Java Grande Benchmarking

The Java Grande Forum (JGF) is working on developing the Java language and class libraries to support ``Java Grande'' applications, which are typically computationally intensive science and engineering applications requiring high-performance computers. To help in this goal, we are working on benchmarking various aspects of Java performance that are relevant for Java Grande applications. We have developed a set of benchmarks to complement the Java Grande Forum Benchmark Suite, and have measured the performance of our benchmarks and the JGF benchmarks on a variety of platforms and JDKS.

DHPC Java Benchmarks

The DHPC Java Grande Benchmarks are divided into two sections: Section 1a contains generic low-level operations, while Section 2a contains short application kernels. These are designed to augment Sections 1 and 2 of the JGF Benchmark Suite.

Section 1a: Low Level Operations

Object Creation
Array Creation
Exception Throwing
Thread Switching
Method Calling
Looping
Generic Code Segments

Section 2a: Kernels

Fibonacci series
Tower of Hanoi
Sieve of Eratosthenes for finding prime numbers
2-D Fast Fourier Transform
NAS EP benchmark generating Gaussian random numbers
NAS IS benchmark for integer sorting

The DHPC Java Grande Benchmarks are available for download as a compressed tar file. (33K)

We are currently working on implementing Java versions of other NAS benchmarks.

Results

Results are available for both the JGF and DHPC Java Grande Benchmarks on a variety of platforms and JDKs.

Technical Reports

DHPC-063: Analysis and Development of Java Grande Benchmarks, J.A. Mathew, P.D. Coddington and K.A. Hawick, January 1999.

Contact Ken Hawick with comments, suggestions, or requests for more information.

This work is being done by Paul Coddington, Ken Hawick and Jesudas Mathew.

Distributed High-Performance Computing Group
Advanced Computational Systems Cooperative Research Centre