2020/09/21 by Kamil A. Khan, Khan, Kamil, Sudeep Pasricha +3 · 2 citations
Computer Science · Engineering · #Advanced Memory and Neural Computing #Cloud Computing and Resource Management #FOS: Computer and information sciences #Ferroelectric and Negative Capacitance Devices #Hardware Architecture (cs.AR)
paper · pdf · doi:10.48550/arxiv.2009.09603
openalex publication_date 2020/09/21 · openalex created_date 2022/07/24 · openalex updated_date 2026/07/28
Due to amount of data involved in emerging deep learning and big data\napplications, operations related to data movement have quickly become the\nbottleneck. Data-centric computing (DCC), as enabled by processing-in-memory\n(PIM) and near-memory processing (NMP) paradigms, aims to accelerate these\ntypes of applications by moving the computation closer to the data. Over the\npast few years, researchers have proposed various memory architectures that\nenable DCC systems, such as logic layers in 3D stacked memories or charge\nsharing based bitwise operations in DRAM. However, application-specific memory\naccess patterns, power and thermal concerns, memory technology limitations, and\ninconsistent performance gains complicate the offloading of computation in DCC\nsystems. Therefore, designing intelligent resource management techniques for\ncomputation offloading is vital for leveraging the potential offered by this\nnew paradigm. In this article, we survey the major trends in managing PIM and\nNMP-based DCC systems and provide a review of the landscape of resource\nmanagement techniques employed by system designers for such systems.\nAdditionally, we discuss the future challenges and opportunities in DCC\nmanagement.\n