通过多位 QD-NVRAM 使用点积进行内存计算

Q4 Engineering

International Journal of High Speed Electronics and Systems Pub Date : 2024-07-25 DOI:10.1142/s012915642440055x

R. Gudlavalleti, J. Chandy, E. Heller, F. Jain

{"title":"通过多位 QD-NVRAM 使用点积进行内存计算","authors":"R. Gudlavalleti, J. Chandy, E. Heller, F. Jain","doi":"10.1142/s012915642440055x","DOIUrl":null,"url":null,"abstract":"This paper presents in-memory computing using fast write/erase quantum dot (QD) nonvolatile random access memory (NVRAM). In comparison to NVMs, multi-state NVRAMs offer enhanced Compute-In-Memory capability for applications in deep neural network architecture. Dot product is the methodology that enables an array structure for multiply and accumulate (MAC) operation. We show an approach to dot product computation using multi-state quantum dot channel (QDC) FETs and QD-NVRAM.","PeriodicalId":35778,"journal":{"name":"International Journal of High Speed Electronics and Systems","volume":null,"pages":null},"PeriodicalIF":0.0000,"publicationDate":"2024-07-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"In-Memory Computing Using Dot-Product via Multi-Bit QD-NVRAMs\",\"authors\":\"R. Gudlavalleti, J. Chandy, E. Heller, F. Jain\",\"doi\":\"10.1142/s012915642440055x\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"This paper presents in-memory computing using fast write/erase quantum dot (QD) nonvolatile random access memory (NVRAM). In comparison to NVMs, multi-state NVRAMs offer enhanced Compute-In-Memory capability for applications in deep neural network architecture. Dot product is the methodology that enables an array structure for multiply and accumulate (MAC) operation. We show an approach to dot product computation using multi-state quantum dot channel (QDC) FETs and QD-NVRAM.\",\"PeriodicalId\":35778,\"journal\":{\"name\":\"International Journal of High Speed Electronics and Systems\",\"volume\":null,\"pages\":null},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2024-07-25\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"International Journal of High Speed Electronics and Systems\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/10.1142/s012915642440055x\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q4\",\"JCRName\":\"Engineering\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"International Journal of High Speed Electronics and Systems","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1142/s012915642440055x","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q4","JCRName":"Engineering","Score":null,"Total":0}

引用次数: 0

摘要

本文介绍了使用快速写入/擦除量子点（QD）非易失性随机存取存储器（NVRAM）的内存计算。与 NVM 相比，多态 NVRAM 为深度神经网络架构中的应用提供了更强的内存计算能力。点积是实现乘法和累加（MAC）操作的阵列结构的方法。我们展示了一种利用多态量子点沟道（QDC）场效应晶体管和 QD-NVRAM 进行点积计算的方法。

本文章由计算机程序翻译，如有差异，请以英文原文为准。

查看原文本刊更多论文

In-Memory Computing Using Dot-Product via Multi-Bit QD-NVRAMs

This paper presents in-memory computing using fast write/erase quantum dot (QD) nonvolatile random access memory (NVRAM). In comparison to NVMs, multi-state NVRAMs offer enhanced Compute-In-Memory capability for applications in deep neural network architecture. Dot product is the methodology that enables an array structure for multiply and accumulate (MAC) operation. We show an approach to dot product computation using multi-state quantum dot channel (QDC) FETs and QD-NVRAM.

求助全文

通过发布文献求助，成功后即可免费获取论文全文。去求助

来源期刊

International Journal of High Speed Electronics and Systems Engineering-Electrical and Electronic Engineering

CiteScore

0.60

自引率

0.00%

发文量

期刊介绍： Launched in 1990, the International Journal of High Speed Electronics and Systems (IJHSES) has served graduate students and those in R&D, managerial and marketing positions by giving state-of-the-art data, and the latest research trends. Its main charter is to promote engineering education by advancing interdisciplinary science between electronics and systems and to explore high speed technology in photonics and electronics. IJHSES, a quarterly journal, continues to feature a broad coverage of topics relating to high speed or high performance devices, circuits and systems.