Sparse waveforms for Spyking Circus 2 #1943
Conversation
… into factoring_omp
for more information, see https://pre-commit.ci
for more information, see https://pre-commit.ci
for more information, see https://pre-commit.ci
| mode = "folder" | ||
|
|
||
| sorting_folder = tmp_folder / "sorting" | ||
| sorting = NumpySorting.from_times_labels(spikes["sample_index"], spikes["unit_index"], fs) |
There was a problem hiding this comment.
This is faster and should be the default when we already have a spike vector.
Note the you need to given also unit_ids
| sorting = NumpySorting.from_times_labels(spikes["sample_index"], spikes["unit_index"], fs) | |
| sorting = NumpySorting(spikes, fs) |
There was a problem hiding this comment.
@yger this is the only change that is required! Can you add the unit_ids?
There was a problem hiding this comment.
Done. However, I have a question for the new waveform speedup, with single or multi buffers. @samuelgarcia @alejoe91 shouldn't this functionnality be exposed at the extract_waveforms() level?
There was a problem hiding this comment.
Yes, this is the plan but not now.
And WaveformEtractor will be refactor a lot anyway not even sure this will be done before the big refactoring.
|
ok for me. |
for more information, see https://pre-commit.ci
|
@samuelgarcia I've checked the behaviour, and this is working fine, so I think you can proceed with merging this PR |
PR to provide the support for sparse waveforms in circus 2, and factorized matching engine. The internal machinery now uses the sparsification mecanisms provided by SI, and thus we should be able to benchmark properly that among sorters