Hi Daniel,
Daniel <daniel.pagan(a)andphysicsforall.com> writes:
I just installed PyCuda and got it working on my
Windows 8.1 laptop. I'm
able to run the examples in VS 2010. My question concerns the 'preferred'
development environment for PyCuda: While it runs on Windows, I'm not able to
debug any kernel code using VS (Nvidia's Nsight) like I can using C/C++. I've
seen a bunch of threads concerning 'cuda-gdb' which leads me to believe folks
are using the command line on Linux to run PyCuda. Is this the 'preferred'
environment to run/debug PyCuda? Is there a way to debug the kernels in
PyCuda in Windows with/without VS Nsight or are we forced to use a command
line interface for debugging purposes?
this question suggests a way to feed PyCUDA programs to Nvidia's GUI
tools (admittedly, in Linux):
https://stackoverflow.com/questions/13188049/how-to-profile-pycuda-code-wit…
Is there perhaps an equivalent Windows way to sell either the script or
the Python interpreter to VS as the 'executable' and to then debug that?
Is VS able/willing to debug executables it hasn't compiled itself?
(Sorry, as you can tell, I'm not much of a Windows guy... Hope this
helps at least a little.)
Andreas