对PiM还是不对PiM

Q3 Computer Science

Queue Pub Date : 2022-12-31 DOI:10.1145/3580503

Gabriel Falcao, João Dinis Ferreira

{"title":"对PiM还是不对PiM","authors":"Gabriel Falcao, João Dinis Ferreira","doi":"10.1145/3580503","DOIUrl":null,"url":null,"abstract":"As artificial intelligence becomes a pervasive tool for the billions of IoT (Internet of things) devices at the edge, the data movement bottleneck imposes severe limitations on the performance and autonomy of these systems. PiM (processing-in-memory) is emerging as a way of mitigating the data movement bottleneck while satisfying the stringent performance, energy efficiency, and accuracy requirements of edge imaging applications that rely on CNNs (convolutional neural networks).","PeriodicalId":39042,"journal":{"name":"Queue","volume":"20 1","pages":"9 - 34"},"PeriodicalIF":0.0000,"publicationDate":"2022-12-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"To PiM or Not to PiM\",\"authors\":\"Gabriel Falcao, João Dinis Ferreira\",\"doi\":\"10.1145/3580503\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"As artificial intelligence becomes a pervasive tool for the billions of IoT (Internet of things) devices at the edge, the data movement bottleneck imposes severe limitations on the performance and autonomy of these systems. PiM (processing-in-memory) is emerging as a way of mitigating the data movement bottleneck while satisfying the stringent performance, energy efficiency, and accuracy requirements of edge imaging applications that rely on CNNs (convolutional neural networks).\",\"PeriodicalId\":39042,\"journal\":{\"name\":\"Queue\",\"volume\":\"20 1\",\"pages\":\"9 - 34\"},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2022-12-31\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Queue\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/10.1145/3580503\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q3\",\"JCRName\":\"Computer Science\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Queue","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1145/3580503","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q3","JCRName":"Computer Science","Score":null,"Total":0}

引用次数: 0

摘要

随着人工智能成为数十亿处于边缘的物联网设备的普及工具，数据移动瓶颈对这些系统的性能和自主性造成了严重限制。PiM（内存处理）正在成为一种缓解数据移动瓶颈的方法，同时满足依赖CNN（卷积神经网络）的边缘成像应用程序的严格性能、能效和准确性要求。

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

To PiM or Not to PiM

As artificial intelligence becomes a pervasive tool for the billions of IoT (Internet of things) devices at the edge, the data movement bottleneck imposes severe limitations on the performance and autonomy of these systems. PiM (processing-in-memory) is emerging as a way of mitigating the data movement bottleneck while satisfying the stringent performance, energy efficiency, and accuracy requirements of edge imaging applications that rely on CNNs (convolutional neural networks).

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

Queue Computer Science-Computer Science (all)

CiteScore

1.80

自引率

0.00%

发文量