Skip to content

README.md

Latest commit

 

History

History
58 lines (49 loc) · 3.84 KB

README.md

File metadata and controls

58 lines (49 loc) · 3.84 KB

Run container

docker run --rm -it -v : --gpus all fullpursuit:latest flags: --rm: removes container once it's stopped -it: for interactive session (-d for detached) -v: mounted volumes (directories) --gpus: enables GPU use within container

Build this image

Build default image: docker build -t fullpursuit-base:latest .

Note: Full binary pursuit relies on opencl. Installation varies by OS and CPU / GPU type. For Intel CPUs, see https://www.intel.com/content/www/us/en/developer/articles/tool/opencl-drivers.html. For configuration on Ubuntu and Docker, see also https://linuxhandbook.com/setup-opencl-linux-docker/. On Ubuntu, you may need to add your user to the video and render groups to access the GPU.

sudo gpasswd -a $USER render
sudo gpasswd -a $USER video

Then log out and back in to apply the changes: su - $USER

Check fullpursuit version: docker run --rm -it spikeinterface/fullpursuit-base:latest bash -c "python -c 'import spikesorting_fullpursuit; print(spikesorting_fullpursuit.version)'"

Check that pyopencl can detect your graphics card: docker run --rm -it --gpus all spikeinterface/fullpursuit-base:latest bash -c "python -c 'import pyopencl as cl; platforms = cl.get_platforms(); print(platforms[0].get_devices(cl.device_type.GPU))'"

Test

Adapted from the readme file's instructions.

First make the voltage data and save ground truth (mounting the host data folder to the docker data folder):

docker run --rm -it -v <host-data-folder>:<docker-data-folder> --gpus all spikeinterface/fullpursuit-base:latest bash -c "python /src/fullpursuit/demos/make_and_save_voltage.py <docker-data-folder>/test_voltage.npy <docker-data-folder>/test_ground_truth.pickle"

e.g.:
Create the data folder: mkdir -p /home/$USER/data/fullpursuit_data Then create the groundtruth data: docker run --rm -it -v /home/$USER/data/fullpursuit_data:/data --gpus all spikeinterface/fullpursuit-base:latest bash -c "python /src/fullpursuit/demos/make_and_save_voltage.py /data/test_voltage.npy /data/test_ground_truth.pickle"

Then run the sorting algorithm on the generated synthetic data:

docker run --rm -it -v <host-data-folder>:<docker-data-folder> --gpus all spikeinterface/fullpursuit-base:latest bash -c "python /src/fullpursuit/demos/test_demo_run_sort.py <docker-data-folder>/test_voltage.npy <docker-data-folder>/sorted_demo.pickle"

e.g.:
docker run --rm -it -v /home/$USER/data/fullpursuit_data:/data --gpus all spikeinterface/fullpursuit-base:latest bash -c "python /src/fullpursuit/demos/test_demo_run_sort.py /data/test_voltage.npy /data/sorted_demo.pickle"

Run automated postprocessing to place sorter output into a dictionary sorted neurons:

docker run --rm -it -v <host-data-folder>:<docker-data-folder> --gpus all spikeinterface/fullpursuit-base:latest bash -c "python /src/fullpursuit/demos/test_demo_postprocessing.py <docker-data-folder>/sorted_demo.pickle <docker-data-folder>/sorted_neurons.pickle"

e.g.:
docker run --rm -it -v /home/$USER/data/fullpursuit_data:/data --gpus all spikeinterface/fullpursuit-base:latest bash -c "python /src/fullpursuit/demos/test_demo_postprocessing.py /data/sorted_demo.pickle /data/sorted_neurons.pickle"

Finally run a quick script that compares sorted results to the ground truth generated data and make a couple figures:

docker run --rm -it -v <host-data-folder>:<docker-data-folder> --gpus all spikeinterface/fullpursuit-base:latest bash -c "python /src/fullpursuit/demos/test_demo_results.py <docker-data-folder>/sorted_neurons.pickle <docker-data-folder>/test_ground_truth.pickle"

e.g.:
docker run --rm -it -v /home/$USER/data/fullpursuit_data:/data --gpus all spikeinterface/fullpursuit-base:latest bash -c "python /src/fullpursuit/demos/test_demo_results.py /data/sorted_neurons.pickle /data/test_ground_truth.pickle"