No, this does not leverage CUDA hardware well. You want a parallel reduction method instead. See developer.download.nvidia.com/compute/cuda/1_1/Website/Data-Parallel_Algorithms.html. Doesn't pyCUDA already have a min/max function? On Aug 24, 2011 4:36 AM, "Tomasz Rybak" <bogom...@post.pl> wrote:
_______________________________________________ PyCUDA mailing list PyCUDA@tiker.net http://lists.tiker.net/listinfo/pycuda