arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2401.16792cs.AR
收藏

WideSA:一种面向Versal ACAP架构上均匀递推的高阵列利用率映射方案

WideSA: A High Array Utilization Mapping Scheme for Uniform Recurrences on the Versal ACAP Architecture

Tuo Dai, Bizhao Shi, Guojie Luo

AI总结:

针对Versal ACAP架构上均匀递推映射的AIE利用率难题,提出WideSA映射方案,结合多面体时空变换、路由感知PLIO分配与自动映射框架,在VCK5000矩阵乘法任务上实现了优于现有最优加速器1.11倍的浮点吞吐量。

摘要:

Versal自适应计算加速平台(ACAP)是一种将AI引擎(AIE)与可重构结构相结合的新型架构。该架构为深度学习、高性能计算、信号处理等多个领域中的均匀递推运算提供了显著的加速潜力。然而,如何将这些计算高效映射到Versal ACAP架构上,同时实现AIE的高利用率,是一项挑战。\n 为解决这一问题,我们提出了名为WideSA的映射方案,旨在通过结合硬件与计算本身的特性,在Versal ACAP架构上加速均匀递推运算。考虑到AIE的阵列架构,我们的方法采用基于多面体模型的时空变换,生成合法且优化的脉动阵列映射。同时,我们开发了一种适配AIE阵列通信的、感知路由的PLIO分配算法,该算法在确保编译成功的同时最大化阵列利用率。此外,我们还提出了一个自动映射框架,该框架可生成均匀递推对应的可执行代码,涵盖AIE内核程序、可编程逻辑比特流以及主机程序。实验结果验证了我们映射方案的有效性。具体而言,将该方案应用于VCK5000板卡上的矩阵乘法运算时,我们在浮点数据类型上实现了4.15TOPS的吞吐量,比Versal ACAP架构上的现有最优加速器高出1.11倍。

英文摘要:

The Versal Adaptive Compute Acceleration Platform (ACAP) is a new architecture that combines AI Engines (AIEs) with reconfigurable fabric. This architecture offers significant acceleration potential for uniform recurrences in various domains, such as deep learning, high-performance computation, and signal processing. However, efficiently mapping these computations onto the Versal ACAP architecture while achieving high utilization of AIEs poses a challenge. To address this issue, we propose a mapping scheme called \fname, which aims to accelerate uniform recurrences on the Versal ACAP architecture by leveraging the features of both the hardware and the computations. Considering the array architecture of AIEs, our approach utilizes space-time transformations based on the polyhedral model to generate legally optimized systolic array mappings. Concurrently, we have developed a routing-aware PLIO assignment algorithm tailored for communication on the AIE array, and the algorithm aims at successful compilation while maximizing array utilization. Furthermore, we introduce an automatic mapping framework. This framework is designed to generate the corresponding executable code for uniform recurrences, which encompasses the AIE kernel program, programmable logic bitstreams, and the host program. The experimental results validate the effectiveness of our mapping scheme. Specifically, when applying our scheme to matrix multiplication computations on the VCK5000 board, we achieve a throughput of 4.15TOPS on float data type, which is 1.11$\times$ higher compared to the state-of-the-art accelerator on the Versal ACAP architecture.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑