Task migration in hypercube multiprocessors

Abstract

Allocation and deallocation of subcubes usually result in a fragmented hypercube where even if a sufficient number of hypercube nodes are available, they do not form a subcube large enough to execute an incoming task. As the fragmentation in conventional memory allocation can be handled by memory compaction, the fragmentation problem in a hypercube can be solved by task migration , i.e., relocating tasks within the hypercube to remove the fragmentation. The procedure for task migration closely depends on the subcube allocation strategy used, since active tasks must be relocated in such a way that the availability of subcubes can be detected by that allocation strategy. In this paper, we develop a task migration strategy for the subcube allocation policy based on the binary reflected Gray code. A goal configuration (of destination subcubes) without fragmentation is determined first. Then, the node-mapping between the source and destination subcubes is derived. Finally, a routing procedure to achieve shortest deadlock-free paths for relocating tasks is developed.