One parameter we can potentially tune is the event chunk size here:
Because the hits are padded to the maximum number of the prompt hits of any event in this chunk. I think it would be helpful to have smaller chunk size.
My naive understanding is that
these three parameters here are not independent.
I think the minimum time time between hits are ~15ticks (please correct me), so it won't merge more than 5 merge steps, and
max_contrib_segments can be 5 x 10 (where 5 is the merge steps and 10 is placeholder for the number of stored segments per prompt hit)
One parameter we can potentially tune is the event chunk size here:
ndlar_flow/yamls/proto_nd_flow/workflows/charge/final_calibration.yaml
Line 22 in f58217e
Because the hits are padded to the maximum number of the prompt hits of any event in this chunk. I think it would be helpful to have smaller chunk size.
My naive understanding is that these three parameters here are not independent.
I think the minimum time time between hits are ~15ticks (please correct me), so it won't merge more than 5 merge steps, and
max_contrib_segmentscan be 5 x 10 (where 5 is the merge steps and 10 is placeholder for the number of stored segments per prompt hit)