-
Bram Veenboer authored
These methods are intended for cases when the Proxy takes ownership over the grid, for instance in the CUDA Generic proxy, where a copy of the grid is kept in GPU memory. Only upon set_grid and get_grid, the host memory is synchronized with the GPU memory. This approach has a number consequences for the other CUDA proxies as well, hence a lot of changes are made throughout the code.
c4f2cb92