Has anyone had trouble running programs on these cards? I have a monte carlo code that I have compiled using pyopencl on Ubuntu Linux 12.04 with Nvidia 780 Ti and CUDA sdk version 6.0 (Not sure if that affects the opencl library). The program runs as expected and gives expected results. I can also change the platform to use the AMD SDK and run the program on the XEON processor in the same computer also with expected results.
I recently tried to compile and run the program on Windows 7 with a core i7 with HD4000 graphics. I am using pythonxy with the pyopencl module downloaded from the pyopencl extras page. All parts of the program work, image processing, interpolations etc. However, the main monte carlo routine fails to return expected results. I have put a variable in the code to read the energy, positions, etc. of the electrons and output that at the end. This variable returns values like I would expect. The positions are within the bounds I set and the energies are normally distributed about the mean I expect. The only thing I have been able to figure out is that the loop which deposits the electron energy is only going through once or twice and then exiting. There seems to be no reason for the exit in terms of the code, since the electrons still have energy after only one loop, so I assumed it is hardware related. Has anyone seen similar behavior with these cards? Is it possible I need a different compiler options or opencl library? Any ideas are appreciated. (I have previously posted the code on this list if you search for my name you can see it.) Joe Reese Haywood, Ph.D. Johnson Family Cancer Center Mercy Health Muskegon 1440. E. Sherman Blvd. Suite 300 Muskegon, MI 49444 Phone: 231-672-2019 Email: [email protected]
_______________________________________________ PyOpenCL mailing list [email protected] http://lists.tiker.net/listinfo/pyopencl
