Public Minutes of 124th Steering Board, 10th May 2021

Present: V. Ivanchenko, W. Pokorski, M. Asai, M. Verderi, A. Ribon, J. Allison (OB Liaison), L. Pandola, D. Sawkey,  A. Howard, I. Hrivnacova, S. Okada, B. Morgan, D. Wright, G. Cosmo, L. Garnier, M. Novak, G. Folger, F. Romano, S. Y. Jun

Excused: K. Murakami

1. Situation wrt pandemic

No specific difficulties have to be reported.

There are clear progress with the vaccination, however, the remaining uncertainties about the pandemic evolution and the difficulties to anticipate what will be the sanitary rules and travel rules across Geant4 participating countries made us to decide to go for an other virtual edition of the Collaboration Meeting (see below).

2. Regular items

A round-table of working group activities was made, as well as a review of pending requirements, SB actions and bug reports. During the round table, the preliminary and recent achievement of electromagnetic shower processing on GPU was reported.

About performance monitoring, the Fermilab system has been moved to a new institutional cluster (mainly, Intel ivy bridge) and a series of performance measurement for release prior to 10.7 redone, to allow normalization.

3. Report from Tasking mini-workshop (26th of March & 22nd of April)

The mini-workshop had two main items : discuss any remaining points needed to make the tasking becoming the default processing mode for this year release, and discuss tasking “by-products”, that is features that the tasking make it possible or make it easier (sub-event parallelism, bridge to hybrid computing, …).

About the first item,  no show stopper have been identified so far and the “tasking” is now the default processing mode since G4 10.7.ref04, in substitution to the “Multi-Threading” one. Massive migration of examples should be ready to start, and any problem should be reported to .

Regarding the second item, discussions were preliminary and will continue; they mostly focused on hybrid computing. The tasking by-products are not planned for this year.

4. Status of 11.0-beta

No major issues are faced so far.

The beta release, geant4-11-00-beta, is planned for June 25th 2021. It is reminded that all developments should be provided for May 28th.

5. Collaboration Meeting Preparation

It has been agreed  to go for another virtual meeting, to avoid surprises related to the Covid and related sanitary rules.

The format will be similar to the one of last year, with one week for plenary sessions (20-24 September) at the same time slots than last year (few hours each afternoon, CERN time), and one additional week for parallel sessions, the week before. The parallel sessions will be organized by the Working Group coordinators. The two weeks will appear under the same Indico page.

To save time for the plenary sessions, some educational items -C++ 17, Jekyll, etc.- will be exposed well before the meeting under the form of “internal seminars”. These will be announced shortly; a C++17 seminar is expected for June.

It has been agreed also to keep the same location (Rennes) for the Collaboration meeting in 2022.

6.  HL-LHC Computing Review Stage-2: Common Software Projects

This review charge states : “The second review will take place in the week of November 1 - 5, 2021 with the intention of an in-person review at CERN. (…) The main goal of this review is to ensure the experiments, WLCG, and the relevant software projects, have common and realistic expectations of requirements and timescales.”

The Collaboration has been requested to conduct the writing of a support document for what the detector simulation is concerned. A mini-workshop took place recently ( as seed of this process.

7. Events

  • Past events
  • Near future events
    • [Virtual] CHEP,  17-21 May 2021
      • No contribution from G4
    • [Virtual] Geant4 Beginners Course @ CERN, 25-31 May 2021
      • Fully booked
    • OB meeting, 14th of June
    • [Virtual] LPCC FastSim Workshop, 22 & 23 November 2021
      • This will be the first time for this meeting, with classic (i.e. parametrized) and ML shower simulations.

8. AOB

  • The KEK team has performed first performance tests for A64FX and Apple silicon (M1 chip). These are preliminary at that time and further work is needed, but the first observations are that the A64FX performances are disappointing compared to Xeon CPUs (about a factor 5 slower), while the ones one on the M1 chip are encouraging, with a speed-up by a factor about 2.
  • A new support for automatic code formatting using clang-format has been exposed and will be an item of discussion at the Collaboration Meeting
  • The HyperNews support has been finished at SLAC, a decision about what to do with the old messages is ongoing.

Meeting Date

Monday, May 10, 2021 - 12:00