Efficient task scheduling in heterogeneous computing systems remains a fundamental challenge in embedded and high-level synthesis flows, according to a new paper on arXiv. The paper introduces MoSAIC (Motif-aware Scheduling with AI-driven Cognition) as a response to this problem.
The full title of the work indicates that MoSAIC learns DAG scheduling heuristics from optimal schedules, using two components: motif-aware recovery and LLM-guided reasoning. The abstract, however, is truncated and does not provide further details on how these components work.
The paper is available as arXiv:2610.15205. No experimental results or comparisons are included in the available abstract.