INSPIRE Track 1: SPINTOP: Spin Torque Nano Oscillator Arrays for "Big Data" Applications

概述：该INSPIRE奖部分由工程局电气、通信和网络系统部门的电子、光子学和磁性设备项目资助；计算机与信息科学与工程学部计算与通信基础学部软硬件基础项目；数学和物理科学理事会材料研究部的电子和光子材料项目。SPINTOP将详细探索一种新的自动机计算范式，该范式基于非布尔、非冯·诺伊曼架构，并通过自旋扭矩纳米振荡器（STNOs）的电耦合实现。该项目的目标是构建一个简单而清晰的演示，说明如何在模式识别方面实现显著的加速，以匹配详细的模拟。该项目通过一种非传统的解决方案来应对“大数据”挑战，这种解决方案可以在飞行中进行计算。INSPIRE项目的跨学科性质将使磁性材料研究与设备设计和测试以及计算机体系结构开发相结合成为可能，涉及三个NSF项目和弗吉尼亚大学（UVa）物理学、材料科学和电子与计算机工程的教师。智力优势：SPINTOP将推动自旋电子器件、电路和计算架构的前沿。该项目成功的关键将是UVa发明的新型混合STNO的演示。具有低磁化强度、低阻尼、高自旋极化和可控各向异性的磁性材料是该混合器件成功的关键。提出的新型STNO结合了现有器件的最佳特性，提供纯正弦波输出和大电压信号。这些STNOs阵列的电耦合将在本项目中进行演示。根据详细的模拟，锁相将发生在它们各自的频率接近时——这种锁相将发生在整个阵列上，而不是像之前使用自旋波耦合所证明的那样，只发生在最近邻锁定。这种相参锁相是设计用于模式识别的联想存储器的关键。SPINTOP还将开发用于模拟STNO阵列创新设计行为的建模工具。该工具的进一步发展是设计和实现电路的关键，因为它将与现有的代码集成在一起，用于执行混合cmos -自旋电子电路的建模，仿真和设计。更广泛的影响：SPINTOP如果成功，将通过创建计算的新范式，特别是模式识别，对科学界产生重大影响。这将有助于解决当今面临巨大挑战的许多大数据应用。然而，其影响将更为广泛；这个项目将解决当今社会面临的许多关键问题。该项目致力于促进劳动力的多样性，并将在项目期间支持至少一名来自代表性不足群体的学生。由于INSPIRE项目将隶属于UVA纳米之星研究所，而PI也是纳米之星的主管，因此该项目将与外展计划紧密结合，并从纳米之星的资源中受益。SPINTOP的教师和学生将参加对弗吉尼亚州高中的外展访问，以突出这项新技术。SPINTOP的教师和学生还将参加每年春天举办的纳米日活动，这是一个高度互动的活动，将650多名当地K-12学生及其父母带到弗吉尼亚大学，体验纳米科学和技术的一些奇迹。这些参观者将看到一张突出SPINTOP的海报，并被邀请参观SPINTOP实验室。SPINTOP和NanoSTAR将共同赞助一个名为NeXT（纳米和新兴技术）的大学生俱乐部，该俱乐部经常举办有嘉宾演讲的研讨会，并将成为许多纳米相关学科的学生相互联系的一种方式。SPINTOP将利用支持本科生研究的项目来招募本科生参加SPINTOP实验室的夏季研究。与弗吉尼亚大学工程多样性中心合作，将为几名少数民族高中生提供在SPINTOP设施中体验研究的机会。SPINTOP也将对弗吉尼亚大学的课程产生影响；在这个项目的支持下，至少会有一个新的班级。这是一门关于基于自动机的计算的课程，这是SPINTOP项目将用于系统级映射应用程序的计算范式。最后，由PI教授的现有自旋电子学课程将更新为包括自旋扭矩振荡器及其在计算中的应用的几堂课。

Overview:This INSPIRE award is partially funded by the Electronics, Photonics, and Magnetics Devices Program in the Division of Electrical, Communications and Cyber Systems in the Directorate for Engineering; the Software and Hardware Foundations program in the Division of Computing and Communication Foundation in the Directorate for Computer & Information Science and Engineering; and the Electronic and Photonic Materials program in the Division of Materials Research in the Directorate for Mathematical and Physical Sciences.SPINTOP will explore, in detail, a new paradigm of automata computing that is based on a non-Boolean, non-Von Neumann architecture and is enabled by the electrical coupling of Spin Torque Nano-oscillators (STNOs). The project's goal is to build a simple, yet clear demonstration, of how this can achieve significant speedups in pattern recognition to match detailed simulations. This project addresses the 'Big Data' challenge with a non-conventional solution that can compute on the fly. The interdisciplinary nature of the INSPIRE program will make it possible to combine magnetic materials research with device design and test and computer architecture development involving three NSF Programs and faculty from Physics, Materials Science and Electronic and Computer Engineering at University of Virginia (UVa).Intellectual Merit:SPINTOP will be pushing the frontiers of spintronic devices, circuits and computing architectures. Key to the success of this project will be the demonstration of a novel hybrid STNO that was invented at UVa. Magnetic materials with low magnetization, low damping, high spin polarization, controlled anisotropy are key to the success of the proposed hybrid device. The proposed novel STNO combines the best features of existing devices to provide both pure sine wave output and a large voltage signal. Electrical coupling of arrays of these STNOs will be demonstrated in this project. According to detailed simulations, phase lock will occur when their individual frequencies are close - this phase locking will take place across the array rather than only nearest neighbor locking as has been previously demonstrated using spin wave coupling. This coherent phase locking is the key to the design for an associative memory for pattern recognition. SPINTOP will also develop the modeling tool used to simulate the behavior of the innovative design for STNO arrays. Further development of this tool is key for designing and implementing circuits as it will be integrated with existing codes for performing the modeling, simulation and design of hybrid CMOS-spintronic circuits. Broader Impacts:SPINTOP, if successful, will have a big impact on the scientific community by creating a new paradigm for computing, especially pattern recognition. This will help solve many of the Big Data applications that are a grand challenge today. However its impact will be much broader; this project will address many of the key issues facing society today. This project is committed to fostering diversity in the workforce and will support at least one student from an underrepresented group for the duration of the project. As the INSPIRE project will fall under the umbrella of the UVA nanoSTAR Institute and the PI is also the director of nanoSTAR, this project will be closely integrated into the outreach programs and benefit from the resources of nanoSTAR. The SPINTOP faculty and students will participate in outreach visits to Virginia high schools to highlight this new technology. SPINTOP faculty and students will also participate in Nanodays, a highly interactive event held each Spring that brings over 650 local students from K-12 and their parents to UVA to experience some of the wonders of nanoscience and technology. These visitors will see a poster highlighting SPINTOP and will be invited to tour the SPINTOP laboratories. SPINTOP along with NanoSTAR will co-sponsor a club for undergraduates called NeXT (for Nano and Emerging Technologies) that hosts frequent seminars with guest speakers and will be a way for students in many of the nano-related disciplines to connect with one another. SPINTOP will leverage programs that support undergraduate research to recruit undergraduates to participate in summer research in the SPINTOP labs. In collaboration with UVA's Center for Diversity in Engineering, opportunities will be provided to several minority senior high school students to experience research in the SPINTOP facilities. SPINTOP will also have an impact on courses at UVA; there will be at least one new class developed under the auspices of this program. It will be a class on automata based computing which is the computing paradigm that the SPINTOP project will use for mapping applications at the system level. Finally, an existing class on Spintronics taught by the PI will be updated to include several lectures on spin torque oscillators and their application to computing.