PIMSYN: Synthesizing Processing-in-memory CNN Accelerators

• DOI: 10.48550/arxiv.2402.18114
• Publication Date: 2024-02-28
• Authors: Wanqian Li; Xiaotian Sun; Xinyu Wang; Lei Wang; Yinhe Han; Xiaohong Chen

Processing-in-memory architectures have been regarded as a promising solution for CNN acceleration. Existing PIM accelerator designs rely heavily on the experience of experts and require significant manual design overhead. Manual cannot effectively optimize explore architecture implementations. In this work, we develop an automatic framework PIMSYN synthesizing PIM-based accelerators, which greatly facilitates helps generate energyefficient accelerators. can automatically transform applications into execution workflows hardware construction To systematically architecture, embed architectural exploration flow synthesis framework, providing more comprehensive space. Experiments demonstrate that improves power efficiency by several times compared with existing works. be obtained from https://github.com/lixixi-jook/PIMSYN-NN.