Brainy users push super Java
(IDG) -- PALO ALTO, California -- Sun Microsystems this week will meet here with some of the brainiest academics and industry gurus in the high-speed computing field hoping to find a way to overcome barriers to using Java in advanced engineering and math-intensive computing.
James Gosling, the Sun engineer who invented Java, has quietly acknowledged Java's limitations in numerics to the group, which is pushing for changes in Java so it can be used in aerospace engineering, car design and scientific modeling.
Calling itself the Java Grande Forum, the group this week will present Sun with a wish list of changes for Java, posing a critical test of the company's willingness to overhaul a core technology to meet high-tech demands.
Can't do the math
At present, Java can't handle arcane-sounding math functions such as floating point, complex numbers, two-dimensional arrays and operator overloading. The functions are accommodated in languages such as Fortran that are used by programmers for engineering applications. This means that Java, great for server-based processing instead of Common Gateway Interface scripts, doesn't find much of a role today in financial modeling, oil-reservoir simulation or physics engineering.
"Java doesn't have features to support these advanced capabilities," said George Thiruvathukal, a computer scientist affiliated with Loyola University and Argonne National Laboratory. He said in some cases Java's deficiencies can be traced to the Java Virtual Machine (JVM) or to Java's method for sharing data remotely, the Remote Method Invocation (RMI).
In spite of its shortcomings, Java has enormous appeal to engineers and researchers because as a programming framework, Java lets you develop network-aware applications that are write-once, run-anywhere.
"We have the same portability problems as everybody else," explained Ron Boisvert, research scientist at the National Institute of Standards and Technology. "Although workhorses like Cray [supercomputers] are still in use, more and more scientific processing is being done on [Reduced Instruction Set Computing]-based servers as parallel machines."
Because of this trend, high-speed engineering applications often need to be ported from one platform to another, and that's where Java could help a lot.
The high-speed computing world is eager to use Java, Boisvert said. "It's new enough that if there are problems with it, we can influence change," he said.
Added Jack Dongarra, computer science professor at the University of Tennessee: "Today, Java is not seriously being used in scientific computing, and Java Grande is an attempt by the community to pose ideas to Sun in the hopes that it eventually will."
To that end, Thiruvathukal, along with Geoffrey Fox, a computer professor at Syracuse University, earlier this year organized the Java Grande Forum to give some of the brightest minds in the computing arena the opportunity to come up with proposed changes to Java that will foster its use in advanced engineering.
At this week's meeting, expected to be attended by more than two dozen academic and research institutions and industry representatives from companies such as IBM and Intel, the Java Grande Forum will seek to nail down its Java-improvement demands based on a completed draft document.
The draft is expected to be a long list that would require an overhaul of the JVM, the addition of the IEEE floating-point standard, a new notion of lightweight objects, functions for vectors and matrices, and a new application framework based on a faster message-passing interface than RMI to satisfy parallel-computing needs.
Based on earlier discussions with Sun, which has quietly posted its own tentative ideas for change on the JavaSoft Web site, the Java Grande Forum's wish list "is more of an ambitious idea" than Sun may immediately be willing to accommodate, Thiruvathukal acknowledged.
Eager to keep the brainy Java Grande Forum in the fold, Sun has mustered a diplomatic effort led by Sia Hassanzadeh, Sun's senior business developer for high-performance computing, to negotiate inevitable change.
Finished by fall?
"They are certain to have a long list, but we want to build consensus," Hassanzadeh said. While Sun naturally hopes the group's demands can be met by simply adding some new APIs to the Java library, he acknowledged it may require more fundamental changes to the JVM - in essence, a new version of Java.
Sun and the Java Grande Forum hope to finalize a list of changes by November for public presentation at the Supercomputer '98 conference, with the wider Java community hearing about Sun's plans at the next JavaOne conference, in the spring of '99.
While the influence exerted by the Java Grande Forum may lead to the positive evolution of the technology, some in the research community are critical of the forum.
"Java has already been developed to accommodate our distributed computing requirements," said Mani Chandy, computer professor at the California Institute of Technology. He praised Sun's new technical effort, called Jini, intended to provide a new way to remotely run applications using Java. "We don't need to put all this pressure on Sun," Chandy concluded.
Back to the top
© 2000 Cable News Network. All Rights Reserved.
Terms under which this service is provided to you.
Read our privacy guidelines.