Su-Hui Chiang

Learn More
The performance potential of run-to-completion (RTC) parallel processor scheduling policies is investigated by examining whether (1) application execution rate characteristics such as <italic>average parallelism (avg)</italic> and <italic>processor working set (PWS)</italic> and/or (2) limited preemption can be used to improve the performance of these(More)
This paper addresses the question of whether more accurate requested runtimes can significantly improve highperformance priority backfill policies, for production workloads running on leading edge systems such as the 1500 Origin 2000 system at NCSA or the new TeraGRID. This question has been studied previously for FCFS-backfill using a limited set of(More)
We define a rewrite strategy for a class of non-confluent constructor-based term graph rewriting systems and prove its correctness. Our strategy and its extension to narrowing are intended for the implementation of non-strict non-deterministic functional logic programming languages. Our strategy is based on a graph transformation, called bubbling, that(More)
To balance performance goals and allow administrators to declaratively specify high-level performance goals, we apply complete search algorithms to design on-line job scheduling policies for workloads that run on parallel computer systems. We formulate a hierarchical two-level objective that contains two goals commonly placed on parallel computer systems:(More)
Fairness is an important issue for parallel job scheduling policies, but has been ignored in most of previous studies. In this paper, we consider two different styles of job fairness: FCFS and EQ. Commonly used summary statistics are applied to different job measures to evaluate the fairness under a wide range of non-preemptive parallel job scheduling(More)
Bubbling, a recently introduced graph transformation for functional logic computations, is well-suited for the reduction of redexes with distinct replacements. Unlike backtracking, bubbling preserves operational completeness; unlike copying, it avoids the up-front construction of large contexts of redexes, an expensive and frequently wasteful operation. We(More)
To balance multiple scheduling performance requirements on parallel computer systems, traditional job schedulers use many parameters that can be configured to define job or queue priorities. Offering many parameters seems flexible, but in reality tuning the values for the parameters is highly challenging. To simplify the task of resource management, we(More)