[hpc-announce] Call for Demos / Papers: Workshop on High Performance Python for Science at Scale (HPPSS)

[Rollin Thomas]

Workshop on High Performance Python for Science at Scale (HPPSS)
https://hppss.github.io/SC23/
Sunday, November 12, 2023

In conjunction with The International Conference for High Performance
Computing, Networking, Storage, and Analysis (SC23)
November 12-17, 2023
Denver, Colorado, USA

*** Important Dates ***

* Submission opens: June 1, 2023
* Paper submission deadline: August 8, 2023
* Decisions: September 8, 2023
* Camera-ready papers: September 29, 2023
* Workshop: November 12, 2023 (Morning)

*** Call for Demos / Papers ***

HPPSS is accepting submissions that demonstrate the application of a technology
for greatly enhancing the performance and capability of Python at scale.
Technologies that are applicable to a wide range of use-cases and accessible
to a broad user base are of particular interest. Showing technologies applied
to specific use-cases can be very helpful for the audience’s understanding, but
be sure to show and discuss how they generalize.

Submissions can be made in either the Demo or Work-in-progress tracks. Both
tracks will feature work that is innovative and of significant interest to the
community:

* Demos: The Demo track is for more mature work and should be based on
  examples/code that is accessible to anyone in the audience. These submissions
  include a 4-page paper, demo recording, and presentation.
* Work-in-progress: The Work-in-progress track is for work that may not be
  ready for community adoption but demonstrates the direction and would benefit
  from community feedback. These submissions include an abstract and demo
  recording.

*** About the HPPSS Workshop ***

Python is the lingua franca of scientific discovery workflows. As the second
most popular language with an estimated 15 million world-wide developers in
2022, Python bridges the worlds of artificial intelligence, data science,
high-performance computing applications and web-application development.

In addition, modern and emerging science workflows make use of shell scripts,
databases, HPC simulation codes, AI frameworks, and data manipulation scripts,
stitched together and often highly customized to the configuration of a
particular system. Python, with the support of its massive user community,
includes the standard interfaces necessary to unify these disparate things into
a common interface for scientists. A single programmable interface for
edge-to-HPC that makes it possible to compose high-performance and scalable
workflows of distributed applications across languages would be a foundation
for the post Exascale era.

HPPSS aims to bring together the Python user community attempting to scale code
for HPC, AI and Data for science:

* Expose new developments in Python for HPC, AI, and Data at Scale to SC
  attendees
* Discuss advantages, disadvantages, challenges, and barriers for Python in
  scientific computing
* Identify improvements required for Python runtimes on future supercomputing
  architectures
* Survey tools/projects in the Python ecosystem for science
* Discuss programming models for HPC, cloud-native, and hybrid-native workflows
  using Python
* Bring together industry, academia, and national labs to identify requirements
  for distributed and parallel computing with Python

Submissions in the following areas are well-suited for HPPSS:
* Python for accelerated computing at scale
* Programming of federated/distributed workflows
* Data management between Python libraries, processes and with storage
* Performance via intelligent data management across memory hierarchies
* High-bandwidth/low-latency network performance for Python
* Multi-processing for on-node and off-node configurations
* Program execution including parallelization and asynchronous communication

*** Submission Instructions ***

* Submit at: https://tinyurl.com/submit-hppss-sc23
* Demo video submission: Video recordings of demos should be submitted through
  Zenodo. Recordings should include audio commentary describing the demo and
  what is being shown. Submissions accepted to the conference will have an
  opportunity to produce an updated video that does not include commentary. The
  presenter should be prepared to describe the demo video as part of their
  presentation.

*** Further Information ***

Visit https://hppss.github.io/SC23/ for more details about this workshop and
feel free to contact the organizing committee directly using the contact
information listed at the website.

-- 
Rollin Thomas
Programming Models and Environments Group
National Energy Research Scientific Computing Center
Lawrence Berkeley National Laboratory