Gyoten proposed a new processor architecture at ICCAD 2018.
Gyoten’s talk was about a processor utilizing the Ising model for solving combinatorial optimization problems. The Ising model is a mathematical model for analyzing ferromagnetism. Its application as an optimization problem solver is gaining increasing attention as one of the next-generation computer technologies. However, the solution quality obtained by existing Ising processors is not sufficient, which makes it difficult to give a feasible solution that satisfies all the constraints placed on the given problem. In this presentation, the application of parallel tempering to the Ising-model solver is proposed to improve the solution quality. Through hardware friendly approximations of the parallel tempering, the performance of the proposed processor is evaluated. When solving max-cut problems, the proposed architecture obtained optimal solutions for the problem that an existing Ising processor could not, and the proposed processor can obtain optimal solutions two orders of magnitude faster than an existing software solver._
- Hidenori Gyoten, Masayuki Hiromoto, and Takashi Sato, “Enhancing the solution quality of hardware Ising-model solver via parallel tempering,” in Proc. IEEE/ACM International Conference on Computer-Aided Design (ICCAD), November 2018.