"I've always been curious about things around me," said Arimilli, who as a teen used to tinker with the family's Firebird and Chevy Impala, along with the familiar electronics kits from Radio Shack and Heath.

That early tendency to "open things up and change them around" led to a career that over the past decade has contributed more than 191 innovations to IBM. During this time, IBM has ranked first on the U.S. Patent and Trademark Office's list of top 10 U.S. patent recipients, with 22,357 patents, including 3,288 this past year. The company's patent portfolio has helped generate an average of more than $1 billion in intellectual-property royalties during each of the past 10 years.

Arimilli is IBM's leading expert on symmetric multiprocessing (SMP) system structures, cache/memory hierarchies and system bus protocols. His career has included work on the Blue Pacific supercomputer for Lawrence Livermore National Laboratory, and he was chief architect and lead engineer of IBM's Power4 GigaProcessor Storage subsystem for its 32-way e-server p690, code-named Regatta.

When asked to explain what motivates him, Arimilli said moving to the United States with his family from Andhra Pradesh, India, in 1969, when he was six, was "a culture shock" that made him grow up fast and "learn to be a survivor." While that experience provided motivation, watching the U.S. Apollo 11 moon landing on television — for Arimilli, a powerful introduction to the medium of TV — provided inspiration.

Arimilli's family settled in Baton Rouge, La., where his father worked as a geologist. Arimilli studied electrical engineering at Louisiana State University, graduating with a BS in 1985.

That same year, he joined IBM to work on the team that was developing the first RISC microprocessor for the IBM RT/PC, a system that ran IBM's version of Unix, AIX 1.0.

Most of the patents Arimilli received in 2002 centered on innovations for the Regatta, which is based on 16 "servers on chip," each integrating dual microprocessor cores, cache, I/O, switches and memory.

According to Arimilli, the Regatta system represented a departure in server design from IBM's and the industry's traditional methods and was an attempt to help IBM gain market share in Unix servers. While at first the plan was to build a run-of-the-mill 128-way SMP server, the strategy shifted after senior management issued a directive to Arimilli and his group to build a system that would propel IBM past its competitors.

The group held brainstorming sessions with the "best of the best from IBM" for three to four months and came up with the early concepts for the Power4 Regatta, Arimilli said.

Arimilli's main contribution to the Regatta was the full subsystem. He chose that area because he believed microprocessor microarchitecture was hitting a wall and soon would no longer be able to achieve 2x to 4x performance increases with each new design. His idea was that "by starting from the system level [and working] backward," the system could be designed to achieve greater performance leaps.

IBM has said that Arimilli's innovative concepts have been central to the industry-leading performance of its RS/6000 and eServer pSeries Unix systems in commercial, technical and Web server applications. His system structures and cache microarchitectures delivered such servers as the IBM RS/6000 Models J30/R30 and F50/H50, "both of which set new performance records at time of announcement," the company said in a press release in 2001.

Those are among the many reasons that IBM named Arimilli an IBM fellow, the company's highest technical honor, in May 2001.

Thick with innovators

Invention and innovation are never far from Arimilli's thoughts. He often consults with seminal thinkers in technology, such as Linus Torvalds, author of the Linux kernel, and Tim Berners-Lee, inventor of the World Wide Web. He says his job is to understand where such visionaries see technology moving and then go back to IBM to innovate and build those ideas into products. It's a job that he said he chose over working in one specific technical field because it leverages his ability to see "the broad picture."

As he looks toward the future, Arimilli believes the medical field will grow increasingly reliant on technology and foresees astrophysicists' using "petaflops boxes that IBM will build to finally solve many scientific problems."

Such milestones could have profound impact not only on the scientific world but also "on every man and woman anywhere, between 2008 and 2020," Arimilli asserted.

As for his role in that future, Arimilli said he likes "to engage people who are involved in specific technologies and broadly take the content, collect it and formulate solutions that [IBM] may not even build but may give to a Sony Corp. or someone else."

Arimilli is confident that IBM will remain at the forefront of technology because it can't afford not to. "We can't back off. That's why IBM has allowed the risky investment in innovation and development of technology."

When Arimilli, who is married, isn't working, he enjoys playing tennis, wine tasting, cooking, traveling and telling stories to his three children.