MultiSky: Dynamic Resource Allocation Framework for High-Throughput CGRA Multitask Execution

Coarse-grained reconfigurable arrays (CGRAs) offer a promising balance between high performance and flexibility, yet dynamic resource allocation in multitask scenarios remains challenging due to unpredictable task creation/destruction. Existing static approaches lack flexibility, while dynamic methods suffer from high latency or limited applicability. This article presents MultiSky, a framework for CGRA multitask dynamic resource allocation, combining a hardware controller and a software premapper. The hardware controller dynamically allocates resources within hundreds of cycles by calculating tile allocation for each task via weighted averaging, and generating tile shapes using a lightweight heuristic algorithm. The software premapper employs incremental compilation to pregenerate configurations, avoiding online transformation overhead. Evaluations on a real-world multitask scenario demonstrate that MultiSky achieves 1.72× higher throughput than baselines by maintaining 82.7% average resource utilization. The framework scales efficiently with larger CGRAs and task counts, with hardware overhead decreasing to 1% for 16\times 16$ CGRAs. These results highlight MultiSky's ability to balance flexibility, efficiency, and practicality in dynamic computing environments.