Abstract
Abstract
3D Network On Chip(NoC) will provide high multi-core data processing with the minimal amount of energy consumption. It is necessary to consider the architectural limits of three-dimensional NoCs. A deep reinforced framework is proposed that uses router-less NoC for evaluation in the case study. The framework uses prior approaches that may be unreliable because of the difficulties with the design and search process, and the inflexibility that results because of space restrictions. The framework has better loop placements for networking chips with various design constraints. A Monte Carlo search tree is used to develop a Deep Neural Network that searches in parallel for a NoC design.
Subject
General Physics and Astronomy
Reference19 articles.
1. Three-Dimensional Integrated Circuit Design;Pavlidis,2009
2. Networks-on-Chip in a Three-Dimensional Environment: A Performance Evaluation;Feero;IEEE Trans. on Computers,2008
3. Predicting the Performance of a 3D Processor-Memory Stack;Jacob;IEEE Design and Test of Computers,2005
4. On-Chip Communication Architecture Exploration: A Quantitative Evaluation of Point-to-Point, Bus, and Network-on-Chip Approaches;Lee;ACM Trans. Design Auto. of Electronic Sys.,2007
5. Design and Management of 3D Chip Multi-processors Using Network-in-Memory;Li,2006