For chronic data, we want to run spike sorters only to the point they estimate templates. Then template matching can be massively parallelized.
The function could also take care and expose slicing options to detect peaks and do the slicing (e.g., what we currently do is to define a chunk_duration - 1m - and inter_chunk_duration - 24h).
For now, let's keep external parallelization separate for the template matching part.
If we want, we could use this existing function: https://github.com/SpikeInterface/spikeinterface/blob/main/src/spikeinterface/generation/hybrid_tools.py#L26
For chronic data, we want to run spike sorters only to the point they estimate templates. Then template matching can be massively parallelized.
The function could also take care and expose slicing options to detect peaks and do the slicing (e.g., what we currently do is to define a
chunk_duration- 1m - andinter_chunk_duration- 24h).For now, let's keep external parallelization separate for the template matching part.
If we want, we could use this existing function: https://github.com/SpikeInterface/spikeinterface/blob/main/src/spikeinterface/generation/hybrid_tools.py#L26