Andrew Gait comments

Results 59 comments of


                                            Andrew Gait

Generate parameters using a "virtual machine" execution

This is a nice idea which requires further consideration in the future. I think it might work relatively well with the delay / weight / synapse / matrix generation that...

set_number_of_neurons_per_core vs already existing Populations

I think I'd say no, because the set_number_of_neurons_per_core does specify which model (not Population) that you wish to do this for. So yes, I would iterate through all current Populations...

Gap junction implementation based on average population membrane voltage

Leaving this issue here for future consideration.

No checking for round to zero or too big for int16 when scaling STDP parameters

Looking at this a bit further, the overflow issue is known about already. I'm not sure anything can be done about it on the Python side that straightforwardly, as you...

get_average_spike_rate could be more efficient

Another good idea that will probably make it into a future release.

Allow user-defined weight scale / left shift value in ring buffer

Noting here that this should probably be dealt with at some point, but it doesn’t need to happen now.

Add C source code to Python install and add support scripts

I agree with Rowley here - option 2 as he sets it out would be a much easier way to support a user like Harry for a start...

Add C source code to Python install and add support scripts

> An alternative here is to have an additional pip install that has no python code to speak of. This would simply contain the source code of sPyNNaker/neural_modelling and the...

Change loading of connections to use CSA

Certain connections are now expanded on the machine (not using CSA). This is probably still an open issue for now in case we decide to implement CSA on-machine rather than...

Change loading of connections to use CSA

This is definitely not something to be considered for the upcoming release.