Probabilistic Approximate Computing at Nanoscales: From data structures to memories

IF 2.3 Q3 NANOSCIENCE & NANOTECHNOLOGY

IEEE Nanotechnology Magazine Pub Date : 2022-02-01 DOI:10.1109/mnano.2021.3126092

Shanshan Liu, P. Reviriego, P. Junsangsri, Fabrizio Lombardi

{"title":"Probabilistic Approximate Computing at Nanoscales: From data structures to memories","authors":"Shanshan Liu, P. Reviriego, P. Junsangsri, Fabrizio Lombardi","doi":"10.1109/mnano.2021.3126092","DOIUrl":null,"url":null,"abstract":"The slowdown of CMOS technology scaling has placed architectures and algorithms on focus for future performance improvements in nanoscale computing systems. Two promising approaches at algorithmic level are approximate computing (AC) and probabilistic data structures (PDSs) that employ the tolerance of an application to small deviations in the results for reducing the complexity of the hardware implementation. AC focuses on applications that process numerical data and relies mostly on approximate (or inexact) low-level arithmetic operations. Instead, PDSs target categorical data and rely on shared data structures and other higher-level simplifications that introduce probabilistic deviations even when all operations are exact. Both AC and PDSs have been able to dramatically reduce the cost in some applications, but they are so far completely disconnected in the application domains, the abstraction levels, and the research communities. In this article, we introduce probabilistic approximate computing (PAC), a new paradigm to use application tolerance for small deviations to reduce the implementation complexity of data structures and hardware when implemented with nanoscale memory technologies. Its goal is to have data structures on which both AC and probabilistic techniques are used in a synergetic way to improve efficiency, while keeping deviations within acceptable margins.","PeriodicalId":44724,"journal":{"name":"IEEE Nanotechnology Magazine","volume":"16 1","pages":"16-24"},"PeriodicalIF":2.3000,"publicationDate":"2022-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"1","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"IEEE Nanotechnology Magazine","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/mnano.2021.3126092","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q3","JCRName":"NANOSCIENCE & NANOTECHNOLOGY","Score":null,"Total":0}

引用次数: 1

Abstract

The slowdown of CMOS technology scaling has placed architectures and algorithms on focus for future performance improvements in nanoscale computing systems. Two promising approaches at algorithmic level are approximate computing (AC) and probabilistic data structures (PDSs) that employ the tolerance of an application to small deviations in the results for reducing the complexity of the hardware implementation. AC focuses on applications that process numerical data and relies mostly on approximate (or inexact) low-level arithmetic operations. Instead, PDSs target categorical data and rely on shared data structures and other higher-level simplifications that introduce probabilistic deviations even when all operations are exact. Both AC and PDSs have been able to dramatically reduce the cost in some applications, but they are so far completely disconnected in the application domains, the abstraction levels, and the research communities. In this article, we introduce probabilistic approximate computing (PAC), a new paradigm to use application tolerance for small deviations to reduce the implementation complexity of data structures and hardware when implemented with nanoscale memory technologies. Its goal is to have data structures on which both AC and probabilistic techniques are used in a synergetic way to improve efficiency, while keeping deviations within acceptable margins.

查看原文本刊更多论文

纳米尺度的概率近似计算：从数据结构到存储器

CMOS技术扩展的放缓使架构和算法成为纳米级计算系统未来性能改进的重点。在算法层面上，两种有前途的方法是近似计算（AC）和概率数据结构（PDS），它们利用应用程序对结果中的小偏差的容忍度来降低硬件实现的复杂性。AC专注于处理数值数据的应用程序，主要依赖于近似（或不精确）的低级算术运算。相反，PDS以分类数据为目标，并依赖于共享数据结构和其他更高级别的简化，即使所有操作都是精确的，这些简化也会引入概率偏差。AC和PDS都能够显著降低某些应用程序的成本，但到目前为止，它们在应用程序领域、抽象级别和研究社区中完全脱节。在这篇文章中，我们介绍了概率近似计算（PAC），这是一种新的范式，当使用纳米级存储器技术实现时，它使用对小偏差的应用程序容差来降低数据结构和硬件的实现复杂性。其目标是使数据结构上以协同的方式使用AC和概率技术来提高效率，同时将偏差保持在可接受的范围内。

本文章由计算机程序翻译，如有差异，请以英文原文为准。

求助全文

约1分钟内获得全文求助全文

来源期刊

IEEE Nanotechnology Magazine NANOSCIENCE & NANOTECHNOLOGY-

CiteScore

2.90

自引率

6.20%

发文量

期刊介绍： IEEE Nanotechnology Magazine publishes peer-reviewed articles that present emerging trends and practices in industrial electronics product research and development, key insights, and tutorial surveys in the field of interest to the member societies of the IEEE Nanotechnology Council. IEEE Nanotechnology Magazine will be limited to the scope of the Nanotechnology Council, which supports the theory, design, and development of nanotechnology and its scientific, engineering, and industrial applications.