Research Summary

Group Member: Viktor Decyk

 

High Performance Computing

The purpose of this research is to develop strategies, algorithms, and techniques to enable effective use of high performance computers for the solution of large-scale scientific problems.


BASIC SCIENTIFIC PROGRAMMING

Why Fortran?
This Op-Ed article describes why Fortran still dominates high performance scientific computing. One reason: it is not your grandfather's FORTRAN.

Fortran 95 Lectures
These lectures are intended to introduce Fortran77 programmers to the new features of Fortran95. Strategies for how to modernize old Fortran77 codes incrementally, as well as object-oriented design in Fortran 95 will be discussed.

Modernization of Fortran Legacy Code
This presentation describes how to use the object-based features of Fortran95 to modernize legacy Fortran77 codes.

Optimization techniques for RISC processors
This preprint describes general strategies for optimizing particle-in-cell-codes written in Fortran for RISC processors. These strategies include data reorganization to improve cache utilization and code reorganization to improve efficiency of arithmetic pipelines. Results show performance improvements of 1.4 to 3.4 times can be achieved. Final version published in Computers in Physics, 10 (3) May/June 1996.

Object-Oriented Programming in Fortran 90
This web site describes modern features of Fortran90, an object-based language useful for complex scientific programs.

A Parallel Macintosh Cluster

 

ALGORITHMS FOR PARTICLE-IN-CELL CODES

UPIC: UCLA Parallel PIC Framework

 

BASIC PLASMA PHYSICS

Simulation of Microscopic Processes in Plasma