The increasing adoption of heterogeneous computing architectures, particularly GPU-accelerated and domain-specific systems, has redefined the landscape of computing, from high-performance computing (HPC) to embedded systems at the edge. However, this architectural heterogeneity introduces significant challenges in achieving performance portability. Vendor-specific programming models (e.g., CUDA, HIP) often result in tightly coupled codebases, limiting cross-platform interoperability, impeding collaborative development, and undermining reproducibility.

As scientific applications scale toward exascale and beyond, the need for performance-portable solutions becomes critical. Achieving near-peak performance across diverse architectures without sacrificing maintainability or extensibility requires abstraction of layers, domain-specific languages, and compiler/runtime innovations. Simultaneously, the energy footprint of large-scale simulations is becoming a first-order concern, necessitating energy-aware optimizations and sustainable computing practices.

This convergence of performance, portability, and sustainability is not merely a technical challenge; it is a foundational requirement for the future of computational science and engineering.

Objectives

This workshop will focus on methodologies and frameworks that enable scalable and portable scientific computing across heterogeneous platforms. Topics will include:

• Compiler and runtime strategies for performance portability.
• Abstractions and Domain Specific Languages for architecture-agnostic optimization.
• Case studies demonstrating cross-platform scalability and performance tuning.
• Lessons learnt from code bases being ported to performance-portable programming models (e.g., Kokkos, SYCL, RAJA, OpenMP offloading).
• Portability approaches for legacy applications.
• Metrics and methodologies for evaluating portability in HPC.
• Co-design approaches for balancing performance, portability, and power efficiency.
• Portability of performance as well as other metrics, including (but not limited to) energy, power, carbon efficiency, and application accuracy.
• Portable applications for future computing systems and architectures.

The session aims to foster discussion on the trade-offs and synergies between these dimensions, highlighting emerging research and practical solutions that address the evolving demands of extreme-scale scientific computing.

Submission Guidelines

All papers must be original and not simultaneously submitted to another journal or conference. At least one full registration and physical attendance at the Computing Frontiers 2026 Conference is required for an accepted paper to be published in the proceedings.

All submissions will undergo single-blind review. Author names and affiliations should be included in the manuscript.Accepted papers will appear in the ACM Digital Library under Gold Open Access.

Submission Procedure

1. Prepare your manuscript
   • Format your paper according to the Author’s GuideCreate or log in to your EasyChair account
2. Visit the EasyChair website
   • If you do not already have an account, you will need to create one
3.  Submit your paper
   • Fill in all required metadata, including title, authors, affiliations, and abstract
   • Upload your paper in PDF format
4. Confirm your submission
   • Review your information carefully before final submission
   • You may update your submission until the paper submission deadline

Committees

Technical Program Committee

• Didem Unat - Koç University
• Flavio Vella - University of Trento
• Ivy Peng - KTH Royal Institute of Technology
• Jussi Enkovaara - CSC
• Marion Weinzierl - University of Cambridge
• Matt Bettencourt - NVIDIA
• Miquel Pericàs - Chalmers University of Technology
• Tobias Weinzierl - Durham University
• Vito Castellana - Pacific Northwest National Laboratory
• Michael Bussmann - Helmholtz-Zentrum Dresden-Rossendorf

Organizing committee

• Nitin Shukla — CINECA — n.shukla@cineca.it
• Biagio Cosenza — University of Salerno — bcosenza@unisa.it
• Elisabetta Boella — E4 Computer Engineering — elisabetta.boella@e4company.com
• Davide Gadioli — Politecnico di Milano — davide.gadioli@polimi.it
• Daniele Gregori — E4 Computer Engineering — daniele.gregori@e4company.com

Publication

Accepted papers will be published in the ACM Digital Library as companion of the Computing Frontiers ‘26 proceedings.

Venue

The workshop will be co-located with the Computer Frontiers 2026 conference, bringing together researchers and practitioners in an eclectic, interdisciplinary, collaborative community. The conference will take place in Catania, Sicily, Italy. Workshop attendance requires full registration with the main conference. Please, refer to the Computer Frontiers 2026 website for more information.

Contact

All questions about submissions should be emailed to n.shukla@cineca.it.