buffer size

[samuelgarcia]

@fpbattaglia @Claybarn

Hi all,
I manage to make my first python plugin and connect realtime tridesclous spike sorting engine.
I have a problem to solve : I need to known the buffer shpae before it start to preallocate buffer on GPU.

In short the def bufferfunction() callback function give an array (n_chan, n_sampl).
n_chan is known at startup but is there a way to known n_sampl before it start (before the frist bufferfunction call) ?

Having a short look at code https://github.com/NeuroNetMem/PythonPlugin/blob/master/PythonPlugin/PythonPlugin.cpp#L146

It appear that this n_sampl is known only when a buffer arrive.
But the getNumSamples(0) suggest that we can know in advance the big buffer size (all channels) and then maybe we could anticipate this n_sampl.

Thanks for the help, this detail is quite important for my need.

And congratulation again for this piece of code.

Samuel

[fpbattaglia]

Hi Sam This has to do with the open ephys side of things and I see that Aaron replied to you on their google groups. In fact, what I see is that most of the times the buffer has its maximum size, and it’s only under special circumstances (eg when stopping acquisition) that you see a smaller buffer. Probably wise to check buffer size in the plugin, and pad it with zeros in case it’s smaller, but that should happens very rarely. I don’t know what happens when CPU load gets very high though (which shouldn’t happen anyway) Cheers, Francesco

…

On 19 May 2020, 09:53 +0200, Garcia Samuel ***@***.***>, wrote: @fpbattaglia @Claybarn Hi all, I manage to make my first python plugin and connect realtime tridesclous spike sorting engine. I have a problem to solve : I need to known the buffer shpae before it start to preallocate buffer on GPU. In short the def bufferfunction() callback function give an array (n_chan, n_sampl). n_chan is known at startup but is there a way to known n_sampl before it start (before the frist bufferfunction call) ? Having a short look at code https://github.com/NeuroNetMem/PythonPlugin/blob/master/PythonPlugin/PythonPlugin.cpp#L146 It appear that this n_sampl is known only when a buffer arrive. But the getNumSamples(0) suggest that we can know in advance the big buffer size (all channels) and then maybe we could anticipate this n_sampl. Thanks for the help, this detail is quite important for my need. And congratulation again for this piece of code. Samuel — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub, or unsubscribe.

[samuelgarcia]

Hi Francesco,
thanks for the anwser.
At the moment in tridesclous with GPU buffer has to be fixe size.

My plan is to add a processing step that equalize buffer size. Make it smaller when it is too big. Concatenate with following buffer when it is too small.

Cheers