Parallel AI game playing approach for faster processor allocation in hypercube systems using Veitch diagram

In this paper, we propose a quick and complete subcube recognition strategy called "parallel AI game playing approach for faster processor allocation in hypercube systems using Veitch diagram (parallel AIPA)" with much less complexity than the other existing allocation policies. The proposed scheme works by assuming Veitch diagram as a Tetris game board and the incoming subcube requests as falling pieces in Tetris game. Incorporating a game playing approach achieves optimality. The cells in the Veitch diagram attribute the processors. This strategy can be implemented efficiently by using a graph coloring approach with a resultant penalty factor computation. This algorithm deals with cubic as well as noncubic allocation. It is not only statically optimal but also optimal in a dynamic environment. Extensive performance analysis with simulation runs are carried out in comparison with other existing allocation strategies and results are discussed. It is shown that our approach proves to be better in allocation and deallocation costs, memory overhead and minimizes the system fragmentation. Furthermore, by incorporating temporal parallelism in our strategy, the time complexity is also improvised. Simulation results illustrate that the parallel AIPA significantly improves the performance.