To implement the GPU routines, we modify the
examplealgorithm_gpu_impl.cu file, which will call our
kernels. The implementation details are irrelevant, and will not be
discussed in detail. The previous section concerned the addition of
two members, threads and blocks, to the
baseclass. These will be initialized by DefaultGPU's
init() member. For the sake of completeness, the entire GPU
implementation is listed below, including the kernel declarations and
implementations.
Joren Heit
2013-12-17