Thomas Nowotny

Very interesting and looks like there is a lot of potential for improvement here. I didn't get the bit "which performs a prefix sum operation on the number of spikes...

Ah, yes of course - across the merged population group ... wouldn't the neuron kernels be able to make the sum via atomic add upon exit?

Sure - of course, but are there actually so many populations that that makes any significant difference I wonder? An atomic add per neuron population seems like a very small...

I was more thinking about the individual populations having a thread each doing the job, kind of like (not sure about the variable names, I hope it is clear what...

we can continue discussion offline - I am not entirely sure I understand how the 'shared library model approach' relates to this ... but I don't think I am aware...

Interesting idea ... and nice intro at this link. I assume that if the kernels are a little less trivial and need to get parameters passed which values change occasionally...

We should probably use one spike buffer for 1) delays, 2) output buffering like here, 3) MUSIC integration buffering

yes ... sounds good. How long do you think we should allow ourselves to run these? Would it be long enough to have some learning in the MBody one? For...

ok ... so on the MBody example, can we require between 1 and N_output/2 spikes for the last last 100 of the 50ms cycles? For the Izhikevich models, I would...

good point - using std (maybe x=3?) a lot more sensible than guessing +/- 10%