The main objective of the processor allocation in hypercube systems is to maximize the resource(processor) utilization as well as to minimize the inherent systems fragmentation. An efficient processor allocation scheme improves the system performance. But the processor allocation problem is nontrivial due to the specific structure of the hypercube, and numerous research efforts have been undertaken.
In this thesis, we propose a processor allocation scheme in hypercube systems that is called CCCC(Cube Coalescing by Cycle Construction). The allocation scheme maintains a graph of free subcubes available in the system. In compared to the other allocation schemes, it uses the information about the released subcube when deallocation subcubes. Particularly, it forms higher dimensional subcubes by detection the cycles starting from the node in the graph which is corresponding to the released subcube.
Extensive simulation runs are carried out to collect experimental measures of different allocation strategies. It is shown from experimental results that the allocation scheme proposed in this thesis outperforms other allocation schemes, the buddy strategy and the gray code strategy, in many situations.