Eric Hunsberger

The short answer is that we're not perfect about estimating the number of axons that will be used by NxSDK, so sometimes our verification thinks the network is OK, but...

Note that this DOES have an effect even if not using the interface directly (i.e. even if doing everything through `nengo_loihi.Simulator`). I wasn't sure if it would. It also shows...

Thanks for tracking those down, @nmnaughton. I was aware of the one with the receiver queue but hadn't got around to fixing it yet, but I wasn't aware of the...

At some point, I'll have to look more into the actual memory leaks. One thing I'll note about your script is that while your `Simulator` object gets closed, it doesn't...

I tried measuring things with a modification of your script (below), which uses heapy to measure the memory usage. I'm not seeing any significant memory leaks (the difference between start...

Thanks @tbekolay, that's good to know. There is still the question of why @nmnaughton sees higher memory usage in `top` in case 1 versus case 4. In terms of things...

What's more important is that setting `on_chip` on the subnet does not appear to work. ``` with nengo_loihi.Simulator(net) as sim: def print_ensembles(name, model): if len(model.params) == 0: return n_ensembles =...

Yeah, related to that, though it's even a bit trickier because the splitting is pre-builder, so even though we're setting/unsetting `model.config` and thus have access to the current network config...

I think I encountered this when using an inappropriate NxSDK version, which didn't have `snip_maker`.

I think this is resolved in some of the networks we trained for the keyword spotting paper. However, this has not been incorporated in this repository, as evidenced by this...