Discussion

In this section, we have explored the performance of the HPJava system on two machines using the efficient node codes previously discussed in [20]. In the results here, the HPJOP2 algorithm was applied systematically by hand to the code generated by the existing translator. The next stage of this work is to include this algorithm in the translator.

The speedup of each HPJava application is satisfactory even with expensive communication methods such as Adlib.writeHalo() and Adlib.sumDim(). Moreover, performances on both machines show consistent behavior similar to that seen on the Linux machine. One machine does not have a big advantage over others. Performance of HPJava is good on all machines we have benchmarked. HPJava has an advantage over some systems because performance of HPJava on Linux machines, shared memory machines, and distributed memory machines are consistent and promising. Thus, we hope that HPJava has a promising future, and can be used anywhere to achieve not only high-performance parallel computing but also grid-enabled applications.