RAISE: The Reconfigurable Optical Computer (ROC)
RAISE: The Reconfigurable Optical Computer (ROC)
批准号:
1748294
负责人:
Tarek El-Ghazawi
金额:
$90.0万
依托单位国家:
美国
项目类别:
Continuing Grant
财政年份:
2017
资助国家:
美国
项目状态:
已结题
起止时间:
2017-09-01 至 2022-08-31
中文摘要
处理器速度和晶体管数量每18-24个月翻一番的时代,正如所谓的摩尔定律所指出的那样,被认为已经结束,因为相关的功耗增加和技术物理限制。然而,计算机及其应用一直是现代生活的支柱,并将继续以更高的速度发挥作用。由于需要处理的数据量的增加,对提高计算能力的需求将继续上升。此外,为了支持科学发现和工程突破不断增长的需求,对更高精度的需求也将继续增长,随着我们走向更智能的城市和更智能的世界,对更小、更节能的计算设备的需求也在迅速增长。因此,迫切需要寻找更快、更小、更节能的改变游戏规则的计算设备,而不是目前似乎接近令人生畏的物理限制的数字计算机技术。该提案介绍了一种基于模拟计算、可重构计算架构、纳米光子学和材料科学进步的变革性计算设备。这种新的计算设备,被称为可重构光学计算机(ROC),旨在实现处理技术的飞跃,并为未来的持续发展打开大门。这种可重构的光学处理器(ROC)能够通过求解那些在科学和工程中用于大多数模拟的偏微分方程(PDEs)来执行第一性原理的计算,而不是像数字计算机那样迭代。这种pde可以通过电子元件网络来实现和直接求解。这些电路元件将使用可编程的纳米光子器件来实现,这些器件被称为可重构光学元件(ROEs)。在电偏压作用下,ROEs可以改变其光学特性,以表示适合解决不同问题的不同元件。ROEs是一种新型电压可编程光子元件,利用电光效应(即调制)和光吸收(即光探测)来模拟电路元件。该方法包括ROE设计的风险规避策略(从低到高)。支持编程/重构存储单元和使能电子设备将被探索形成一个集成的可重构架构,即可重构光学计算机(ROC)。本文还将研究一种适合这种新范式的软件堆栈,包括编程和执行模块,为领域用户提供一个高效的平台。如果成功,与现代最先进的处理器相比,ROC系统可能在速度、能耗和占地面积方面实现数量级的改进,并可能简化软件,对从电子工业到智能生活的许多领域产生重大影响。
英文摘要
The era of observing processor speed and number of transistors doubling every 18-24 months, as noted in the so-called Moore's law is believed to be over, due to the associated increase in power consumption and technology physical limitations. Computers and their applications, however, have been the backbone of modern life and will continue to be so at even a higher rate. The demands for increased computing capabilities will continue to be on the rise due to the increase in data volumes that need to be processed. Further, the increase in demand for higher accuracy will also continue in order to support the rising needs for scientific discoveries and engineering breakthroughs, and the demand for smaller and more energy efficient computing devices is also rapidly growing as we move towards smarter cities and a smarter world. It is therefore imperative to search for game changing computing devices that are orders of magnitude faster, smaller and more energy efficient as opposed to the current digital computer technology that seem to be approaching daunting physical limitations. This proposal introduces a transformative computing device based on analog computing, reconfigurable computing architectures, nanophotonics, and the advances in material science. This new computing device, which is referred to as the reconfigurable optical computer (ROC), aims to achieve a leap forward in processing technology in addition to opening the doors for continued future developments.This reconfigurable optical processor (ROC) that is capable of performing computations out of first principle by solving those partial differential equations (PDEs) that are used for most simulations in science and engineering in one shot, rather than iteratively as in digital computers. Such PDEs can be realized and directly solved using a network of electric components. Those electric circuit components will be implemented using programmable nanophotonic devices, referred to as the reconfigurable optical elements (ROEs). ROEs can change their optical characteristics under electrical bias to represent different components suitable for solving different problems. ROEs are novel voltage programmable photonic components utilizing electro-optic effects (i.e. modulation) and optical absorption (i.e. photodetection) to mimic electrical circuit elements. The approach includes a risk-tired strategy for ROE design (low-to-high). Supporting programming/reconfiguration memory cells and enabling electronic devices will be explored to form an integrated reconfigurable architecture, the Reconfigurable Optical Computer (ROC). An appropriate software stack including programming and execution modules for such new paradigm will be also investigated to provide a productive platform for domain users. If successful, the ROC system can potentially achieve orders of magnitude improvements in speed, energy consumption and footprint as compared to the modern state-of-the-art processors and potentially simplify software, providing substantial impacts on many domains from the electronic industry to smart life.
期刊论文(6)
专著(0)
科研奖励(0)
会议论文
登录
查看更多内容
DOI:
10.1038/s42005-021-00683-4
发表时间:
2021-08
期刊:
Communications Physics
影响因子:
5.5
作者:
[M. Miscuglio;Yaliang Gui;Xiaoxuan Ma;Zhizhen Ma;Shuai Sun;T. El Ghazawi;T. Itoh;A. Alú;V. Sorger]
通讯作者:
M. Miscuglio;Yaliang Gui;Xiaoxuan Ma;Zhizhen Ma;Shuai Sun;T. El Ghazawi;T. Itoh;A. Alú;V. Sorger
DOI:
10.1515/nanoph-2020-0655
发表时间:
2021-03
期刊:
Nanophotonics
影响因子:
7.5
作者:
[Shuai Sun;M. Miscuglio;Xiaoxuan Ma;Zhizhen Ma;Chen Shen;Engin Kayraklioglu;Jeff Anderson;T. El Ghazawi;V. Sorger]
通讯作者:
Shuai Sun;M. Miscuglio;Xiaoxuan Ma;Zhizhen Ma;Chen Shen;Engin Kayraklioglu;Jeff Anderson;T. El Ghazawi;V. Sorger
Virtualizing a Post-Moore’s Law Analog Mesh Processor: The Case of a Photonic PDE Accelerator
虚拟化后摩尔定律模拟网格处理器:光子 PDE 加速器的案例
DOI:
10.1145/3544971
发表时间:
2022
期刊:
ACM Transactions on Embedded Computing Systems
影响因子:
2
作者:
[Anderson, Jeff, Kayraklioglu, Engin, Imani, Hamid Reza, Shen, Chen, Miscuglio, Mario, Sorger, Volker J., El-Ghazawi, Tarek]
通讯作者:
El-Ghazawi, Tarek
DOI:
10.1145/3380944
发表时间:
2020-03-01
期刊:
ACM TRANSACTIONS ON PARALLEL COMPUTING
影响因子:
1.6
作者:
[Anderson, Jeff, Kayraklioglu, Engin, El-Ghazawi, Tarek]
通讯作者:
El-Ghazawi, Tarek
Software stack for an analog mesh computer: the case of a nanophotonic PDE accelerator
模拟网状计算机的软件堆栈:纳米光子 PDE 加速器的案例
DOI:
10.1145/3387902.3394030
发表时间:
2020
期刊:
17th ACM International Conference on Computing Frontiers
影响因子:
--
作者:
[Kayraklioglu, Engin, Anderson, Jeff, Imani, Hamid Reza, Sorger, Volker, El-Ghazawi, Tarek]
通讯作者:
El-Ghazawi, Tarek
共 6 条
IDIEA-DC: An Infrastructure for Distributed Intelligence Experimentation and Architectures in the Digital Continuum: from IoTs to the Cloud
-
批准号:2038682
-
项目类别:Standard Grant
-
资助金额:$29.99万
-
财政年份:2020
-
负责人:Tarek El-Ghazawi
-
依托单位:
Architecture Support for Advancing PGAS (ASAP)
-
批准号:1547980
-
项目类别:Standard Grant
-
资助金额:$23.0万
-
财政年份:2015
-
负责人:Tarek El-Ghazawi
-
依托单位:
I/UCRC: FRP: Productive Scientific Computing on Heterogeneous Systems
-
批准号:1230815
-
项目类别:Standard Grant
-
资助金额:$10.0万
-
财政年份:2012
-
负责人:Tarek El-Ghazawi
-
依托单位:
I/UCRC Phase II(Renewal): NSF Center for High-Performance Reconfigurable Computing (CHREC)
-
批准号:1161014
-
项目类别:Continuing Grant
-
资助金额:$20.0万
-
财政年份:2012
-
负责人:Tarek El-Ghazawi
-
依托单位:
Collaborative Research: Development of efficient petascale algorithms for inhomogeneous quantum-mechanical systems
-
批准号:0904887
-
项目类别:Standard Grant
-
资助金额:$37.5万
-
财政年份:2009
-
负责人:Tarek El-Ghazawi
-
依托单位:
MRI: Acquisition of a High-Performance Instrument for Interdisciplinary Computational Science and Engineering
-
批准号:0923256
-
项目类别:Standard Grant
-
资助金额:$78.72万
-
财政年份:2009
-
负责人:Tarek El-Ghazawi
-
依托单位:
US-Egypt Workshop on Software Development for Multicore and Heterogeneous Processors, Cairo, Egypt, June 22-24, 2009
-
批准号:0854959
-
项目类别:Standard Grant
-
资助金额:$4.0万
-
财政年份:2009
-
负责人:Tarek El-Ghazawi
-
依托单位:
Collaborative Research (I/UCRC): CHREC -- the Center for High-Performance Reconfigurable Computing
-
批准号:0555794
-
项目类别:Standard Grant
-
资助金额:$0.0万
-
财政年份:2006
-
负责人:Tarek El-Ghazawi
-
依托单位:
CHREC -- The NSF Center for High-Performance Reconfigurable Computing
-
批准号:0706352
-
项目类别:Continuing Grant
-
资助金额:$35.0万
-
财政年份:2006
-
负责人:Tarek El-Ghazawi
-
依托单位:
US-Egypt Cooperative Research: Grid Computing for Authentication of Arabic Text
-
批准号:0109038
-
项目类别:Standard Grant
-
资助金额:$2.5万
-
财政年份:2001
-
负责人:Tarek El-Ghazawi
-
依托单位:
海外基金