Collaborative Research: Reversible Computing and Reservoir Computing with Magnetic Skyrmions for Energy-Efficient Boolean Logic and Artificial Intelligence Hardware

合作研究：用于节能布尔逻辑和人工智能硬件的磁斯格明子可逆计算和储层计算

• 批准号: 2343606
• 负责人: Jean Anne Incorvia
• 金额: $ 25万
• 依托单位: University of Texas at Austin
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2024
• 资助国家: 美国
• 起止时间: 2024-03-15 至 2027-02-28
• 项目状态: 未结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=2343606&HistoricalAwards=false
• 关键词: Collaborative; Research; Reversible; Computing; Reservoir

As the increasing pervasiveness of computers throughout society has led to drastic increases in the energy consumed by computers, there is a strong need to improve the energy efficiency of computers. In addition to minimizing the economic and environmental costs resulting from computer energy consumption, enhancing the energy efficiency of computers will also allow for increased computing capabilities with beneficial impacts throughout society. This project will therefore design and experimentally demonstrate computing systems with extreme energy efficiency, both for high-performance computing and for artificial intelligence. This extreme energy efficiency will be achieved by leveraging magnetic skyrmions, which are magnetic quasiparticles that have been predicted to be suitable for energy-efficient conventional computing in the reversible computing paradigm as well as for energy-efficient artificial intelligence through the reservoir computing paradigm. This project will significantly advance the development of computers with extreme energy efficiency, thereby reducing environmental harm, facilitating economic development, and enabling revolutionary computing applications that require minimal power dissipation. This project will also have beneficial impacts on workforce development through the inclusion of undergraduate research participants and the vertical training of graduate students from devices to systems.This project will design and experimentally demonstrate reversible and reservoir computers with magnetic skyrmions. Magnetic skyrmions are swirls of magnetic spin texture that are energy-protected once created. They are tunable in size, can operate at room temperature, and have dynamical response to current, voltage, and field, making them a good choice for use in future computing paradigms. Based on preliminary designs and simulations of the PIs on the efficient use of skyrmions in reversible and reservoir computing, this project has four main objectives. Firstly, skyrmion reversible computer co-design will be carried out to efficiently drive the skyrmions, to determine optimal parameters, and to develop a roadmap for the future of skyrmion reversible computing. Secondly, skyrmion reversible computer fabrication will be carried out to demonstrate and analyze skyrmion stability, voltage-driven skyrmion propagation, skyrmion interactions mediated by the skyrmion-Hall effect, and reversible skyrmion logic gates. Thirdly, skyrmion reservoir computing co-design will be carried out to maximize the reservoir expressivity and energy efficiency, to determine optimal parameters, and to develop a roadmap for the future of skyrmion reservoir computing. And fourthly, skyrmion reservoir computer fabrication will be carried out to measure multi-skyrmion interactions, to control repeatability through pinning sites, and to demonstrate and characterize skyrmion reservoir computers. This project will advance the science of thin film magnetism, improve understanding of the dynamics of magnetic nanostructures, develop new device designs and fabrication methods for skyrmion-based devices, and develop and implement circuits and systems to leverage those dynamics. This project will thus advance knowledge in materials, devices, and computing.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

随着计算机在社会中的日益普及，导致计算机消耗的能源急剧增加，因此迫切需要提高计算机的能源效率。除了最大限度地减少计算机能源消耗造成的经济和环境成本外，提高计算机的能源效率还可以提高计算能力，从而对整个社会产生有益的影响。因此，该项目将设计并实验演示具有极高能效的计算系统，无论是高性能计算还是人工智能。这种极端的能源效率将通过利用磁性斯格明子来实现，磁性斯格明子是磁性准粒子，预计适用于可逆计算范式中的节能传统计算以及通过储层计算范式的节能人工智能。该项目将显着推进具有极高能源效率的计算机的开发，从而减少环境危害，促进经济发展，并实现需要最小功耗的革命性计算应用。该项目还将通过纳入本科生研究参与者以及对研究生从设备到系统的垂直培训，对劳动力发展产生有益的影响。该项目将设计并实验演示具有磁性斯格明子的可逆计算机和储层计算机。磁性斯格明子是磁性自旋结构的漩涡，一旦产生就受到能量保护。它们的尺寸可调，可以在室温下运行，并且对电流、电压和场具有动态响应，这使得它们成为未来计算范例的良好选择。基于对斯格明子在可逆和储层计算中的有效利用的 PI 的初步设计和模拟，该项目有四个主要目标。首先，将进行斯格明子可逆计算机协同设计，以有效驱动斯格明子，确定最佳参数，并制定斯格明子可逆计算未来的路线图。其次，将进行斯格明子可逆计算机制造，以演示和分析斯格明子稳定性、电压驱动的斯格明子传播、斯格明子霍尔效应介导的斯格明子相互作用以及可逆斯格明子逻辑门。第三，将进行斯格明子储层计算协同设计，以最大化储层表现力和能量效率，确定最佳参数，并制定斯格明子储层计算未来的路线图。第四，将进行斯格明子储层计算机制造，以测量多个斯格明子相互作用，通过钉扎位点控制可重复性，并演示和表征斯格明子储层计算机。该项目将推进薄膜磁性科学，提高对磁性纳米结构动力学的理解，开发基于斯格明子的器件的新器件设计和制造方法，并开发和实现利用这些动力学的电路和系统。因此，该项目将推进材料、设备和计算方面的知识。该奖项反映了 NSF 的法定使命，并通过使用基金会的智力价值和更广泛的影响审查标准进行评估，被认为值得支持。