Benchmark Results

For the two applications described above, we have sequential and parallel programs to compare performance. The sequential programs were written in Java and/or Fortran 95. The parallel programs, of course, were written in HPJava. For multigrid we also compare with an available HPF code (taken from [4]).

The experiments were performed on the SP3 installation at Florida State University. The system environment for SP3 runs were as follows:

• System: IBM SP3 supercomputing system with AIX 4.3.3 operating system and 42 nodes.
• CPU: A node has Four processors (Power3 375 MHz) and 2 gigabytes of shared memory.
• Network MPI Settings: Shared ``css0'' adapter with User Space(US) communication mode.
• Java VM: IBM 's JIT
• Java Compiler: IBM J2RE 1.3.1

For best performance, all sequential and parallel Fortran and Java codes were compiled using -O5 or -O3 with -qhot or -O (i.e. maximum optimization) flag.