nt0 is hard-coded

[WeissShahaf]

Hello, I've been experimenting with increasing the sample window nt0 to =81 in kilosort3 (but this issue probably exists in kilosort2+), as my spikes are very long. unfortunately, I've had to recursively change many lines in the code to do that, as the 61 value is assumed at various points in the code.

not sure I got it all.

perhaps you can fix this more easily than me.

thanks

[JonathanAMichaels]

Were you able to get the code working for nt0 > 61 @WeissShahaf? I changed all the hardcoded entries, but I'm getting CUDA errors. So, there may be aspects of the CUDA code that do not allow this.

[WeissShahaf]

81 worked for me. but i did have to change alot of code

[JonathanAMichaels]

I changed all locations where a "61" was hard-coded to ops.nt0, but these were only a few cases. Do you remember what else you had to change?

[WeissShahaf]

unfortunately no. sorry. i quit using kilosort since, maybe will return one day

[JonathanAMichaels]

@WeissShahaf do you still have that code lying around anywhere that you could send for me to look at? It seems that nt0 > 64 breaks mexGetSpikes2.cu

[WeissShahaf]

I used a value of 81 specifically.

Shahaf Weiss, PhD

Max Planck Institute for Brain Research

NSK - Laurent department Max-von-Laue-Str. 4 60438 Frankfurt am Main, Germany www.brain.mpg.de

@.***

@.***

Tel:+49 17657814689

On Fri, Jan 6, 2023, 23:45 Jonathan A Michaels @.***> wrote:

@WeissShahaf https://github.com/WeissShahaf do you still have that code lying around anywhere that you could send for me to look at? It seems that nt0 > 64 breaks mexGetSpikes2.cu

— Reply to this email directly, view it on GitHub https://github.com/MouseLand/Kilosort/issues/401#issuecomment-1374225813, or unsubscribe https://github.com/notifications/unsubscribe-auth/AK4J42D5KQXGUNF5BUPPZLTWRCOBRANCNFSM45CBUK4Q . You are receiving this because you were mentioned.Message ID: @.***>