Exascale machines and clouds are now available, based on several different architectures and arithmetic. Supercomputing and their programming paradigms are no longer only lead by historical computational science applications. Brain-scale applications, such as those arising from machine learning and AI manipulate very large datasets that often lead to distributed and parallel linear algebra problems (such as very sparse non-symmetrical matrix computation, with dynamic patterns), resulting in performance far from peak ones. The notion of benchmarking and performance evaluation, and their evolution, must be adapted to these modifications and to the emergence of new applications from domains which would dominate a large part of the future hardware and software industry worldwide

Therefor AI, including LLM and machine learning applications in general, is currently redesigning the architectures and the distributed and parallel programming approaches: from the chip design, the new arithmetic, the new accelerators to new data structures and programming paradigms, for example. Nevertheless, linear algebra and supercomputing expertise are still required. Convergence between those domains is not guaranteed and we need to build bridges and be able to propose interoperability allowing to avoid separate roadmaps and ecosystems. Decade long researches on distributed and parallel computational science, linear algebra and middleware may be adapted and optimize these new application deployments.

On order to discuss on these challenges, we organize a two-day workshop at Ecole de Mines de Paris to bring together scientists in the fields of HPC, AI and Linear Algebra, or involved in associated applications. The program is composed of Keynote talks, dedicated sessions on distributed and parallel programming programing for AI methods, added with a panel and a general discussion on benchmarking and performance evaluation.

Program

Mines Paris - PSL University

Room: V107

Monday, June 16th

• Opening (09:30 - 10:00)
  Paolo Stringari, Director of Research at Mines Paris
  Corinne Ancourt, Senior Research Scientist
  Serge Petiton, Professor
  [Slides]
• Keynote: Subject. (10:00 - 10:45)
  Jack Dongarra
  Chair: Serge G. Petiton
  [Abstract]
  Abstract ....
  [Bio]
  Jack Dongarra is ...
  [Slides]

• Eigenvalue, dimension reduction.
  Nahid Emad.
  [Abstract]
  Abstract ....
  [Slides]
• Mixed accuracies (.
  Mines + Perpignan.
  [Abstract]
  Abstract ..
  [Slides]
• Sparse matrix computation .
  Nvidia ?
  [Abstract]
  Abstract ..
  [Slides]

• Keynote: Subject. (14:00 - 14:45)
  keynote 2 name
  Chair: xxxx
  [Abstract]
  Abstract ....
  [Bio]
  the keynote is ...
  [Slides]
• title 1. (14:45 - 15:10) [20min + 5min Q&A]
  Author names.
  [Abstract]
  Abstract ...
  [Slides]
• Title2. (15:10 - 15:35) [20min + 5min Q&A]
  Author names.
  [Abstract]
  Abstract ...
  [Slides]

• Title xx.
  Author names.
  [Abstract]
  Abstract ...
  [Slides]
• Ttile yy.
  Authors.
  [Abstract]
  Abstract
  [Slides]

Tuesday, June 17th

• Welcome (09:00 - 9:15)
  XXXXX
  
• Keynote: Subject. (9:15 - 10:00)
  Names 3
  Chair: xxxx
  [Abstract]
  Abstract ....
  [Bio]
  Name 3 is ...
  [Slides]

• Title xxx.
  Author names.
  [Abstract]
  Abstract.
  [Slides]
• Title.
  Author names.
  [Abstract]
  Abstract
  [Slides]
• Title.
  Author names.
  [Abstract]
  Abstract
  [Slides]

• [Slides]

How to Get from Orly or Roissy-Charles de Gaulle (CDG) Airports to the École des Mines, 60 Boulevard Saint-Michel, Paris

The École des Mines is located in the heart of Paris, in the 6th arrondissement, near the Jardin du Luxembourg. Here are the ways to reach this address from Paris's main airports: Orly and Roissy-Charles de Gaulle (CDG).

• From Orly Airport:


• By Public Transport (Approximately 35 to 60 minutes):
• Option 1: Line 14 + RER B (New Route)
1. Take Line 14 (automatic metro) from Orly Airport (Orly 1-2-3 station) to Châtelet-Les Halles.
2. At Châtelet-Les Halles, transfer to the RER B (direction Mitry-Claye or Charles de Gaulle Airport).
3. Get off at the Luxembourg station.
4. The École des Mines is a 5-minute walk from the station.
5. Duration: ~35-40 minutes.
6. This is the fastest public transport option.
• Option 2: Orlyval + RER B
1. Take the Orlyval shuttle train to Antony.
2. At Antony, take the RER B (direction Paris, Mitry-Claye, or Charles de Gaulle).
3. Get off at the Luxembourg station.
4. Duration: ~40-50 minutes.
• Option 3: Orlybus + RER B
  1. Take the Orlybus from Orly Airport to Denfert-Rochereau.
  2. At Denfert-Rochereau, take the RER B (direction Mitry-Claye or Charles de Gaulle).
  3. Get off at the Luxembourg station.
  4. Duration: ~50-60 minutes.
• By Taxi or Rideshare (Approximately 30 to 45 minutes):
• Take a taxi or rideshare (e.g., Uber) directly from the airport to 60 Boulevard Saint-Michel.
• Approximate cost: €30–€40 (fixed rate for Paris's left bank).
• Travel time depends on traffic, generally 30–45 minutes.


• From Roissy-Charles de Gaulle Airport (CDG):


• By Public Transport (Approximately 45 to 70 minutes):
  1. Take the RER B from Charles de Gaulle Airport (follow signs for "Paris by train").
  2. Get off at the Luxembourg station.
  3. The École des Mines is a 5-minute walk from the station.
  4. Frequency: every 10–15 minutes.
  5. Duration: ~45 minutes.
• By Taxi or Rideshare (Approximately 45 to 60 minutes):
  • Take a taxi or rideshare (e.g., Uber) directly from the airport to 60 Boulevard Saint-Michel.
  • Approximate cost: €55–€65 (fixed rate for Paris's left bank).
  • Travel time depends on traffic, generally 45–60 minutes.

Scientific Coordination

	Serge Petiton	Lille University, France
	Jack Dongarra	Uiversity of Tenessee, USA
	Nakajima Kengo

Local Organizers

Corinne Ancourt

Mines Paris - PSL University, France

Contact Us

Please send any questions related to the DPPEIA'25 workshop to xxxxx