[QST] Best practices for using with_cupy_rmm vs rmm_cupy_ary

[beckernick]

It's often desirable to use RMM as the memory allocator for CuPy functions. At the function execution level, this can be done by wrapping the function call with rmm_cupy_ary. At the function definition level, this can be done by using the with_cupy_rmm decorator.

Both of these approaches are used across the codebase. For developers looking to contribute, are there guidelines as to when they should opt for execution level RMM allocation vs definition level allocation? Should all cuML function definitions calling CuPy functions be decorated with with_cupy_rmm?

[github-actions]

This issue has been marked rotten due to no recent activity in the past 90d. Please close this issue if no further response or action is needed. Otherwise, please respond with a comment indicating any updates or changes to the original issue and/or confirm this issue still needs to be addressed.

[github-actions]

This issue has been marked stale due to no recent activity in the past 30d. Please close this issue if no further response or action is needed. Otherwise, please respond with a comment indicating any updates or changes to the original issue and/or confirm this issue still needs to be addressed. This issue will be marked rotten if there is no activity in the next 60d.

[beckernick]

As noted in #3409 (comment) , we can now close this issue. Both of these can be deprecated in favor of @cuml.internals.api_return_any(). More detailed information can be found in the Estimator Guide.