Development of an Optimizer for Vortex Transitional Memory Using Particle Swarm Optimization
| Autor: | Masamitsu Tanaka, Ali Bozbey, Mustafa Eren Celik, Seda Demirhan, Mustafa Altay Karamuftuoglu, Akira Fujimaki, Yuto Komura |
|---|---|
| Přispěvatelé: | TOBB ETU, Faculty of Engineering, Department of Electrical & Electronics Engineering, TOBB ETÜ, Mühendislik Fakültesi, Elektrik ve Elektronik Mühendisliği Bölümü, Bozbey, Ali, E-2738-2010 |
| Jazyk: | angličtina |
| Rok vydání: | 2016 |
| Předmět: |
Josephson effect
Working set 02 engineering and technology 01 natural sciences Computational science superconductor Margin (machine learning) Memory 0103 physical sciences Electrical and Electronic Engineering 010306 general physics Electronic circuit particle swarm optimization (PSO) vortex transitional memory (VTM) cell Physics Digital electronics business.industry Process (computing) Particle swarm optimization 021001 nanoscience & nanotechnology Condensed Matter Physics Electronic Optical and Magnetic Materials Vortex 0210 nano-technology business optimization |
| Popis: | High-performance computing that involves superconducting digital circuits is one of the promising technologies. A number of groups have already demonstrated working prototypes of CPUs or ALUs. However, one of the bottlenecks of these circuits is that it is very difficult to have large memories with very high speed and low-power consumption. One of the potential candidates compatible with the already available superconducting foundries that might enable on-chip memory is the vortex transitional memory (VTM) cell. However, VTM operation is mainly based on dc I/O rather than single-flux quantum I/O. VTM cell is mainly based on four Josephson junctions and it is a relatively simple circuit. Nevertheless, optimization of such a cell is a hectic process, as VTM cells are not used as a single cell but combined all together inside a connected network. In this study, we report an optimizer that uses particle swarm optimization algorithm and the results of optimization of VTM cells. By starting from random circuit parameters, the optimizer is able to converge on a working set of parameters fabricated using AIST standard process 2 and advanced process 2 processes with a minimum margin of +/- 30% and +/- 15%, respectively. Optimizations are completed in less than 100 h on a 12-core computer. Fabricated VTM cells have 15% operation margin. This work is supported in part by TUBITAK with the Project 111E191 and ALCA-JST entitled "Superconductor Electronic System Combined with Optics and Spintronics." This paper was recommended by Associate Editor C. J. Fourie. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|