Hi,
I am using KS 3.0 for spike sorting of very long (days) Neuropixels recording. I am using quite a strong machine ~400GB RAM, 2 GPUs, and 20 CPUs, however, it seems that KS does not utilize the power the computer offers. I was wondering if there are some processes currently running serially that can speed up by utilizing parallel computing. I am not familiar enough with the algorithm of KS, but if there are processes that work on distinct batches of recording, these can be paralleled. If so, I will be happy to take part in a programming team to make it happen. Regardless, any suggestion to reduce the sorting time of neuropixels data will be appreciated (e.g. is it really necessary to high-pass filter NP data?)
Hi, I am using KS 3.0 for spike sorting of very long (days) Neuropixels recording. I am using quite a strong machine ~400GB RAM, 2 GPUs, and 20 CPUs, however, it seems that KS does not utilize the power the computer offers. I was wondering if there are some processes currently running serially that can speed up by utilizing parallel computing. I am not familiar enough with the algorithm of KS, but if there are processes that work on distinct batches of recording, these can be paralleled. If so, I will be happy to take part in a programming team to make it happen. Regardless, any suggestion to reduce the sorting time of neuropixels data will be appreciated (e.g. is it really necessary to high-pass filter NP data?)
Thanks in advance Anan Moran Tel Aviv University