2026 International Workshop on Performance, Portability & Productivity in HPC

Program

The P3HPC workshop will be held on Sunday, 15 November 2026 from 9:00AM - 5:30PM Central.

Session 1

09:05 - 09:40 CST
Featured Speaker
A Study in High Performance Portability Through Layered Abstractions
Stephen Jones

09:40 - 10:00 CST
14 GPU Programming Paradigms Put to the Test: Performance Portability and Halstead Difficulty in Scientific Computing
Jonas Schuhmacher, Hans-Joachim Bungartz

Paper

Morning Break

10:00 - 10:30 CST

Session 2

10:30 - 10:54 CST
Achieving Performance Portability for Semi-Lagrangian Simulations via Automated Layout and Loop Tuning
Etienne Malaboeuf, Julien Bigot, Kévin Obrejan, Virginie Grandgirard, Emily Bourne, Thomas Padioleau

Paper

10:54 - 11:18 CST
Performance Portability of Multi-GPU Hydrodynamics: SCALSALE with MPI+OpenMP Target Offload
Avitay Geltman, Shahar Maimon, Yehonatan Fridman, Ron Wagner, Shay Shani, Re'em Harel

Paper

11:18 - 11:42 CST
Meganeura: Vulkan and Metal as a Performance-Portability Layer for GPU Training and Inference
Dzmitry Malyshau

Paper

11:42 - 12:06 CST
A Performance-Portable GPU-Accelerated Solution for Particle-Particle Particle-Mesh Method using Cabana and FFTX
Zoe Barbeau, Het Menkad, Phillip Colella, Sanil Rao, Franz Franchetti, Jeffrey Vetter, Keita Teranishi

Paper

12:06 - 12:30 CST
A Multi-Layer Workflow for Diagnosing Performance Portability in RAJA Applications
Befikir Bogale, Ian Lumsden, Stephanie Brink, Olga Pearce, Michela Taufer

Paper

Lunch Break

12:30 - 14:00 CST

Panel

14:00 - 15:00 CST
The Future of P3HPC in the Age of Agentic Development
Moderator:Balint Joo

Afternoon Break

15:00 - 15:30 CST

Session 3

15:30 - 15:54 CST
Extending the Performance Portability of Julia to Long-Vector Architectures
Rintaro Togo, Keichi Takahashi, Tomonori Hayami, Kota Sakakura, Takashi Soga, Arata Endo, Susumu Date

Paper

15:54 - 16:18 CST
The Price of a Single Source: Performance Portability of Chapel and Julia across CPUs and GPUs
Andrew Morgan

Paper

16:18 - 16:42 CST
Performance Improvement of Multi-Dimensional Parallel Loops in Kokkos
Adrien Taberner, Hariprasad Kannan, Yuuichi Asahi, Dong Hun Lee, Trévis Morvany, Jean-François David, Nevin Liber, Nicolas Morales, Christian R. Trott, Julien Bigot, Damien Lebrun-Grandie

Paper

16:42 - 17:06 CST
Mojo-Driven Performance Portability for LBM
Giulio Malenza, Matthew A. Smith, Zadok Storkey, Massimiliano Meneghin, Marco Aldinucci, Robert Birke, Steven A. Wright

Paper

17:06 - 17:30 CST
Triton Attention Backend: Portable High-Performance Paged Attention for Production-Grade LLM Inference
Burkhard Ringlein, Jan van Lunteren, Radu Stoica, Thomas Parnell

Paper