This fixes #41 by rounding the global work-size down to a multiple of the local work-size.
Possible drawback: The value is now guaranteed to be a multiple of the local work-size, not a power of it!
As far as I can tell, this should have no negative consequences.
This fixes #41 by rounding the global work-size down to a multiple of the local work-size.
Possible drawback: The value is now guaranteed to be a multiple of the local work-size, not a power of it! As far as I can tell, this should have no negative consequences.