You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The usage model needs to be consistent. Earlier the approach was to import numba and enable numba_dpex if inputs are usm ndarray.
Lately we revised the approach, and enable numba_dpex explicitly via import. By going this route, the implementation is numba_dpex need to be self-contained, and alias necessary attributes from numba.
The goal is to have a model that minimizes code changes by going from CPU script to heterogeneous script. It means in particular that we have to avoid situations when user needs to import both numba and numba_dpex.
The usage model needs to be consistent. Earlier the approach was to import numba and enable numba_dpex if inputs are usm ndarray.
Lately we revised the approach, and enable numba_dpex explicitly via import. By going this route, the implementation is numba_dpex need to be self-contained, and alias necessary attributes from numba.
The goal is to have a model that minimizes code changes by going from CPU script to heterogeneous script. It means in particular that we have to avoid situations when user needs to import both numba and numba_dpex.
Examples of aliases needed in numba_dpex
The text was updated successfully, but these errors were encountered: