I am an experienced Python hacker. I year ago I had never used CUDA or PyCUDA. I had tried to learn CUDA through many different tutorials and finally found that udacity course and I immediately understood several things with GPUs that I never got the hang of before.
After the course I looked at the examples provided by PyCUDA and I had what I needed to start playing. Now I have a full fledged parallel ODE solver implemented for arbitrary ODEs using PyCUDA. Thanks for the excellent software :) Johan On Sun, Feb 15, 2015 at 5:51 PM, Jerome Kieffer <[email protected]> wrote: > On Sat, 14 Feb 2015 19:47:17 +0100 > Alessandro Barracco <[email protected]> wrote: > > > Thank for fast reply. But OpenCL is not cuda (and haven't the same > > performances) > > Actually OpenCL is not Cuda but the OpenCL driver for nvidia hardware > declares itself as: > > Platform Version: OpenCL 1.1 CUDA 6.5.33 > Platform Name: NVIDIA CUDA > Platform Vendor: NVIDIA Corporation > > The difference in speed is still to be proven (above the error bar) > > > and udacity is a web service while I need a printed book. > > Udacity offers online courses ... in a couple of hours you get > bootstrapped. > I guess it would be faster than reading a book. > If you want a book, print the documentation of the nvidia SDK. > > -- > Jérôme Kieffer > Data analysis unit - ESRF > > _______________________________________________ > PyCUDA mailing list > [email protected] > http://lists.tiker.net/listinfo/pycuda >
_______________________________________________ PyCUDA mailing list [email protected] http://lists.tiker.net/listinfo/pycuda
