NEST Hackathon 2025/06

23 Jun 2025, 09:00 → 26 Jun 2025, 23:59 Europe/Oslo

Alpha (NMBU, Ås, Norway)

Hans Ekkehard Plesser, Susanne Kunkel

This application-oriented hackathon is an in-person event allowing NEST users to bring their ideas for new NEST features and get started implementing them with the help of experienced NEST developers.

Please note that while participation is free of charge, costs for traveling, accommodation, food/drinks, etc. are at your own expense.

Monday 23 June

1 Welcome & planning

Potential topics to discuss:

• Structural plasticity
  • NEST / Node refactoring (Abigail, Johanna, Parul, Camillo, Jan, Tim, Ady, Susanne)
  • IPU (Cath, Ady, Parul)
• JEDI/JUPITER thread scaling (HEP, Markus, Guido, Alp, Jan, Pooja, Susanne, (Johanna))
  • Investigate origin of 3-4 socket catastrophic threading. Start with ci-bennch Fig. 3B reprodction on hambach also to exclude JURECA variability problems. What is a good tool to investigate false sharing of memory, or cache collisions
  • JURECA jemalloc variability problem. Exclude hardware settings using hambach
• PyNEST-NG / Porting tests to PyTest
• cMAM and EXABRAINPREP (Renan, Lies, Cath, Melissa, Markus, HEP)
• review PR on documentation of timers including graphics on state propagation cycle (Jan, Melissa, Jessica)
• look at status of long e-prop simulations -> plenary
• kernel doc (Alp, Jessica, Susanne) -> ???
• PRs/issues meetings /format of regular meetings
• NESTML (Camillo, Pooja)
• Alexandria S.C. (Elide, Robin)

2 PyNEST-NG

plenary

12:30 Lunch

3 review PR on documentation of timers including graphics on state propagation cycle

(Jan, Melissa, Jessica)

4 NESTML

(Camillo, Pooja)

5 Barbecue at Ås

Tuesday 24 June

6 Structural plasticity in NEST / Node refactoring

plenary

7 NEST Initiatives General Assemblies

12:30 Lunch

8 look at status of long e-prop simulations

plenary

9 cMAM and EXABRAINPREP

(Renan, Lies, Cath, Melissa, Markus, HEP)

10 NEST Conference Debriefing

Wednesday 25 June

11 JEDI/JUPITER thread scaling

plenary

12 Eprop code staring

Speakers: Agnes Korcsak-Gorzo, Camilo Jara Do Nascimento (University of Sussex), Hans Ekkehard Plesser

13 IPU SP

(Cath, Ady, Parul, Susanne, (Marie))

14 IPU comm

(Cath, Jan, Markus, Susanne)

15 NESTML

(Johanna, Camilo/Parul, Pooja)

12:30 Lunch

16 Alexandria S.C.

plenary
(presentation by Elide, Robin)

Thursday 26 June

17 NEST Board (closed meeting)

(Abigail, Hans Ekkehard, Markus Susanne)

18 PRs/issues meetings /format of regular meetings

plenary

12:30 Lunch

19 ci-beNNch

Jan, (Anno) et al.

20 cMAM code staring

Speakers: Catherine Schofmann, Hans Ekkehard Plesser

21 Hackathon wrap-up

Plenary

22 Wrap-up

• Structural plasticity
  • NEST / Node refactoring (Abigail, Johanna, Parul, Camillo, Jan, Tim, Ady, Susanne)
    • bug happened between 3.5->3.6 (def branch merge)
    • segfault in deliver_events_  by SpikeData.get_lcid() or assertion triggered in connector (access with too large lcid)
  • IPU (Cath, Ady, Parul)
    • discussed requirements for Ady's and Parul's models
    • first steps on modelling and implementation side
    • be in communication
• JEDI/JUPITER thread scaling (HEP, Markus, Guido, Alp, Jan, Pooja, Susanne, (Johanna)
  • managed to get immediate access
  • in contact with Andreas Herten and an NVIDIA expert
  • hints on pinning
• PyNEST-NG / Porting tests to PyTest
  • merged some converted tests
  • Jessica auto-ported more tests
  • Robin reversed some extremely advanced test tech that was to error prone
• cMAM and EXABRAINPREP (Renan, Lies, Cath, Melissa, Markus, HEP)
  • definition prototype clustered connect rule to be implemented by Lies and Renan with help by Cath
  • discussion of implementation specifics
• review PR on documentation of timers including graphics on state propagation cycle (Jan, Melissa, Jessica)
  • last checks about to pass, then merge
• look at status of long e-prop simulations -> plenary
  • worked on bug fixed
  • worked on MPI problem -> not related to eprop
  • problem could be happening when reading from buffers
  • recorders doc: files written MPI ranks should be read by same ranks
• kernel doc PR (Alp, Jessica, Susanne) 
  
  • meeting after hackathon
• PRs/issues meetings /format of regular meetings
  • decided to move back to a scheme where we don't discuss projects but concrete PRs
• NESTML (Camilo, Ariel, Parul, Pooja)
  • intro of NESTML to new particpants 
  • two-compartment model in NESTML
  • now work on defining custom synapse model
• Alexandria S.C. (Elide, Robin)
  • thumps up
  • collaboration on doc and tests

17:00 Social in Oslo