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
Is your feature request related to a problem? Please describe.
cudf.pandas currently falls back to pandas whenever an operation cannot be run on the GPU. The profiler is the typical approach for users to determine when this fallback happens. For developers, however, it will be useful to see fallback more immediately so that we can start patching differences between the libraries.
Describe the solution you'd like
We should add a mode in cudf.pandas to disable fallback to host and replace it with an error. Depending on the specifics of an investigation, we may want to disable fallback only in certain cases, for instance only when OOMs occur.
The text was updated successfully, but these errors were encountered:
Is your feature request related to a problem? Please describe.
cudf.pandas currently falls back to pandas whenever an operation cannot be run on the GPU. The profiler is the typical approach for users to determine when this fallback happens. For developers, however, it will be useful to see fallback more immediately so that we can start patching differences between the libraries.
Describe the solution you'd like
We should add a mode in cudf.pandas to disable fallback to host and replace it with an error. Depending on the specifics of an investigation, we may want to disable fallback only in certain cases, for instance only when OOMs occur.
The text was updated successfully, but these errors were encountered: