An MLIR-Based Compilation Framework for Control Flow Management on Coarse Grained Reconfigurable Arrays

Coarse-Grained Reconfigurable Arrays (CGRAs) present both high flexibility and efficiency, making them well-suited for the acceleration of intensive workloads. Nevertheless, a key barrier towards their widespread adoption is posed by CGRA compilation, which must cope with a multi-dimensional space spanning both the spatial and the temporal domains. Indeed, state-of-the-art compilers are limited in scope, as they mostly handle data flow while providing little or no support for control flow. Hence, they mostly focus on mapping single loops and/or delegate the management of control-flow divergences to ad-hoc hardware units. Conversely, in this paper, we show that control flow can be effectively managed and optimized at the compilation level, enabling a broad set of applications to be targeted while remaining hardware-agnostic and achieving high performance. We embody our methodology in a modular compilation framework consisting of transformation and optimization passes, enabling support for applications with arbitrary control flows running on homogeneous CGRA meshes. We also introduce a novel mapping methodology that serves as a compilation back-end, addressing limitations in available CGRA hardware resources and ensuring a feasible solution during compilation. Our framework achieves up to 2.1 $\times$ runtime speed-up over state-of-the-art approaches, purely through compilation optimizations.