The @mvapich team takes on the @PyTorch 2026 conference! Lab member Quentin Anthony showcases the group's growing #HPC-#AI project and its current directions to use 25+ years of #MPI research for AI #training and #inference on modern #CPU, #GPU, and #interconnects!

Members of the @mvapich team will be attending CARLA 2026 in Cordoba, Argentina next week! We'll be presenting our tutorials on #HPC networking tech and #Distributed #AI #training and #Inference alongside a workshop keynote talk! We hope to see you there!!

Live demo time at #MUG2026! @OhioState’s Assistant Professor Dr. @harisubramoni & Lead AI Engineer Brijesh Nanda showcase @ICICLE_AI's end-to-end pipeline: Intelligent Edge Management Service, No Code Image Lab, and Smart Labeler! #EdgeAI #ComputerVision #DataScience #HPC

Next at #MUG2026: @cwru Department Chair & Professor Dr. Vipin Chaudhary gives an overview of ICICLE AI and CI services, exploring 100+ standardized, reusable software components and portals bridging AI to domain applications! #Cyberinfrastructure #AI #HPC

Happening now: XScaleSolutions' Principal HPC Engineer Soham Ghosh & Senior Software Engineer Kyle Schaefer discuss accelerating #HPC and #AI workloads with MVAPICH2-DPU, X-ScaleSecureMPI, X-ScalePETSc, and X-ScaleAI packages! #MPI #DPU #MachineLearning #MUG2026

Welcoming back @uflorida Associate Professor Dr. Xiaoyi Lu at #MUG2026! He presents on scalable and secure AllReduce for modern #HPC and #AI, highlighting DPAR for differential privacy and MPCAR for secure multi-party computation in collective communication! #Supercomputing

Next up at #MUG2026: Mahidhar Tatineni from @SDSC_UCSD presenting on application performance with MVAPICH-Plus across different partitioning modes on the Cosmos system with AMD MI300A #HPC #MPI #SDSC #AMD #MI300A

Now at #MUG2026: HooYoung Ahn from ETRI presenting on KCCL, the K-Collective Communication Library — a new collective communication library built for NPUs and DPUs developed in South Korea. #HPC #MPI #NPU #DPU

Next up at #MUG2026: Natalie Beams and Ahmad Abdelfattah from @UTKnoxville presenting on performance and scalability of 3D FFT using heFFTe and @mvapich

Next up at #MUG2026. Sameh Abdulah from @KAUST presenting high-performance federated and distributed modeling of spatial data at scale

Coming up at #MUG2026 — @NVIDIA's Jeff Hammond presents VibeMPI: what happens when you let agentic AI (Claude, Codex) develop and test MPI. #HPC #MPI

Kicking off the final day of #MUG2026 is a keynote talk from Prof. Yutong Lu of Sun Yat-Sen University, where she talks about experiences from building the #LineShine #Exascale supercomputer and what to expect in the post-exascale #HPC and #AI worlds!

Spotlighting NOWLAB research at #MUG2026! Graduate Research Associates Lang Xu (characterizing communication patterns in distributed #LLM inference) and Sungjae Lee (HAT-MPI hierarchical auto-tuning on InfiniBand clusters) present their short talks! #HPC #MPI #AI #OhioState

Focusing on the Trinetra communication stack at #MUG2026: @cdacindia 's Scientist E Rakesh Kumar Yadav presents on accelerating HPC & AI applications using MVAPICH 4, detailing topology- and GPU-aware optimizations on heterogeneous systems! #HPC #AI #MPI

Next on stage at #MUG2026: Brian Hwang, Principal Software Engineer at @CornelisHPC , breaks down how the BTS/HFISVC architecture provides a common services layer across Cornelis 5K/6K fabrics to accelerate both #MPI and IP communication paths! #HPC #AI

Happening now: Longtime #MUG collaborator Dr. Sameer Shende, Director of PRL at @UOregon & President of @ParaToolsInc, breaks down advanced performance introspection with the TAU Performance System®, highlighting GPU sampling, MPI_T tools, & @mvapich support! #HPC #Profiling

Continuing #MUG2026: Sangamesh Kodge, Senior AI Systems Architect at @d_matrix_ai, explores disaggregating compute across GPUs & custom accelerators to redefine the interactivity vs throughput frontier for scalable #LLM inference! #AI #HPC #Accelerators

In addition to our amazing talks at #MUG2026, fresh off the presses is @osc 's press release about this year's edition of MUG: lnkd.in/gsUX-B2s We're extremely grateful to them for allowing us to host MUG year after year!

Up now: Jithin Jose, Principal Software Engineer at @Azure, explores transport-aware collective communication with MRC and MSCCL++ for massive-scale AI training and inference! #AI #HPC #Supercomputing #CCL #MUG2026