Shared-memory parallelization of the semi-ordered fast iterative method

نویسندگان

  • Josef Weinbub
  • Florian Dang
  • Tor Gillberg
  • Siegfried Selberherr
چکیده

The semi-ordered fast iterative method is used to compute a monotone front propagation of anisotropic nature by solving the eikonal equation. Compared to established iterative methods, such as the fast iterative method, the semi-ordered fast iterative method (SOFI) offers increased stability for variations in the front velocity. So far, the method has only been investigated in a serial, two-dimensional context; in this paper we investigate a parallel implementation of SOFI (using OpenMP) and evaluate the method for three-dimensional realworld type problems. We discuss the parallel algorithm and compare its performance and its computed solutions with an OpenMP-powered fast iterative method. Different speed functions together with varying problem sizes are used to investigate the impact of the computational load. Although the semi-ordered fast iterative method is inferior to the fast iterative method with respect to parallel efficiency, we show that its execution performance is significantly faster. Author

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

منابع مشابه

Comparison of the Parallel Fast Marching Method, the Fast Iterative Method, and the Parallel Semi-Ordered Fast Iterative Method

Solving the eikonal equation allows to compute a monotone front propagation of anisotropic nature and is thus a widely applied technique in different areas of science and engineering. Various methods are available out of which only a subset is suitable for shared-memory parallelization, which is the key focus of this analysis. We evaluate three different approaches, those being the recently dev...

متن کامل

Strategies to hide communication for a classical molecular dynamics proxy application

Issakar Ngatang and Masha Sosonkina Summary Co-designing applications and computer architectures has become of major importance due to the growing complexity of both applications and architectures and the need to better match application characteristics to the available hardware. Thus, “miniapplications”, which serve as proxies of large-scale ones by highlighting their most intensive parts and ...

متن کامل

Shared-memory parallelization of the fast marching method using an overlapping domain-decomposition approach

The fast marching method is used to compute a monotone front propagation of anisotropic nature by solving the eikonal equation. Due to the sequential nature of the original algorithm, parallel approaches presented so far were unconvincing. In this work, we introduce a shared-memory parallelization approach which is based on an overlapping domain decomposition technique. We introduce our paralle...

متن کامل

Solution of Singular Perturbation Problems via the Domain Decomposition Method on Serial and Parallel Computers

Domain decomposition methods for the solution of semi-linear selfadjoint and non-selfadjoint singularly perturbed di erential equations are considered. An iterative domain decomposition algorithm suitable for parallelization is described and convergence of the algorithm is established. The implementation on a shared memory multiprocessor is described and numerical examples are presented to demo...

متن کامل

A Semi-Ordered Fast Iterative Method (SOFI) for monotone front propagation in simulations of geological folding

This paper present a novel algorithm for monotone front propagation of anisotropic nature. In several examples the new algorithm is shown to be fast and able to solve a general class of front propagation problems. The algorithm is inspired by Huygens’ principle in that the front is described using a list of nodes that are used as source points to evolve the front. Nodes affected by the source p...

متن کامل

ذخیره در منابع من


  با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

عنوان ژورنال:

دوره   شماره 

صفحات  -

تاریخ انتشار 2015