DP2E-AI 2026

2nd International Workshop on Distributed and Parallel Programming for Extreme-scale AI

Exascale machines and clouds are now available across several different architectures and arithmetic. Supercomputing and its programming paradigms are no longer only led by historical computational science applications. Brain-scale applications, such as those arising from machine learning and AI, require the manipulation of very large datasets that often lead to distributed and parallel linear algebra problems (e.g., very sparse non-symmetric matrix computations with dynamic patterns), where performance can be substantially improved. The concepts of benchmarking and performance evaluation, along with their evolution, must adapt in response to these changes and to the emergence of new applications in domains that are expected to dominate a large share of the future hardware and software industry worldwide.

Therefore, artificial intelligence (AI), encompassing both popular large language models (LLMs) and machine learning applications more broadly, is currently reshaping architectures as well as distributed and parallel programming approaches. AI is catalyzing developments ranging from chip design, arithmetic, and accelerators, to new data structures and programming paradigms. Since the first edition of this workshop, agentic AI has seen widespread deployment, new brain-inspired AI models have been proposed, and new projects exploring alternative paradigms for AI architectures have emerged.

Nevertheless, linear algebra and distributed-parallel programming expertise remains and will continue to be required. Convergence between these domains is not guaranteed, and we therefore need to build bridges to enable interoperability, thereby preventing separate roadmaps and ecosystems. Decades-long research in distributed and parallel computational science, linear algebra, and middleware may be adapted to optimize these new application domains.

In order to discuss these challenges, we are organizing a two-day workshop at École des Mines de Paris to bring together scientists working in the fields of HPC, AI, linear algebra, and related application domains. The program is composed of keynote talks, dedicated sessions on distributed and parallel programming for AI methods, along with panels and general discussion sessions.

Sponsors

Contact Us

Please send any questions related to the DP2E-AI 2026 workshop to dp2eai2026@gmail.com