SpikeInterface / SpikeInterface/spikeinterface

Kilosort4 is much slower when run from Spikeinterface than from Kilosort 4 GUI

Open
#4,768 2 comments 0 reactions 1 assignee View on GitHub

@chrishalcrow is already working on this.

Since Sep 8, 2026.

performance sorters
Dominant language
Python
Stars
847
Forks
280
Avg merge
3d 9h
Merged PRs (30d)
29

Description

Hi all,
I have seen others mention this, but have not been able to find the solution.
When we sort a neuropixels recording with Kilosort 4 using run_sorter (default settings, but with delete_recording_dat = False) processing takes about 6 hours, but when we use the Kilosort 4 GUI (same settings) processing takes only about 2 hours. Why is this?

I believe that the GPU is used for processing: we had a faulty installation previously and CPU-based sorting takes about 4 days with the spikeinterface-based scripts. The difference in duration is striking nonetheless.
Is there anything I need to change in our functions (see attached) to make them perform 'normally' (like the GUI)?

I've tested if saving the preprocessed data (tem_wh.dat) helps speed things up, but this doesn't seem to make a difference.

Our system is:
HP Elite Tower 800 with Windows 11 pro
96GB RAM
GPU: Nvidia RTX 4000 Quadro
We generally run the our scripts from within an anaconda prompt.

clusterKSort_NPX.py

Contributor guide

No contributing guide indexed for this repository

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Assessment

This issue has not been assessed yet.

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.