Author

Anh Trinh

Publication Date

Fall 2011

Degree Type

Master's Project

Department

Computer Science

Abstract

To avoid overheating the chip, chip designers have switched to multi-cores. While multicore CPUs reserve instruction-level parallelism features that help existing applications run as if they were running under single core, applications do not reach speeds two or four times faster. Instead of relying on compiler and hardware to figure out parallelism in source code, software developers now must control parallelism explicitly in their programs. Many programming languages and libraries, such as Java, C# .NET, and OpenMP, are trying to help programmers by providing rich concurrency API. X10 is the new experimental language from IBM Research, which has been under development since 2004 targeting multi-core programming ranging from multi-cores single machine to cluster. This project examines the X10 parallel constructs, compares its usability with the Java language, the OpenMP library, and then compares the performance between X10 and Java language.

Share

COinS