First working state of CudaTrackJob, CudaController, CudaArgument, etc. . This includes the python bindings, updated tests and some fixes to the examples
Please consult the tests/python/cuda/ files and examples/python/cuda_particle_addr.py for some templates as to how to use the new structures.
All tests pass in both release and debug mode
Addresses:
34 -> fully implements NS(CudaController), NS(CudaArgument) in C99, C++ and Python -> close
42 -> implements track_lines for cuda, opencl and cpu in c99, c++ and python but does not (yet) allow to only load a split-fraction of the kernel
TODO:
add additional tests, especially for changing nodes and resetting buffers
list operations on nodes including printing are limited, expand and harmonize them
implement automatically calculation and optimization of threads/block and block dimensions to optimize hardware occupancy
First working state of CudaTrackJob, CudaController, CudaArgument, etc. . This includes the python bindings, updated tests and some fixes to the examples
Please consult the tests/python/cuda/ files and examples/python/cuda_particle_addr.py for some templates as to how to use the new structures.
All tests pass in both release and debug mode
Addresses:
34 -> fully implements NS(CudaController), NS(CudaArgument) in C99, C++ and Python -> close
42 -> implements track_lines for cuda, opencl and cpu in c99, c++ and python but does not (yet) allow to only load a split-fraction of the kernel
TODO: