There are a couple of bottle-necks in the filter step, abs_bkp and remap_micro.
These are only a problem in very noisy samples (e.g. 400k+ events on entry to abs_bkp) but can push the runtime into weeks/months so need to be investigated.
Likely that the filter step needs to be broken up as the individual elements in this case have variable resource requirements.
There are a couple of bottle-necks in the filter step,
abs_bkpandremap_micro.These are only a problem in very noisy samples (e.g. 400k+ events on entry to abs_bkp) but can push the runtime into weeks/months so need to be investigated.
Likely that the
filterstep needs to be broken up as the individual elements in this case have variable resource requirements.