The massively parallel computing model GCA

Abstract

The Global Cellular Automata Model (GCA) is an extension of the Cellular Automata Model (CA). Whereas in the CA model each cell is connected via fixed links to its local neighbors, in the GCA model each cell is connected via data dependent dynamic links to any (global) cell of the whole array. The GCA cell state does not only contain data information but also link information. The cell state is synchronously updated according to a local rule, modifying the data and the link information. Similar to the CA model, only the own cell state is modified. Thereby write conflicts cannot occur. The GCA model is related to the CROW (concurrent read owners write) model and it can be used to describe a large range of applications. GCA algorithms can be described in the language GCA-L which can be compiled into different target platforms: a generated data parallel multi-pipeline architecture, and a NIOS II multi-softcore architecture. © 2011 Springer-Verlag Berlin Heidelberg.