EuroMPI 2026 will continue to focus on advancing the ubiquitous Message Passing Interface (MPI) model and specification for parallel programming. It will also encompass extensions and alternative interfaces for high-performance heterogeneous and hybrid systems, benchmarks, tools, parallel I/O, fault tolerance, and parallel applications using MPI and other interfaces.

EuroMPI 2026 will be hosted by TU Wien, in the center of Vienna, Austria and will be co-located with IWOMP 2026 (October 7 - October 9, 2026), with joint tutorials, and opportunities for interaction between the two communities.

Through the presentation of full papers, position papers, posters, and invited talks, the meeting will provide ample opportunities for attendees to interact and share ideas and experiences to contribute to the improvement and furthering of message-passing and related parallel programming paradigms.

Having EuroMPI, IWOMP, and the MPI Forum grouped together enables attendees to engage across communities and contribute to the advancement of message passing, shared-memory parallelism, and related parallel paradigms -- including their combined use.

We invite high-quality, full and short paper submissions on all topics related to message-passing parallel programming with MPI and related or competing models. Topics of interest include, but are not limited to:

    Implementation Issues and Algorithms: Efficient and scalable implementations of message-passing constructs.
    Architectures and systems: Towards exascale computing, efficient use of accelerators and other features of modern, large-scale systems, hardware-software interaction.
    Message Passing for accelerator-based compute architectures, including multi-GPU, TPU, and AI-accelerator systems.
    Programming models and paradigms related to MPI targeting large-scale, parallel, distributed memory systems, hierarchical and hybrid models, partitioned global address space (PGAS) models.
    Extensions to and shortcomings of MPI: alternative interfaces and solutions.
    Parallel I/O: New mechanisms and optimizations for (MPI-)I/O.
    Hybrid and heterogeneous programming with MPI and other interfaces.
    Message passing interface support for data-intensive parallel applications.
    Fault tolerance in message-passing implementations and systems.
    MPI parallel programming in clouds and non-dedicated systems.
    Performance evaluation for MPI and MPI-based applications.
    Automatic performance tuning of applications and implementations.
    Verification of message-passing applications and protocols.
    Applications using advanced message passing, in particular in computational science and scientific computing.
    Parallel algorithms in the message-passing paradigm.
    Interactions between the message-passing library and runtime systems.
    Use of MPI in Artificial Intelligence (AI) and Large Language Models (LLM) and its challenges
    Competing standards: Integration with and comparison to competing industry standards (e.g., NCCL, RCCL)

Papers will be reviewed based on quality, originality, clarity, correctness, technical strength, and relevance. Each paper submission will receive at least three reviews from members of the program committee. Reviews are single-anonymous (but not double-anonymous); the identities of authors will be visible to the program committee.

Papers not accepted for presentation may be recommended for poster presentations. Posters will be published informally on the website, but are not intended to appear in the regular conference proceedings. Posters provide an excellent opportunity to present early-stage work to the community, which can lead to discussions, constructive feedback, and collaboration opportunities.