Dynamic vs. Static Quantum-Based Parallel Processor Allocation
نویسندگان
چکیده
This paper improves upon previous synthetic workload models and compares the performance of dynamic spatial equipartitioning (EQS) and the semi-static quantum-based FB-PWS processor allocation deened in 23], under synthetic workloads that have not previously been considered. These new workloads include realistic repartitioning overheads and job characteristics that are consistent with system measurement, anticipated trends, and experience. The overall conclusion from the results is that the EQS policy is generally superior to the FB-PWS policy even under realistic repartitioning overheads. We nd cases where the EQS system saturates earlier than the FB-PWS system, and vice versa. This leads to the deenition of a modiied EQS policy, called EQS-PWS, which has performance equal to or better than EQS and FB-PWS for all workloads examined in this paper.
منابع مشابه
Dynamic Processor Allocation in Large Mesh-Connected Multicomputers
Current processor allocation techniques for highly parallel systems are based on centralized front-end based algorithms. As a result, the applied strategies are restricted to static allocation, low parallelism and weak fault tolerance. To lift these restrictions we are investigating a distributed approach to the processor allocation problem in large meshconnected multicomputers. A noncontiguous...
متن کاملDynamic vs . Static Quantum -
This paper improves upon previous synthetic workload models and compares the performance of dynamic spatial equipartitioning (EQS) and the semi-static quantum-based FB-PWS processor allocation deened in 23], under synthetic workloads that have not previously been considered. These new workloads include realistic repartitioning overheads and job characteristics that are consistent with system me...
متن کاملMaximizing Speedup through Self-Tuning of Processor Allocation - Parallel Processing Symposium, 1996., Proceedings of IPPS '96, The 10th International
We address the problem of maximizing application speedup through runtime, self-selection of an appropriate number of processors on which to run. Automatic, runtime selection of processor allocations is important because many parallel applications exhibit peak speedups at allocations that are data or time dependent. We propose the use of a runtime system that: (a) dynamically measures job eficie...
متن کاملParallel Virtual Memory for Time Shared Environments
This paper analyses the issues involved in providing virtual distributed shared memory for time-shared parallel machines. We study the performance of two di erent page management policies, namely, static and dynamic page allocation under two widely accepted scheduling policies: Gang scheduling and independent processor scheduling. The performance of each page management policy is studied under ...
متن کاملMaximizing Speedup through Self-Tuning of Processor Allocation
We address the problem of maximizing the speedup of an individual parallel job through the selection of an appropriate number of processors on which to run it. If a parallel job exhibits speedup that increases monotonically in the number of processors, the solution is clearly to make use of all available processors. However, many applications do not have this characteristic: they reach a point ...
متن کامل