Spiral: A heuristic mapping algorithm for network on chip

Abstract

Abstract:In this paper, a heuristic core mapping algorithm for 2-D mesh topologies called Spiral is proposed. To compare the efficacy of the algorithm, the results of the proposed algorithm are compared with those of the genetic and random mapping algorithms. The experimental results of synthetic traffic profiles reveal that the Spiral algorithm improves the reduction in the energy consumption as well as the speed of performing application compared to those of the genetic algorithm mapping.

DSM: A Heuristic Dynamic Spiral Mapping algorithm for network on chip

“…Regarding the application mapping optimization, the mapping and routing path allocation problem for tile based architectures have been addressed in [3,4,5,6,7,8]. In [5], an energy aware mapping and scheduling approach has been addressed.…”

“…This is performed based on minimizing the reconfiguration time, the overall energy consumption over the communication links, reducing the algorithm complexity in terms of reducing the algorithm execution time. In this paper our focus is on the first measure because using the Spiral mapping will guaranty the other optimization measures [6]. First lets consider some definitions.…”

“…DSM uses the Spiral mapping algorithm where the priority assignment policies during the mapping process is expressed in [6].…”

“…The pre-mapping process during the design time is executed by the Static Spiral Mapping (SSM) [6] during the runtime, DSM will reconfigure and optimize this configuration and will result better performance in executing the application or even changing into another application. For dynamic Mapping problem, we have developed two different approaches based on the DSM concept which are called Full Dynamic Spiral Mapping (FDSM) and Partial Dynamic Spiral Mapping (PDSM) respectively.…”

“…This is the time needed for an application to reorganize the current task configuration to a new one. Since this algorithm is based on the static Spiral mapping [6], other metrics such as the cumulative energy consumption over the communication links, the mapping algorithm complexity, and the overall system completion time have been considered implicitly.…”