[hpc-announce] Call for Participation: Open Day at the Joint Laboratory on Extreme Scale Computing Virtual Workshop

[George Bosilca]

Dear HPC community,

We are delighted to announce the OpenDay of the 12th workshop of the
Joint-Laboratory on Extreme Scale Computing (JLESC): February 26, starting
@ 8:00AM EST via Zoom. Unlike the rest of the workshop, the participation
in the OpenDay is open to any attendee without registration or registration
fees.

The OpenDay is one of the 3 days of the JLESC workshop and features 3
invited talks from leaders in the field presented along with 3 success
stories from JLESC teams. The OpenDay speakers and topics are:

Invited talk 1: Fugaku: the first 'Exascale' supercomputer
Prof. Satoshi Matsuoka, Riken CCS

Invited talk 2: How a Community Software Ecosystem Perspective Helps to
Advance Science Goals in the Exascale Computing Project
Dr. Lois Curfman McInnes, Argonne National Laboratory

Invited talk 3: High-Performance Deep Learning
Prof. Torsten Hoefler, ETH Zurich

Success story 1: A Story About Data: Advancing Storage, I/O and Processing
at Challenging Scales
Dr. Gabriel Antoniu, INRIA

Success story 2: Resilience for Extreme Scale Computing
Dr. Leo Bautista, BSC

Success story 3: Developer tools for porting & tuning parallel applications
on extreme-scale systems
Dr. Brian Wylie, JSC


Please see below for the abstracts of our invited talks and success
stories. More information, including the bio of the speakers, as well as
the schedule for the Open Day can be found on the workshop webpage.

The timing of the OpenDay has been selected to accommodate a simultaneous
participation from America, Europe, and Asia. The schedule and virtual
conferencing link to attend the OpenDay will be published on:
http://icl.utk.edu/jlesc12/



About the JLESC:

The purpose of the JLESC is to be an international, virtual organization
whose goal is to enhance the ability of member organizations and
investigators to make the bridge between Petascale and Extreme computing.
The founding partners of the JLESC are INRIA and UIUC. Further members are
ANL, BSC, JSC and R-CCS. UTK is an associate member.

JLESC involves computer scientists, engineers and scientists from other
disciplines as well as from industry, to ensure that the research
facilitated by the Laboratory addresses science and engineering's most
critical needs and takes advantage of the continuing evolution of computing
technologies. Find out more at https://jlesc.github.io/.



The JLESC community is looking forward to welcoming you at the 12th JLESC
workshop OpenDay!





Invited talks:

Invited talk 1: Fugaku: the first 'Exascale' supercomputer

Prof. Satoshi Matsuoka, Riken CCS


Abstract: Fugaku is the first ‘exascale’ supercomputer of the world, not
due to its peak double precision flops, but rather, its demonstrated
performance in real applications that were expected of exascale machines on
their conceptions 10 years ago, as well as reaching actual exaflops in new
breed of benchmarks such as HPL-AI. But the importance of Fugaku is
“applications first” philosophy under which it was developed, and its
resulting mission to be the centerpiece for rapid realization of the
so-called Japanese ‘Society 5.0’ as defined by the Japanese S&T national
policy. As such, Fugaku’s immense power is directly applicable not only to
traditional scientific simulation applications, but can be a target of
Society 5.0 applications that encompasses conversion of HPC & AI & Big Data
as well as Cyber (IDC & Network) vs. Physical (IoT) space, with immediate
societal impact with its technologies utilized as Cloud resources. In fact,
Fugaku is already in partial operation a year ahead of schedule, primarily
to obtain early Society 5.0 results including combatting COVID-19 as well
as resolving other important societal issues and also go into full
production in moments time.





Invited talk 2: How a Community Software Ecosystem Perspective Helps to
Advance Science Goals in the Exascale Computing Project

Dr. Lois Curfman McInnes, Argonne National Laboratory


Abstract: Teams in the U.S. Exascale Computing Project (ECP) are working
toward scientific advances on forthcoming exascale platforms, across a
diverse suite of applications in chemistry, materials, energy, Earth and
space science, data analytics, optimization, artificial intelligence, and
national security. In turn, these applications build on software
components, including programming models and runtimes, mathematical
libraries, data and visualization packages, and development tools that
comprise the Extreme-scale Scientific Software Stack (E4S).  E4S represents
a portfolio-driven effort to collect, test, and deliver the latest in
reusable open-source HPC software products, as driven by the common needs
of applications. E4S establishes product quality expectations and provides
a portal as a starting point for access to product documentation. This
presentation will discuss early experiences with how this software
ecosystem approach delivers the latest advances from ECP software
technology projects to applications, thereby helping to overcome software
collaboration challenges across distributed aggregate teams. A key lesson
learned is the need for close collaboration between teams developing
applications and reusable software technologies, as well as the need for
crosscutting strategies to increase developer productivity and software
sustainability, thereby mitigating technical risks by building a firmer
foundation for reproducible, sustainable science.





Invited talk 3: High-Performance Deep Learning

Prof. Torsten Hoefler, ETH Zurich


Abstract: Deep Learning is as computationally expensive as the most
challenging scientific computing applications. In this talk, we outline the
biggest challenges in training deep learning workloads and show how HPC
techniques can be used to improve the performance of training workloads. We
focus on model sparsity in the training process. This will be even more
important once the scientific computing community uses deep learning in
their workflows.





Success story 1: A Story About Data: Advancing Storage, I/O and Processing
at Challenging Scales

Dr. Gabriel Antoniu, INRIA


Abstract: Looking back over more than 10 years of collaboration within
JLESC involving Inria, the University of Illinois at Urbana-Champaign and
Argonne National Lab, this talk will highlight a few achievements on hot
topics related to data storage, I/O management and in situ visualisation
and processing. From the initial challenges in this areas posed by the
expected arrival of Exascale systems, new ones emerged as frontiers started
to blur between High-Performance Computing and Big Data analytics. We will
also discuss upcoming open problems triggered by the increasingly complex
workflows that are mixing simulations, analytics and AI, which emphasize
new requirements and opportunities created by their potential execution on
the HPC/Cloud/Edge computing continuum.





Success story 2: Resilience for Extreme Scale Computing

Dr. Leo Bautista, BSC


Abstract: Resilience has been one of the main research topics of the JLESC
since its conception over a decade ago. We have covered multiple types of
failures and errors which has led to completely different fault tolerance
techniques, some of them at the intersection of HPC and ML. The research
work, carried out by JLESC researchers from five different institutions,
shows a strong interaction between theoretical analysis and practical
implementations. The results of this endeavor had led to multiple
collaboration visits, dozens of publications and hundreds of citations; but
more interestingly, it has opened new questions and it has shown
connections between HPC fields that we didn't know were connected before.
In this talk we will go over this trajectory and get a quick glance of what
might come in the future for HPC resilience.





Success story 3: Developer tools for porting & tuning parallel applications
on extreme-scale systems

Dr. Brian Wylie, JSC


Abstract: Application developers targeting extreme-scale HPC systems such
as Fugaku, and modular supercomputing architectures such as JUWELS, need
effective tools to assist with porting and tuning for these unusual
systems. This collaborative project brings together developers of such
tools from JLESC partners to investigate their integration and support
joint training activities as the tools are deployed and applied to a
variety of application codes.