HEP-SCORE deployment TF meeting #28

Wednesday 18 May 2022, 16:00 → 17:00 Europe/Zurich

Zoomland

Helge Meinhard (CERN)

Minutes (approved) [PDF]

16:00 → 16:05 Welcome, note-taking, notes from previous meeting

Minutes (approved) of previous meeting [PDF]

16:05 → 16:15 Status of workloads

Speaker: Domenico Giordano (CERN)

1. Alice gen-sim-digi
   1. validated (see https://rjsobie.web.cern.ch/rjsobie/Workloads/fits/plots/alice_gen_sim.gif )
   2. included in the script "D" (with IGWN), circulated by Gonzalo few days ago in the TF Operation egroup
   3. Possible improvements
      1. the digi part is small and is not running all digi processes (due to a misconfiguration in the script)
         1. could be extended to all digi processes, or removed
      2. the WL report is a single number, the event throughput of the whole chain (gen-sim-digi). The report could be improved with a breakdown of the throughput in gen, sim and digi. This would enable further studies and comparisons with other WLs 
          
2. Alice reco
   1. Issues seen when using cvmfs snapshot, even outside the container (i.e. it's not a container issue)
   2. Alice experts are debugging the call to llvm which shows this problem. This requires to isolate some pieces of code (from Stefano Piano)
       
3. Atlas Real Reco:
   1. the DB snapshot has been provided.
   2. A first standalone container has been build. refinement is needed (ETA: this week)
   3. We can validate it in the following 7 days and, if validation is fine, make it available in 2 weeks for the matrix

16:15 → 16:25 Benchmark run campaign

Speaker: Gonzalo Menendez Borge (CERN)

2022-05-18_TF_Measurement_Collection_Update@TF.pdf

2022-05-18_TF_Measurement_Collection_Update@TF.pptx

CPUs_missing_entries-2022-05-18.png

Entries per CPU-data-2022-05-18 14_45_39.csv

16:25 → 16:35 Any other business

Workshop in autumn/fall

* September?

* As a pre- (or post-) GDB?