Sam Yates
Sam Yates
Some notes: * Probe id tags could be strings only in the Python interface, or we could make them strings in the C++ API too. For Python only, we'd associate...
In order to avoid twice-computing the currents in each integration time step, we need to be able to apply a 'current delta' on event delivery. Work on this is ongoing...
Having a look at what is required, I believe we need: * The `external_connections_` code in `communicator.hpp` (though I would keep the source unsigned). * `external_spike_callback` in `simulator` and implementation...
Oh, and I think we should extend the `external_spike_callback` interface so that it takes an argument for the minimum window of external events, i.e. a `t` such that any subsequent...
Might be a stretch goal, but I'd like to see this interface in the 0.2 release; adding the tag and we'll see how we go.
Removing 0.2 tag; will aim for 0.3.
I believe that its serial nature is not the fundamental problem: a multithreaded version of the existing approach would still be hamstrung by bad cache behaviour. We'll need to address...
Does indeed look like L2! We should run perf and get some stats.
My vague voice from beyond says: I'd like to keep the 'm' prefix because the terms segment etc. are so overloaded in the literature that leaving the 'm' off invites...
If we're aligning the APIs, we're going to have a version incompatibility regardless; perhaps it is better to break both in a way that will fail to compile (C++) or...