WRITTEN BY: Audrey Woods
In a 2014 commencement address to students in Ahmedabad India, Professor Arvind emphasized how important it is to “enjoy what you are learning or doing. One cannot just work for the pot of gold at the end of the rainbow – one has to enjoy the journey itself.” By his jovial laugh and passion for computer technology, it’s easy to see that Prof. Arvind has lived out his own advice. He speaks about programming languages, microprocessors, and computer architecture the way some people talk about sports, with the excitement of imagined potential.
Prof. Arvind’s interest in machines dates back to 1969 when he earned a degree in electrical engineering from the Indian Institute of Technology in Kanpur. He soon went on to earn a Master of Science and then a PhD in Computer Science from the University of Minnesota. Recalling his first summer job as a graduate student working in the marketing department of Pillsbury, Prof. Arvind describes learning the real-world applications of computer technology. “Every extra penny we could charge without losing the sales meant several million dollars of profit per month.” At first disillusioned by his hard work being “put to use in selling cookies and cakes,” he soon understood that “ultimately people determine what they want and if they want to buy cakes then, by God, we are going to produce the best cakes.”
The business lessons from his graduate school days were no doubt pivotal in Prof. Arvind’s various industry ventures. In 1992, after 14 years on the MIT faculty, Prof. Arvind and his Lab for Computer Science (as CSAIL was known then) group had a joint project with Motorola funded by DARPA to create the Monsoon dataflow machine, a parallel, general-purpose computer. It was used in the Los Alamos National Laboratory and other research institutions before being retired to the Computer History Museum. Later, Prof. Arvind took a two-year leave of absence in 2000 to start Sandburst, a fabless semiconductor company that was acquired by Broadcom in 2006. Not done with industry yet, he started Bluespec Inc. in 2003 to design semiconductor tools. While Prof. Arvind has since returned to being a full-time faculty member of MIT’s CSAIL, he still serves on the Bluespec board.
Speaking about his business ventures, Prof. Arvind says he “learned that timely work in industry is far more important than creative work without time restraints. It is important to be able to explain how a product may help a customer without explaining how the product actually works. As an academic I took pride in explaining things clearly but suddenly discovered that no one outside the company wanted to know how cool the inside of our product was! The customer was merely interested in how the product may improve his/her life. All this understanding has had profound impact on my research presentations ever since.”
Now, Prof. Arvind is passionate about chip verification systems, saying it is his “dream that when you buy a microprocessor or another piece of hardware it’ll come with a proof that it actually works.” In a joint project with Professor Adam Chlipala, Prof. Arvind is working to verify the functional correctness of processors. He gives the example of, “if you say that I have a RISC-V processor, I would really like to give you a certificate with a formal proof that indeed this is a RISC-V processor.” Describing the difficulty of building a trustworthy stack of computing hardware, Prof. Arvind imagines that “these things will be of extraordinary use.”
Another of his related interests is the potential of flash storage. To Prof. Arvind, the progress in flash technology has been “dramatic,” leading to a thriving new field of research pushing the boundaries of the technology. He says, “what people do not know is that even a tiny pen drive is a full-blown system. It has a four-core processor and several gigabytes of DRAM inside… there is a whole lot of stuff going on there, and that’s just innovation on top of innovation.”
The field of flash storage is exciting to him for many reasons, one of which is personalized medicine. Because of the “humongous” amount of data in the genome, getting to a point where a person can not only look at their DNA but also run calculations over it to determine cancer risks, potential mutations, etc. is going to require enormous amounts of processing power, and that comes down in part to hardware requirements. Prof. Arvind imagines a future where we could run these calculations both quickly and cheaply, in a doctor’s office or maybe even at home.
In short, Prof. Arvind hopes to “bring down the cost of computing,” imagining a future where things previously run on supercomputers can be done “on very small machines, provided they have enough flash storage.”
Perhaps due to his industry background, Prof. Arvind recognizes that there will be some compromises in his dreams of bringing vast processing power to the market cheaply. One such trade-off is security. He admits that “very secure computing is important” for things like banks and governments and that secure computation will “remain expensive.” There will always be situations requiring costly security measures, but not all circumstances need be so restricted.
Prof. Arvind compares the scenarios to “the way we use water. We use bottled water for drinking, but we can use tap water for watering the lawn.” In his mind, there’s no reason we can’t have equivalent options in computer technology.
With plenty of experience in both hardware and software, Prof. Arvind, a valued member of CSAIL, is currently the Johnson Professor of Computer Science and Engineering and the Head of Computer Science Faculty in EECS. Having delivered more than a hundred keynote and distinguished lectures, his impact has reverberated across the field of Computer Science and beyond. He’s served on various editorial boards, chaired many conference committees, been a visiting professor in Tokyo and Seoul, managed major research collaborations, published a book, and earned a long list of accolades and awards. One might say he reached the pot of gold, but even now, Prof. Arvind’s still in it for the joy.
“It’s my dream to sell chips with proofs,” he emphasizes with his trademark laugh. “That’ll make a big difference [and] be the ultimate trustworthy computing.”