登入選單
返回Google圖書搜尋
Scalability Analysis of Static Load Balancing Under Unpredictable Subproblem Sizes
註釋Abstract: "We investigate the balance of load between processors in parallel execution, in which a given problem consists of many subproblems of unpredictable different sizes. If we solve each subproblem at a different processor with a polynomial-time algorithm of degree d [> or =] 1, unevenness in the subproblem size is translated into larger unevenness (according to d) in the load between processors. However, we have found that we can almost balance the load between processors by assigning only a modest number of subproblems to each processor. Namely, an [omega](log [superscript d] p) number of subproblems per processor is sufficient, where p denotes the number of processors.