Ravi Arimilli (IBM's prolific inventor)-Working on computer history
Shivram:
Another Desi rising techno-wiz.
================================================
(IBM inventor Ravi Arimilli more excited about company's big project than his personal awards)
--------------------------------------------------------------------------------
Ravi Arimilli
--------------------------------------------------------------------------------
Age: 35
Company: IBM Corp.
Job: Chief scientist for one of two design teams involved with IBM's Giga Processor project. Arimilli's team is in charge of the memory subsystem.
Family: His wife, Padma Sunkara, is a high-tech job recruiter. They have a 1-year-old daughter.
Education: Bachelor's degree in electrical engineering from Louisiana State University.
Background: Born in India. His family moved to Baton Rouge, La., when he was 6.
Accomplishments: Has won 18 patents with more than 100 patent applications. His team invented Smart Cache, a way for memory to anticipate the next information the processor will want.
Vivid childhood memory: The Apollo 11 moon landing in 1969 was his first experience watching television. Those memories aroused his curiosity in new technologies.
High school hobbies: Worked on cars, including a 1968 Opel Kadette and a '72 Chevrolet Impala.
What he learned from his parents: Strong work ethics, values and competitiveness.
==========================================
By Kirk Ladendorf
American-Statesman Staff
Posted: Jan. 18, 1999
Ravi Arimilli could have a great trophy wall in his office at IBM Corp.'s Austin campus if he wanted one.
Arimilli, 35, is among the most prolific inventors at IBM. He has been awarded 18 patents and has more than 100 patent applications in the works. His other honors include election to the IBM Academy, which is made up of the company's brightest technical minds.
Those things don't excite him. He says there's no need for what he calls an "I love me" wall full of awards in his office. He keeps his patent awards at home in a box.
What does excite him is working with other smart engineers on technology that has a chance to make computer history.
And that's what he is doing as chief scientist for one of two IBM engineering teams in Austin that are pushing hard on an ambitious development project they call Giga Processor, or GP.
Vijay Lund, IBM's vice president for server technology and head of the project, says GP is the company's big bet for next-generation, high-performance computing.
"Our entire future is at stake here," Lund says. "We undertook a high-risk mission two years ago. We have no backup plans. We are going all out."
The processor, when it becomes reality, will operate at speeds of 1 gigahertz (1,000 megahertz) and faster.
Arimilli's team of 100 engineers is working on chips that surround the processor and will enable it to talk to its memory very efficiently. Another similarly sized team, headed by Jim Kahle, is working on the processor core.
The two teams, both at work in a squat building nicknamed the Dungeon on the IBM campus, interact constantly.
High stakes
The project is big stuff. IBM will invest hundreds of millions of dollars just on the design of the new processor, one analyst estimated. The company is building a chip factory in East Fishkill, N.Y., to manufacture it.
The stakes for IBM are enormous. GP, if it is successful, will be an engine for a full range of high-performance computers at IBM: desktop workstations, RS/6000 network servers, AS400 mid-range computers and even its advanced parallel processing supercomputers that can be made up of dozens or hundreds of high-performance processors working together.
"It's a big deal," said analyst Brian Richardson with the META Group in Stamford, Conn., of the Giga Processor. "They have a lot riding on this."
That's a big part of why Arimilli likes working for IBM.
"You can exploit the strength of a big company," he says. "You have good scientists and engineers at your fingertips, and you can use those people to catapult your thinking forward."
He has friends who have struck it rich with start-up companies in Silicon Valley, but he's not tempted by the entrepreneurial life. "I am more of a technical spirit," he says.
Arimilli talks about engineering challenges in an articulate, philosophic fashion that his colleagues say is a key to his leadership skills.
"With these engineers, the ideas are just coming out like you wouldn't believe -- one after another, one after another," he said. "The question is: Which of these ideas really makes sense and is profitable for the company?"
A spark
Lund said he decided Arimilli was the right person for the job in the first few minutes of their initial conversation two years ago.
"There was a spark that he had in that discussion," Lund said. "It's the spark of: 'Let's go build the best. Let's go do something that's never been done before.' "
Arimilli says he isn't the only engineer who is excited by the project.
"I come into work and these guys are jump pumped, and they pump me up," he said. "They are thinking and really questioning things. When you have a team like that, that is really charged, not only will they execute something big, but they will come up with some pretty ingenious ways of solving problems."
Fifteen years into his IBM career, Arimilli has come a long way from being the kid from India who was too short and picked on by his elementary school classmates in Baton Rouge, La.
Arimilli was 6 when his family moved from India after his father, a geologist, won a fellowship at Louisiana State University. He remembers staring at a flight attendant on the plane ride over. She was the first blonde he had seen.
In his late teens, Arimilli grew to be a trim 6-footer, but he says those early school years of teasing toughened him and steeled his competitiveness.
He concentrated on tennis as a youngster, practicing and playing with a ferocity that eventually won him a spot on the LSU tennis team.
When he decided that tennis could take him only so far, he turned his energy to electrical engineering, graduating from LSU at 20.
He had six job offers from which to choose at IBM. He chose one at Austin because it put him right in the middle of the development of IBM's first computer to be based on a streamlined design called RISC, for reduced instruction set computing. That project turned out to be a commercial dud, but a follow-up project turned into IBM's well-received RS/6000 product family that was introduced in 1990.
While processors are the brains of computers, their performance is sometimes bottled up because they can't communicate fast enough with support chips such as memory chips.
Much of Arimilli's work has been in devising ways to speed communication between the processor and other chips.
Arimilli designed improved data paths -- called system buses -- that enabled a series of fast processors to move instructions to the rest of the computer.
More recently, he has worked on creating fast, new ways for computers with more than one processor to interact rapidly with memory.
He has led teams that have developed high-speed, innovative ways for the processor to interact with its fastest memory, called cache memory. IBM calls the innovation smart cache because the memory actually anticipates what information the processor will need next.
Arimilli's team must find clever and efficient ways to keep new instructions and information moving constantly to the processor.
"That is the punch line," he said. "Just keep the processor cruising."
'Pushing the envelope'
Jim Rymarczyk, an IBM Fellow who has worked with Arimilli, says he is one of IBM's leading minds in devising ways for processors to interact rapidly with memory.
"He is really exceptional in terms of pushing the envelope, taking risks, but with good judgment," Rymarczyk said. "He is willing to go out on a limb relying on his confidence in his ability to pull it off. He commits to deliver products, and he comes through on aggressive designs. He's also a natural leader. His growth potential is tremendous, far beyond his current role."
Arimilli's team works hard -- and parties regularly. The engineers gather about once a month to celebrate, often at local brew pubs.
Blowing off steam, he said, is an important way to avoid burnout.
Arimilli said his style is to give his team leaders freedom, but to challenge their thinking, to explore options and to keep sight of how engineering details can affect overall system performance.
"My strength is that I have a tendency to make people think," he said. "People who are around me are always thinking and have a passion to do what they do.
"Coming in Monday morning and working is not laborious. I'm working with people who are just energized, stimulated and enthused, day after day. And, to top it off, if you put out something that makes history, that's what I love to do."
austin360.com |
