SBIR Phase I: Hybrid Computing Techniques for Quantum-inspired Ising Machines

SBIR 第一阶段：受量子启发的 Ising 机器的混合计算技术

• 批准号: 2233642
• 负责人: William Moy
• 金额: $ 27.5万
• 依托单位: LEVIOSA TECHNOLOGIES LLC
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2023
• 资助国家: 美国
• 起止时间: 2023-05-01 至 2024-10-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=2233642&HistoricalAwards=false
• 关键词: SBIR; Phase; Hybrid; Computing; Techniques

The broader impact/commercial potential of this Small Business Innovation Research (SBIR) Phase I project will result from the development of a high-speed, low-power solver based on standard semiconductor technologies that can directly solve combinatorial optimization problem (COP) problems. COP problems have broad applicability across manufacturing optimization, semiconductor wire routing, logistics planning and execution, and financial portfolio management. However, COP problems are notoriously difficult or even impossible to solve via classical computers on a scale suitable for commercial application. Current state-of-the-art COP solvers need tremendous computing power, are unsuitable for edge computing, rely on undeveloped technology, or use similar algorithms to classical computers. The goal of this project is to provide consumers with a drop-in replacement COP solver that is faster, more precise, more mobile, and more energy efficient than state-of-the-art classical computers and COP solvers. This Small Business Innovation Research (SBIR) Phase I project seeks to develop a complementary metal–oxide–semiconductor (CMOS) based parallel combinatorial optimization problem (COP) computing cluster accessible through a cloud interface. The proposed solution will directly solve COP problems, increasing the speed, precision, and power efficiency compared to classical computers or current quantum-based COP solvers. Additionally, the cluster uses all standard semiconductor technology allowing for near-term hardware manufacturing, unlike current quantum computing competitors. The device also works at room temperature, making it the only suitable edge device for directly solving COP problems. A hybrid computing algorithm combining the custom and classical methods will parallelize the COP solving across numerous custom chips. Many custom chips will be combined into a single cluster to maximize the speed and efficiency of the hybrid algorithms. An introductory cloud service interface will be incorporated with the custom computer cluster to facilitate outside access. The end goal of this project is to create a state-of-the-art scalable, accessible, and economical COP solver that overcomes the inherent disadvantages of quantum and classical 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.

小型企业创新研究(SBIR)第一阶段项目的更广泛影响/商业潜力将来自于基于标准半导体技术的高速、低功耗求解器的开发，该求解器可以直接解决组合优化问题(COP)问题。COP问题广泛适用于制造优化、半导体布线、物流规划和执行以及金融投资组合管理。然而，众所周知，COP问题很难甚至不可能通过适合商业应用的经典计算机来解决。目前最先进的COP解算器需要巨大的计算能力，不适合边缘计算，依赖未开发的技术，或者使用与经典计算机类似的算法。该项目的目标是为消费者提供一种替代COP解算器，它比最先进的经典计算机和COP解算器更快、更精确、更机动、更节能。这个小型企业创新研究(SBIR)第一阶段项目寻求开发一个基于金属氧化物半导体(CMOS)的并行组合优化问题(COP)计算集群，该集群可通过云接口访问。提出的解决方案将直接解决COP问题，与经典计算机或当前基于量子的COP解算器相比，提高了速度、精度和功率效率。此外，与目前的量子计算竞争对手不同，该集群使用所有标准半导体技术，允许短期硬件制造。该设备还可以在室温下工作，使其成为唯一适合直接解决COP问题的边缘设备。一种结合了定制方法和经典方法的混合计算算法将在众多定制芯片上并行求解COP。许多定制芯片将被组合到一个集群中，以最大限度地提高混合算法的速度和效率。将在定制计算机集群中加入介绍性云服务界面，以方便外部访问。该项目的最终目标是创建一种最先进的可扩展、可访问和经济的COP解算器，克服量子计算和经典计算的固有缺点。该奖项反映了NSF的法定使命，并通过使用基金会的智力优势和更广泛的影响审查标准进行评估，被认为值得支持。