Controlling computationally intensive heterogeneous computational tasks with directive deadlines

et al. 2014

Self Cite

The CORSAR model software complex designed to analyze scheduling rules for resource intensive heterogeneous problems of search type that can be parallelized with respect to data is described. In the course of searching for solutions, the original arrays are decomposed into separate informative fragments so that one can use sliding scheduling methods and optimize parallelizing pro cesses when processing requests. Tasks are executed on a high performance heterogeneous specialized computing system. The developed multiparametric model control scheme is invariant to types of com putational resources, ensures their operation efficiency and scalability when the system is updated and evolving.

show abstract

Section: Introductionmentioning

confidence: 99%

Control of resource-intensive computations under uncertainty. II. Scheduling complex

Kupalov-Yaropolk

et al. 2014

Self Cite

show abstract

“…The sched uling system based on the MP model supports the sliding scheduling mode, performs data parallelization and optimizes the composition of the batch of current jobs so that tasks can be completed by their dead lines [5,6] and production loss can be minimized [7].…”

Section: Introductionmentioning

confidence: 99%

Control of resource-intensive computations under uncertainty. III. Dynamic concurrent resource allocation

2015

Self Cite

Control problems for a heterogeneous system executing heterogeneous search tasks are considered. Data parallelization of tasks under objective uncertainty-for the unknown and rapidly changing flow of incoming requests-is studied. Critical modes of operation under big system work loads are analyzed. The modified sliding scheduling method allows correcting the processing proce dure in real time. A way of dynamic construction of a sequence of non discriminating concurrent allo cations of limited different type resources between users with equal rights is proposed. A concept of normative maximal performance of the system processing one type of tasks in the single user mode is introduced. Parametric guaranteed estimates are used to decompose tasks with respect to their type. Scheduling procedures according to which tasks that de facto require less computational costs receive the processing priority and are completed first of all are developed.

show abstract

Data aggregation for scheduling of resource-intensive computations under uncertainty

2014