Closed texadactyl closed 2 years ago
This is not necessary and seems to slow down find_et by adding extra cupy/cuda overhead.
When calling run_pipeline directly, an attach must be made.
When calling run_pipeline via find_et, the attach should be made only once.
This is not necessary and seems to slow down find_et by adding extra cupy/cuda overhead.
When calling run_pipeline directly, an attach must be made.
When calling run_pipeline via find_et, the attach should be made only once.