Technical Roadmap - Chapter 8

Wednesday 22 Jul 2026, 16:00 → 17:00 Europe/Zurich

31/2-028 (CERN)

Alessandro Di Girolamo (CERN), Doug Benjamin (Brookhaven National Laboratory (US)), Ianna Osborne (Princeton University), James Letts (Univ. of California San Diego (US)), Oxana Smirnova (Lund University), Stefano Piano (INFN (IT)), Tommaso Boccali (INFN Sezione di Pisa)

Chapter 8 - working draft

Notes to Frame the Discussion, and Live Notes

Meeting invitation:

On 19/06/2026 09:00, James R Letts wrote:

Dear Doug, Oxana, Ianna, and Stefano,

First of all, we’d like to welcome Stefano as one of the facilitators
for Chapter 8 of the WLCG Technical Roadmap on New Architectures and
Infrastructures. He brings extensive GPU development experience and
was recently a long-time computing coordinator of the ALICE
experiment, which is quite advanced in its utilization of GPUs. This
is part of a wider WLCG effort to make sure WLCG systems and
procedures are in place to integrate heterogeneous resources, which
Stefano will convene.

As part of this effort, WLCG management identified four key areas
where technical evolution is needed in order to be technically ready
to define GPU pledges by Run 4:
    • Benchmarking
    • Information system / accounting
    • Provisioning, and
    • Online and Offline Experiments’ Software and Workflows

Note that experiment requirements for GPUs are out of scope of this
activity, and there is no pressure from WLCG to ask experiments to
use GPUs, but we need to be technically ready when requests do
materialize.

Now of course, Doug, Oxana, and Ianna have been working on the
Chapter 8 draft for some months now, which can be found here [1], and
of course the scope is not strictly confined to GPUs but new
architectures and infrastructures generally. While the topics above
certainly factor into the chapter, there are other areas where focus
is needed (and here we have looked at [1] for reference), such as
cost vs. benefit studies (GPU processing relative to CPU, HPC vs.
Cloud vs. traditional Grid infrastructures), security-related
aspects, and very importantly Analysis Infrastructures and Analysis
Facilities, defining their role and purpose. Lately as well, we are
all aware of AI-related developments and discussions, and of course
the infrastructures needed for this (in WLCG, however) could also
greatly influence our deployment models.

We hope that this email serves as an introduction, and it would be
great if you four could meet to discuss Chapter 8, and as always we
are available to join if you would find it helpful. We aim to have a
more or less complete draft of the chapter in two months, in time for
OTF#12 in late August [2] so that it can be socialized and submitted
to the LHCC in Q3 of 2026.

Regards,

Ale & James

[1]
https://docs.google.com/document/d/1-KCdNvf1yD2SoshVBuPHEimVBxDy7YDAJOLac_JVAcw/edit?tab=t.0*heading=h.mzrg6omwpbwk

[2] https://indico.cern.ch/event/1683255/