Task assignment in heterogeneous computing systems

Abstract

The problem of task assignment in heterogeneous computing systems has been studied for many years with many variations. We consider the version in which communicating tasks are to be assigned to heterogeneous processors with identical communication links to minimize the sum of the total execution and communication costs. Our contributions are three fold: a task clustering method which takes the execution times of the tasks into account; two metrics to determine the order in which tasks are assigned to the processors; a refinement heuristic which improves a given assignment. We use these three methods to obtain a family of task assignment algorithms including multilevel ones that apply clustering and refinement heuristics repeatedly. We have implemented eight existing algorithms to test the proposed methods. Our refinement algorithm improves the solutions of the existing algorithms by up to 15% and the proposed algorithms obtain better solutions than these refined solutions.

Correlation-Aware Object Placement for Multi-Object Operations

“…Our problem is also similar to a variant of the task assignment problem [6,7,12], which allocates a number of tasks to a distributed system of heterogeneous processors such that the total execution and communication cost are minimized while per-processor constraints are satisfied. Task assignment problems typically have heuristic solutions that focus on online efficiency for small problem sizes.…”

Edge-pancyclicity and edge-bipancyclicity of faulty folded hypercubes

“…The pancyclicity is an important measurement of whether a network is suitable for an application inquiring cycles of any length within the network [6]. In a heterogeneous computing system, each edge and each vertex may be assigned with distinct computing power and distinct bandwidth, respectively [18]. Thus, it is worthwhile to extend pancyclicity to edge-pancyclicity and vertex-pancyclicity.…”

Task distribution based adaptation in mobile patient monitoring systems

“…• The majority of the existing algorithms are proposed based on a rather abstract cost model, e.g., the total computation cost and communication cost [Lo88,NT93,UAKI06]. However, when dealing with a real application in an MPMS, we are interested in a measure that reflects more realistically the system performance.…”

“…In [UAKI06], a set of "Multilevel Clustering" heuristic algorithms are proposed. This algorithm works in the same problem model as Lo's "Grab-Lump-Greedy" algorithm and Kopidakis's "Max Edge" algorithm.…”

“…The edge set of each supertask is set to the weighted union of the edge sets of its member tasks, where the internal edges are deleted. Four different clustering algorithms are proposed in [UAKI06]. Among them the "Matching-based clustering" offers the best performance and it works as follows: For each edge (i, j) in G l , a clustering profit for tasks i and j is calculated.…”