Denis Alevi

Alright, I can't stop thinking about supporting concurrent kernel execution. It is such an obvious next step in brian2CUDA development, which would increase performance for many simulations, especially those with...

In the `synapses` kernel we have a `thread synapse` correspondence, where in heterogeneous delay mode the synapses are sorted by delays and in homogeneous delay mode by synapse IDs. Therefore...

Also check again what I wrote in #30

Special case to take into account: If there is only one synaptic pathway direction (either pre->post or post->pre, not both), then all synapses should just be relabeled once they are...

Any sorting of synaptic variables would require some mechanism for sorting variables (and potentially monitor recordings) back into the format expected on Python side, see [my comment in #271](https://github.com/brian-team/brian2cuda/issues/271#issuecomment-1057254828).

Hi @bdevans, did you want to open this issue or was this a mistake while creating your Project (each item in your list opened an issue in the respective repository)?...

1. You seem to have used the version that was used for the consistency tests (that means with `TimedArray` implementations of the randomness). Not sure what the effect is on...

Another special case: - 1-to-1 connectivity (or more generally: every post neuron is targeted at most by 1 synapse) Here, we could store the target variable (e.g. post voltage) directly...

My suggestion is for the special case of 1-to-1 connectivity. Here, we just sort the "main array" in the way the connectivity matrix is sorted (which we can do since...

> > My suggestion is for the special case of 1-to-1 connectivity. Here, we just sort the "main array" in the way the connectivity matrix is sorted (which we can...