The pressure equation, generated while solving the incompressible Navier–Stokes equations with the segregated iterative algorithm such as PISO, produces a series of linear equation systems as the time step advances. In this paper, we target at accelerating the iterative solution of these linear systems by improving their initial guesses. We propose a weighted group extrapolation method to obtain a superior initial guess instead of a general one, the solution of the previous linear equation system. In this method, the previous solutions that are used to extrapolate the predicted solutions are carefully organized to address the oscillatory solution on each grid. The proposed method uses a weighted average of the predicted solutions as the new initial guess to avoid over extrapolating. Three numerical test results show that the proposed method can accelerate the iterative solution of most linear equation systems and reduce the simulation time up to 61.3%.

Designing and implementing a large-scale parallel system can be time-consuming and costly. It is therefore desirable to enable system developers to predict the performance of a parallel system at its design phase so that they can evaluate design alternatives to better meet performance requirements. Before the target machine is completely built, the developers can always build an symmetric multi-processor (SMP) for evaluation purposes. In this paper, we introduce an SMP-based discrete-event execution-driven performance simulation method for message passing interface (MPI) programs and describe the design and implementation of a simulator called SMP-SIM. As the processes share the same memory space in an SMP, SMP-SIM manages the events globally at the granularity of central processing units (CPUs). Furthermore, by re-implementing core MPI point-to-point communication primitives, SMP-SIM handles the communication virtually and sequential computation actually. Our experimental results show that SMP-SIM is highly accurate and scalable, resulting in errors of less than 7.60% for both SMP and SMP-Cluster target machines.

Current parallel programs are often composed of lots of processes that communicate through message passing. The mapping of the program processes onto the underlying processor. By analyzing the MPI library, we find that the communication protocols have great influence upon the communication characteristics. The profiling information used by the existing process placement methods does not include the communication protocols. We propose and implement a method called Protocol-aware process placement (PaPP) to incorporate the influences of communication protocols into the mapping problem. We demonstrate the validations of PaPP by the experiments using NPB3.3-MPI benchmarks. The experimental results show that compared to basic process placement, the average execution time when using PaPP scheme is reduced by 9.5%.