一句话：提出 Neura 框架，以分层时空 CGRA 架构配合迁移感知映射算法，实现动态资源分配与内核迁移，显著提升多内核场景的吞吐与利用率。
粗粒度可重构阵列（CGRA）在能效加速方面前景广阔，但面临可扩展性瓶颈，论文将其归纳为三方面：映射可扩展性差（设计复杂度上升时难以找到可行或最优映射）、体系结构限制（映射粒度与访存方式僵化，制约灵活性与性能）、以及动态多内核支持不足（多内核的动态并发执行缺乏探索）。为此作者提出开源统一框架 Neura，引入新颖的分层时空 CGRA 架构，并配合迁移感知的映射算法，二者结合使动态资源分配与内核迁移成为可能，从而缓解多内核场景下的可扩展性与利用率问题。评估在多种场景下与同规模传统 CGRA 对比，Neura 取得 1.64× 至 3.85× 的吞吐提升，并显著提高资源利用率。
Coarse-Grained Reconfigurable Arrays (CGRAs) are a promising solution for energy-efficient acceleration across multiple application domains. Yet, CGRAs face significant scalability challenges that hinder their widespread adoption, stemming from three main concerns: (1) Mapping Scalability — existing mapping algorithms struggle to find feasible and optimal solutions as the design complexity grows; (2) Architectural Limitations — rigid mapping granularity and memory access restrict flexibility and performance; and (3) Dynamic Multi-Kernel Support — dynamic and simultaneous execution of multiple kernels are not thoroughly explored, limiting the applicability of CGRAs in complex multi-kernel scenarios. In this paper, we propose Neura, an open-source unified framework towards scalable CGRAs to address these challenges. Neura introduces a novel hierarchical spatial-temporal CGRA architecture combined with a migration-aware mapping algorithm. This combination uniquely enables dynamic resource allocation and kernel migration, addressing key scalability and utilization challenges in multi-kernel scenarios. Our experimental results demonstrate that Neura achieves a throughput improvement of 1.64× to 3.85× and significant higher utilization, across different scenarios over a conventional CGRA at the same scale.
@inproceedings{0002JSYOZ0026,
title = {Neura: A Unified Framework for Hierarchical and Adaptive CGRAs},
author = {Cheng Tan and Miaomiao Jiang and Yuqi Sun and Ruihong Yin and Yanghui Ou and Qing Zhong and Lei Ju and Jeff Zhang},
booktitle = {ASPLOS},
year = {2026},
doi = {10.1145/3779212.3790193}
}