Workshop Program (June 3 - 4, 2025)
The conference will be held at Politecnico di Milano – located at Piazza Leonardo da Vinci 32, Milan 20133.Workshops and Tutorials will be held on Tuesday and Wednesday in Building 3 (Edificio 3), and main conference events will be held the last three days in Trifoglio (250m away).
Day One (June 3, 2025)
(Important: all regular talks are scheduled for ~20 minutes + 10 minutes for discussion)8:45 - 9:00 AM - Workshop Opening
- Workshop Opening. [talk slides] Dalibor Klusáček
9:00 - 10:30 AM - Session 1
- Power-Aware Scheduling for Multi-Center HPC Electricity Cost Optimization. [talk slides] Abrar Hossain, Abubeker Abdurahman, Mohammad Atiqul Islam and Kishwar Ahmed.
- Job Grouping Based Intelligent Resource Prediction Framework. [talk slides] Beste Oztop, Benjamin Schwaller, Vitus Leung, Jim Brandt, Brian Kulis, Manuel Egele and Ayse Coskun.
- Kubernetes Scheduling with Checkpoint/Restore: Challenges and Open Problems. [talk slides] Viktória Spišaková, Radostin Stoyanov, Lukáš Hejtmánek, Dalibor Klusacek, Adrian Reber and Rodrigo Bruno.
10:30 - 11:00 AM - Coffee Break
11:00 - 12:30 AM - Session 2
- Adaptive Carbon-Aware scheduling policies for HPC systems. Abdessalam Benhari and Denis Trystram.
- Resource elasticity for scientific platforms on HPC infrastructure. [talk slides] Maxime Martinasso and Thomas Schulthess.
- More for Less: Integrating Capability-Predominant and Capacity-Predominant Computing. [talk slides] Zhong Zheng, Michael Papka and Zhiling Lan.
12:30 - 2:00 PM - Lunch Break
2:00 - 4:00 PM - Session 3
- Workflow Batch Job Scheduling with Considering Task Dependencies. Kaito Yanai, Keichi Takahashi, Yoichi Shimomura and Hiroyuki Takizawa.
- Quality-Aware Energy-Efficient Scheduling of Moldable-Parallel Streaming Computations on Heterogeneous Multicore CPUs with DVFS. Sajad Khosravi, Sebastian Litzinger, Christoph Kessler and Jörg Keller.
- Optimizing Energy Efficiency in Heterogeneous Computing via Multi-Objective Scheduling with Reinforcement Learning. [talk slides] Ezgi Nur Alisan and Ismail Akturk.
- Static powercap vs EAR hard-powercap: Performance evaluation. Julita Corbalan and Lluis Alonso.
4:00 - 4:30 PM - Coffee Break
Day Two (June 4, 2025)
(Important: all regular talks are scheduled for ~20 minutes + 10 minutes for discussion)9:00 - 10:30 AM - Session 4
- Deep RC: A Scalable Data Engineering and Deep Learning Pipeline. [talk slides] Arup Kumar Sarker, Aymen Alsaadi, Alexander James Halpern, Prabhath Tangella, Mikhail Titov, Gregor Von Laszewski, Shantenu Jha, Mills Staylor, Niranda Perera and Geoffrey Fox.
- Fedsort: An Optimized Federated Scheduling Strategy for Cloud Workloads with Inter-task Dependencies. [talk slides] Suhas Gowda Harish, Sparsh Bk, Shresht V G, Meghana Thiyyakat and Prafullata K Auradkar.
- Evaluating the Impact of Algorithmic Components on Task Graph Scheduling. [talk slides] Jared Coleman, Ravi Vivek Agrawal, Ebrahim Hirani and Bhaskar Krishnamachari.
10:30 - 11:00 AM - Coffee Break
11:00 - 12:30 AM - Session 5
- Communication-balanced Job Allocation using SLURM. [talk slides] Gagandeep Mangat and Preeti Malakar.
- Performance Models to support HPC Co-Scheduling. [talk slides] Athanasios Tsoukleidis-Karydakis, Efstratios Karapanagiotis, Nikolaos Triantafyllis, Nectarios Koziris and Georgios Goumas.
- ELiSE: A tool to support algorithmic design for HPC co-scheduling. [talk slides] Efstratios Karapanagiotis, Nikolaos Triantafyllis, Athanasios Tsoukleidis-Karydakis, Georgios Goumas and Nectarios Koziris.
12:30 - 2:00 PM - Lunch Break
2:00 - 3:00 PM - Keynote
How to make the ultimate goal of energy-efficient data centers a reality. [keynote slides] Julita Corbalán (Barcelona Supercomputing Center)
Abstract: Having energy-efficient data centers is a key topic in the agenda of Europe for the current EuroHPC data centers but even most important for the coming Exascale EuroHPC supercomputers and new AI factories.
To reach the ultimate goal of having the most efficient possible computing resources, lot of requirements have been introduced in the procurements.
These requirements have to be fulfilled from hardware vendors to win the contracts. However, once the systems are up and running, what can we do to guarantee the systems are used in a really efficient way and not only reporting the highest scores when running HPL?
During the talk we will comment about the current challenges to make energy-efficient data centers a reality. We will comment most from three points of view: first from a software perspective, second from the final user perspective, and finally from the system management point of view.
Traditional HPC policies and metrics must be deprecated and must give place to new metrics and criteria to redefine what a "good" or "bad" job is, new criteria for job accounting, new strategies for job optimization, etc.
Biography: Julita Corbalán is the team leader of the system software for energy management in HPC group at Barcelona Supercomputing Center (BSC). She received the engineering degree in computer science in 1996 and the PhD degree in computer science in 2002, both from the Technical University of Catalunya (UPC), Spain. Her research interests include processor management of parallel applications, parallel runtimes, scheduling policies and energy efficiency solutions for data centers. She has been the advisor of several PhD in computer science and she has participated in several long-term research projects with other universities and industries, mostly in the framework of the European Union ESPRIT, IST programs and HORIZON EUROPE 2020. She is currently a tenure lecturer in the Computer Science Department and Team leader in the "System software for energy management in HPC" group at BSC. The main project in this group is the EAR software, an energy management framework for data centers developed in a BSC-Lenovo collaboration project. Since 2020 she's the CTO of the Energy Aware Solutions company.
Biography: Julita Corbalán is the team leader of the system software for energy management in HPC group at Barcelona Supercomputing Center (BSC). She received the engineering degree in computer science in 1996 and the PhD degree in computer science in 2002, both from the Technical University of Catalunya (UPC), Spain. Her research interests include processor management of parallel applications, parallel runtimes, scheduling policies and energy efficiency solutions for data centers. She has been the advisor of several PhD in computer science and she has participated in several long-term research projects with other universities and industries, mostly in the framework of the European Union ESPRIT, IST programs and HORIZON EUROPE 2020. She is currently a tenure lecturer in the Computer Science Department and Team leader in the "System software for energy management in HPC" group at BSC. The main project in this group is the EAR software, an energy management framework for data centers developed in a BSC-Lenovo collaboration project. Since 2020 she's the CTO of the Energy Aware Solutions company.
3:00 - 4:00 PM - Session 6
- Deadline Miss Minimization Scheduling for License-Constrained CAE Jobs in Hybrid Cloud Infrastructure. [talk slides] Mohamed Noaman, Srishti Dasgupta and Michael Gerndt.
- Workshop Closing. Dalibor Klusáček
4:00 - 4:30 PM - Coffee Break
JSSPP 1995-2025 – the Workshop on Jobs Scheduling Strategies for Parallel Processing. Contact email: jssppw@gmail.com