2020 09 28 Open NEST Developer Video Conference
- Welcome
- Review of NEST User Mailing List
- Project team round
- In-depth discussion Benchmarking setups for CoSim and the NRP-NEST coupling (Kim Sontheimer and Cristian Jimenez-Romero)
- https://nest-simulator.org/mailinglist
- all threads answered since last meeting
Here we discuss topics that need broader attention, for example questions that came up but are outside a single project's scope, larger planned changes/PRs that affect all teams or pending work that is blocked by external factors.
- Models / NESTML
- no report
- PyNEST
- please have a look at #1773, note your ideas/comments in the issue
- Kernel
- nothing to report
- Installation
- nothing to report, closed couple of issues, still looking for MacOS developers.
- Infrastructure
- no report
- Documentation
- no major updates, cleaup and PRs as usual
- EBRAINS
- nothing to report
- (Feature) Random number generation
- still waiting, currently no updates
- (Feature) Automated Testing
- no report
Special guests are Kim Sontheimer and Cristian Jimenez-Romero of the SimLab Neuroscience (Jülich)
Introduction of the discussion by Kim Sontheimer
- In-transit coupling of NEST and Elephant, stream simulation output into a synchronous analysis with Elephant
- Current Implementation, MPI send spikeevents to in-transit server (sorting, transposition, …) coupled to down-stream analysis
- Plans are to scale up to test till which size of network the synchronous analysis can keep pace
- neurons per rank are known
- clarification of "real-time" meaning, that simulation time advances at the same speed as the wall-clock.
- which scalable model could be used for testing?
-
hpc_benchmarklong known - structured models available, but difficult to scale. need to keep reasonable activity
-
- real-time not relevant for analysis, but only for real-world applications like robotics, sound-click, …
Introduction of the discussion by Cristian Jimenez-Romero
- is a NEST simulator-simulator available for testing? – not necessary, computationally inexpensive simulations can create sufficiently large streams of data. NEST's dry-run mode is not yet usable for this in the new versions.
- NRP is free of real-time constraint. closed-loop simulations should be able to run as fast as possible in a simulated environment.
- interactions with NEST are coordinated also with NEST Desktop
