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
Currently, the code wrapping mpi4py for torch.Tensor's is contained in heat/core/communication.py and not fully separated from constructions related to the DNDarray-class (e.g., computation of chunk sizes is also introduced there).
I would suggest to separate this clearly, which may also make #482 and similar work easier:
code wrapping mpi4py for torch.Tensor's becomes an own module of heat
in heat/core/communication.py this module can be used
Diclaimer: feel free to discuss other ideas for refactoring and/or that status-quo does not need changes
The text was updated successfully, but these errors were encountered:
Currently, the code wrapping
mpi4py
fortorch.Tensor
's is contained inheat/core/communication.py
and not fully separated from constructions related to theDNDarray
-class (e.g., computation of chunk sizes is also introduced there).I would suggest to separate this clearly, which may also make #482 and similar work easier:
mpi4py
fortorch.Tensor
's becomes an own module of heatheat/core/communication.py
this module can be usedDiclaimer: feel free to discuss other ideas for refactoring and/or that status-quo does not need changes
The text was updated successfully, but these errors were encountered: