This thread is for discussing how energy evaluation may be enhanced using GPU's. Some ideas: - STL for GPU: https://developer.nvidia.com/thrust - see i.e. their "fancy iterators" which can be used to [zip up arrays of data](https://github.com/thrust/thrust/blob/master/examples/dot_products_with_zip.cu) - also runs on CPUs using OpenMP or TBB: https://github.com/thrust/thrust/wiki/Device-Backends. Use e.g. `-DTHRUST_DEVICE_SYSTEM=THRUST_DEVICE_SYSTEM_X` where `X=CPP, OMP, CUDA, TBB` - header only - thrust is also used in the [Aboria particle library](https://github.com/aboria/Aboria), see also #162 - force biased MC where forces are calculated using GPU code, either internally or via OpenMM
This thread is for discussing how energy evaluation may be enhanced using GPU's.
Some ideas:
-DTHRUST_DEVICE_SYSTEM=THRUST_DEVICE_SYSTEM_XwhereX=CPP, OMP, CUDA, TBB