Workshop Program

ExHetAI 2025 will be held on Sunday, November 16. The schedule is provided below. All times are in CST.

9am - 9:10am CST	Opening Remarks
9:10am - 10am CST	Invited Talk 1: Memorization vs Reasoning in MoEs and Estimating Memory Consumption in Distributed Training Rio Yokota Institute of Science Tokyo
10am - 10:30am CST	Morning Break
10:30am - 11:10am CST	Invited Talk 2: TBD Matthew Dearing Argonne National Laboratory
11:10am - 11:30am CST	Enabling Unstructured Sparse Fine-Tuning and Inference for Foundation Models on Wafer-Scale Engine Haoyu Zheng, Yifan Zeng, Linghao Song, Murali Emani, and Wenqian Dong Session chair: Seyong Lee
11:30am - 11:50am CST	WAGES: Workload-Aware GPU Sharing System for Energy-Efficient Serverless LLM Serving Tianyu Wang, Gourav Rattihalli, Aditya Dhakal, Xulong Tang, and Dejan Milojicic Session chair: Seyong Lee
11:50am - 12:10pm CST	OmniFed: A Modular Framework for Configurable Federated Learning from Edge to HPC Sahil Tyagi, Andrei Cozma, Olivera Kotevska, and Feiyi Wang Session chair: Pedro Valero-Lara
12:10pm - 12:30pm CST	Enhancing ChatPORT with CUDA-to-SYCL Kernel Translation Capability Zheming Jin, Swaroop Pophale, and Keita Teranishi Session chair: Pedro Valero-Lara
12:30pm CST	Closing Remarks

Call for Papers

Topics will include but will not be limited to:

AI and HPC Convergence: Application of AI techniques to accelerate scientific computing and enhance HPC workflows; integration of machine learning and data-driven methods into traditional simulation pipelines.;

Heterogeneous Architectures for HPC and AI: Design, deployment, and utilization of diverse compute platforms including GPUs, TPUs, FPGAs, neuromorphic processors, quantum devices, and domain-specific accelerators in high-performance environments.;

Programming Models and Portability: Frameworks, languages, and tools that enable performance portability and efficient code execution across heterogeneous and evolving architectures.;

AI-Assisted System Optimization: Use of AI and machine learning for compiler optimizations, workload scheduling, autotuning, performance modeling, and intelligent resource management in complex systems;

Memory and Data Management: Innovations in hierarchical memory systems, data movement optimization, locality-aware computing, and high-performance I/O strategies tailored for heterogeneous platforms;

Energy Efficiency and Sustainable HPC: AI-enhanced techniques for reducing energy consumption, managing power-performance trade-offs, and enabling environmentally sustainable supercomputing;

Submission

Authors are invited to submit manuscripts in English structured as technical papers up to 6 pages 2-column pages, min 5 pages (U.S. letter – 8.5″x11″), excluding the bibliography, using the ACM proceedings template that is available at here. The manuscripts are single-blind. Word authors can use the “Interim Layout”. Submissions not conforming to these guidelines may be returned without review.

All manuscripts will be peer-reviewed and judged on correctness, originality, technical strength, and significance, quality of presentation, and interest and relevance to the workshop attendees. Submitted papers must represent original unpublished research that is not currently under review for any other conference or journal. Papers not following these guidelines will be rejected without review and further action may be taken, including (but not limited to) notifications sent to the heads of the institutions of the authors and sponsors of the conference. Submissions received after the due date, exceeding length limit, or not appropriately structured may also not be considered. At least one author of an accepted paper must register for and attend the workshop. Authors may contact the workshop organizers for more information.

Papers should be submitted electronically at: https://submissions.supercomputing.org, SC25 Workshop: ExHetAI'25: Workshop on Extreme Heterogeneity and AI Convergence in HPC".

The final papers will be published in the SC Workshops Proceedings.

Important Dates