Task
Scheduling is important for the performance of applying STAPL in various
applications. STAPL Scheduler is responsible of mapping tasks onto
multiple processors to gain the maximum performance out of the system.
The STAPL scheduler should also provide an interface for the user to have
control of the scheduling process. The scheduling strategy can be internally
selected, or the user can select a specific way, like wave front execution,
or ultimately provide their own scheduling strategies for their best interests.
The scheduling results are passed to the STAPL
Executor. The Executor is responsible for the real dispatching and
execution of the problem.
One of the applications
of STAPL is the ASCI
project. Due to the specific requirement of physical computation, we
develop a scheduling package,
ASCI Scheduler, which uses the STAPL scheduler in its construction and is
an example of a user-defined scheduler.
The following pages present a short
introduction
to task scheduling problems, and some
algorithms used in STAPL Scheduler.
Load Balancing Techniques for Scalable Parallelization of Sampling-Based Motion Planning Algorithms, Adam Fidel, Sam Ade Jacobs, Shishir Sharma, Lawrence Rauchwerger, Nancy M. Amato, Technical Report, TR13-002 , Parasol Laboratory, Department of Computer Science, Texas A&M University, Mar 2013. Technical Report(pdf, abstract)
Task Scheduling and Parallel Mesh-Sweeps in Transport Computations, Nancy M. Amato, Ping An, Technical Report, TR00-009, Department of Computer Science and Engineering, Texas A&M University, Jan 2000. Technical Report(ps, pdf)