PolyCIM：基于多面体编译提升数字CIM加速器中的数据复用
PolyCIM: Improving Data Reuse in Digital CIM Accelerators with Polyhedral-Based Compilation
浏览论文内容
中文总结 AI 辅助
针对数字CIM加速器中DNN映射导致的阵列利用不足问题，提出基于多面体的编译框架PolyCIM，通过仿射变换暴露非轴向数据复用超平面，实现最高4倍宏利用率和3.2倍加速。
中文摘要 AI 辅助
数字计算存储（CIM）通过将计算逻辑直接集成到存储阵列中，为加速深度神经网络（DNN）提供了一种有前景的解决方案。然而，由于刚性CIM阵列结构所施加的严格数据复用约束，将现代DNN算子映射到CIM加速器往往导致阵列严重利用不足。我们观察到，现代DNN中的数据复用形成了通常沿非轴向方向定向的超平面结构，这使得它们对仅利用轴向对齐复用的传统映射方法不可见。在这项工作中，我们提出了PolyCIM，一个基于多面体的CIM架构编译框架，通过仿射变换系统地暴露并重新对齐这些超平面。PolyCIM提供了一种统一的抽象，能够高效地表示多样化的DNN工作负载和数字CIM架构。通过数据复用暴露、计算映射和数据移动优化，PolyCIM为CIM架构生成映射，实现了优越的阵列利用率和性能。实验结果表明，PolyCIM在宏单元利用率上实现了高达4倍的提升，并获得了3.2倍的加速，有效弥合了现代DNN算子与CIM架构之间的差距。
英文摘要
Digital Compute-in-Memory (CIM) presents a promising solution for accelerating deep neural networks (DNNs) through the integration of computational logic directly within memory arrays. However, mapping modern DNN operators to CIM accelerators often results in severe array underutilization, due to the strict data reuse constraints imposed by the rigid CIM array structure. We observe that data reuse in modern DNNs forms hyperplane structures often oriented along non-axial directions, rendering them invisible to conventional mapping methods that only exploit axis-aligned reuse. In this work, we propose PolyCIM, a polyhedral-based compilation framework for CIM architectures that systematically exposes and realigns these hyperplanes through affine transformations. PolyCIM provides a unified abstraction capable of efficiently representing both diverse DNN workloads and digital CIM architectures. Through data reuse exposure, computation mapping, and data movement optimization, PolyCIM generates mappings for CIM architectures that achieve superior array utilization and performance. Experimental results show that PolyCIM delivers up to $4\times$ improvement in macro utilization and $3.2\times$ speedup, effectively bridging the gap between modern DNN operators and CIM architectures.
发表机构
- Beihang University（北京航空航天大学）
机构由 AI 辅助整理，请以论文原文为准。
