Dear all,
There were so many discussions and feature requests that I finally put together a module for delay continuous variables (aka summed synapses). This module (seems) works for numpy, cython, and cpp-standalone implementations. This code follows a very helpful discussions here, here, and here.
I ran a few performance assessments. For randomly connected 10000 neurons with a 30% probability of connections, the performance is good, but not excellent:
| Numpy | Cython | CPP standalone | |
|---|---|---|---|
| With delays | 16m 19s | 6m 7s | 3m 48s |
| Without delays | 7m 5s | 1m 39s | 37s |
The code on GitHub, but I don’t have time to write any documentation, run better benchmarks, and so on. I’ll truly appreciate if someone finishes this module (setup.py, readme.md, MANIFEST.in, etc.) and puts it on pypi for the community.
-rth