HSF Generator WG Meeting

Europe/Zurich
Vidyo

Vidyo

Videoconference Rooms
HSF_Generator_WG
Name
HSF_Generator_WG
Description
HSF Physics Event Generator WG
Extension
10799316
Owner
Andrea Valassi
Auto-join URL
Useful links
Phone numbers
    • 16:00 16:05
      News and general matters 5m
    • 16:05 16:20
      ATLAS & CMS accounting updates 15m
      Speakers: Alexander Josef Grohsjean (Deutsches Elektronen-Synchrotron (DE)), Efe Yazgan (Chinese Academy of Sciences (CN)), Josh McFayden (CERN), Qiang Li (Peking University (CN)), Simone Amoroso (Deutsches Elektronen-Synchrotron (DE))

      CMS numbers in Table1   (Updates marked in Red)

      (1) Sample Statistics: 9.4->12.3B  (we missed to count some samples)

      (2) Reminder: Time/Evt numbers from Computing Experts

               G+S  = 50 sec /ev/core    - D: 10 sec/ev/ecore   - R: 25 sec/ev/core :      - Tot: 85sec/ev/core

      (3) GEN fraction: updates from https://indico.cern.ch/event/804266/

               Old GEN faction: (1.41*0.05+1.96*0.17+0.43*0.05+1.94*0.04+0.712*0.05+0.154*0.05+0.876*0.05+1.95*0.05)/9.4      ~7.3%
               New GEN faction:  (1.62*0.05+4.58*0.17+0.43*0.05+1.94*0.04+0.712*0.05+0.154*0.05+0.876*0.05+1.95*0.05)/12.2 ~9.3%

             where 0.05, 0.17 and 0.04 ....are from numbers achieved from local tests.

             Thus for GEN we have 85*9.3%~7.9 sec/ev/core.

              And the total GEN CPUs is  7.9*12.26B~96.9B

      (4)  In Table 1, we also provided GEN CPUs for each sample category:

             For example,  DY MG MLM~1.9s/event for GEN step based on local tests, thus the total CPUs is 1.9*1618M=3.07B.

      Summing over the numbers we get in Table1 for each sample category, we get 95.5B, which is quite close to 96.9B

       

      ATLAS numbers (Josh)

      (1) Changes to numbers since last time came from two places:

      - Double counting of a small number of samples

      - Accidental removal of samples that should be included

      --> Relatively small effect

      (2) HS06s vs s

      All the ATLAS numbers are raw seconds. Would be preferable to have in HS06seconds instead. Technically this is possible but will require relatively major update to the code, in the end it's just a matter of finding the time to do it.

       

       

       

    • 16:20 16:35
      Report from Les Houches discussions 15m
      Speaker: Josh McFayden (CERN)
    • 16:35 16:50
      ATLAS-CMS ttbar sample sharing 15m
      Speakers: Javier Fernandez Menendez (Universidad de Oviedo (ES)), Kyle James Read Cormier (University of Toronto (CA))
    • 16:50 17:00
      Proposal for PDGID extension 10m
      Speaker: Zach Marshall (Lawrence Berkeley National Lab. (US))
    • 17:00 17:05
      AOB 5m
      • Next meeting 1m