In and near-memory computing using DRAM

Enas Emad Abulibdeh, Baker Mohammad, Hani Saleh

    Research output: Chapter in Book/Report/Conference proceedingChapterpeer-review

    Abstract

    Dynamic random access memory (DRAM) is one type of semiconductor memory technology that has been used in electronic systems for decades and will survive into the future for its outstanding specifications, including simplicity, low cost, and high-density architecture. However, the access time of DRAM degrades the performance of the utilized system. In- and near-memory computing resolves the memory wall problem, where off-the-shelf, unmodified, and commercial DRAM with minimal modifications are used. Bitwise operations like logic-AND, bulk initialization, computation acceleration, true random number extraction, and secure key generation are different approaches for DRAM computing. The implementation of DRAM computing paradigms involves programming the memory controller, updating the peripheral circuits, integrating low-overhead blocks near the memory, or a combination of these.

    Original languageBritish English
    Title of host publicationIn-Memory Computing Hardware Accelerators for Data-Intensive Applications
    PublisherSpringer Nature
    Pages39-55
    Number of pages17
    ISBN (Electronic)9783031342332
    ISBN (Print)9783031342325
    DOIs
    StatePublished - 25 Sep 2023

    Fingerprint

    Dive into the research topics of 'In and near-memory computing using DRAM'. Together they form a unique fingerprint.

    Cite this