Parallel and vectorial solving of finite element problems on a shared-memory multiprocessor

A finite element method is presented to compute time harmonic microwave fields in three dimensional configurations. Nodal-based finite elements have been coupled with an absorbing boundary condition to solve open boundary problems. This paper describes how the modeling of large devices has been made possible using parallel computation, New algorithms are then proposed to implement this formulation on a cluster of workstations (10 DEC ALPHA 300X) and on a CRAY C98. Analysis of the computation efficiency is performed using simple problems. The electromagnetic scattering of a plane wave by a perfect electric conducting airplane is finally given as example

show abstract

“…L t (L and L t are computed). The building of the incomplete Cholesky matrix is performed by column (5) [15], [16] (fig. 7).…”

Section: Incomplete Cholesky Preconditioning (Icp)mentioning

confidence: 99%

Parallel computing for the finite element method

Vollaire

Nicolas

1998

Eur. Phys. J. AP

View full text Add to dashboard Cite

show abstract

A Vectorial Finite Element Procedure for Solving Three-Dimensional Field Problems

Cannistra

Minenna

Negros

et al. 1995

Electric and Magnetic Fields

View full text Add to dashboard Cite

Implementing finite element methods on supercomputers, workstations and PCs: a bibliography (1985‐1995)

Mackerle

1996

Engineering Computations

View full text Add to dashboard Cite

IntroductionThe power and range of capabilities of computers is increasing all the time. In retrospect, hardware development from the 1960s to the 1990s can be divided into several phases. At the beginning, there were mainframe computers, mostly UNIVAC 1 and IBM 704, 705. The next phase of hardware development opened with IBM System 360; also UNIVAC continued with new models. Other manufacturers, such as Burroughs, Honeywell, CDC and NCR, were also in the market. In the third period minicomputers have appeared. They are much cheaper than mainframes but are of roughly comparable capacity. Minicomputers led to the development of distributed information systems.In 1971, microprocessors were introduced. In general, they were able to achieve more impressive performance gains. Early microprocessors did not incorporate all the architectural advances which are present today in supercomputers, and their speeds were slow. After 1978, PCs and, later on, super-minicomputers start to appear, bringing the computer to the user's desk. Many new firms entered the market at this time. In the late 1970s old and wellknown vendors were starting in the field of personal computing, office automation and later on in local networking.During the 1980s the main emphasis in hardware development was on supercomputers, with pioneers such as Control Data, Cray Research, Amdahl, Hitachi, etc. The traditional supercomputers were equipped with a singleprocessor. The supercomputer sector continued with the application of multiprocessors, working in parallel, and with the development of superior software. The first commercial parallel computers containing microprocessors were available on the market in the mid-1980s. The late 1980s saw relatively complex applications of supercomputers.

show abstract

Parallel and vectorial solving of finite element problems on a shared-memory multiprocessor

Cited by 6 publications

References 5 publications

Parallel computing for the finite element method

Parallel computing for the finite element method

A Vectorial Finite Element Procedure for Solving Three-Dimensional Field Problems

Implementing finite element methods on supercomputers, workstations and PCs: a bibliography (1985‐1995)

Contact Info

Product

Resources

About