pycuda-2013.1.1-py2.7-macosx-10.6-x86_64.egg/pycuda/cuda > kernel.cu': [Errno 2] No such file or directory > > I have no idea why this should be as the identical code file runs fine when I > run it from the Terminal (I am on OS X), all the files and directories that I > am aware of are in the correct places, nvcc is on the path. > > All suggestions or solutions gratefully received. > > Thanks in advance, > > Ben > _______________________________________________ > PyCUDA mailing list > PyCUDA@tiker.net > http://lists.tiker.net/listinfo/pycuda _______________________________________________ PyCUDA mailing list PyCUDA@tiker.net http://lists.tiker.net/listinfo/pycuda
