GitLab

Tony's changes to error handling in the CUDA platform so it throws exceptions instead of calling exit()

Removed debugging code

Check if aewald has been set; if not, then call NonbondedForceImpl::calcPMEParameters() to set aewald and grid dimensions

TorsionTorsion grids reordered, if needed, so that first angle is the 'slow' index
Fixes for MonteCarloBarostat 

Windows compiler did not like the following: std::max(std::max(gridSize.x, gridSize.y), gridSize.z);