Alumni Profiles

Wisniewski hiking

Bob Wisniewski PhD '96

Interview from 2013 Multicast Newsletter

Bob Wisniewski PhD ’96 is an ACM Distinguished Scientist and the Chief Software Architect for Extreme Scale Computing and a Senior Principal Engineer at Intel Corporation. He has published over 60 papers in the area of high-performance computing, computer systems, and system performance, and has filed over 50 patents. Before coming to Intel, he was the chief software architect for Blue Gene Research and manager of the Blue Gene and Exascale Research Software Team at the IBM T.J. Watson Research Facility, where he was an IBM Master Inventor and lead the software effort on Blue Gene/Q, which was the fastest machine in the world on the June 2012 Top 500 list, and occupied 4 of the top 10 positions. Prior to working on Blue Gene, he worked on the K42 Scalable Operating System project targeted at scalable next generation servers and the DARPA HPCS project on Continuous Program Optimization that utilizes integrated performance data to automatically improve application and system performance. Before joining IBM Research, and after receiving a Ph.D. in Computer Science from the University of Rochester, Bob worked at Silicon Graphics on high-end parallel OS development, parallel real-time systems, and real-time performance monitoring. His research interests lie in experimental scalable systems with the goal of achieving high performance by codesign between hardware and software. He is interested in how to structure and design systems to perform well on parallel machines, and how those systems can be designed to allow user customization. Bob was gracious enough to answer our editor’s questions below:

What inspired you to go into this field?

RW: Growing up, I was always fascinated by the internals of calculators. I would take them apart and examine the green circuit boards and wonder how the wires and chips could produce the computation results. As personal computers developed, this interest grew into wanting to understand them, and from there, wanting to be able to understand and program supercomputers.

What is the most interesting part of your research?

RW: There are a couple axes worth defining to best describe the areas I like best. The first is how pure versus connected to usage the research is, and the second is at what level the work occurs, spanning vision, strategy, algorithms, and implementation. I believe there are interesting challenges that cross the dimensions on both axes. What have personally, and it has been a bit of a migration over time, come to find most reward on the first access, is research is targeted at solving particular users’ challenges. In computing this is a roughly five-year time horizon. For example, on Blue Gene/Q, five to seven years before it was released, we were meeting with users, understanding the problems they were trying to solve, and then trying to understand how the coming new hardware architecture as driven by technologies were going to change their model. To me, it was rewarding to spend a couple years with the hardware and software teams brainstorming about how to address the challenges, put forth some strategy for directions that may pan out. And then as the architecture starting solidifying, to translate the vision and strategy into algorithms and implementations that solved those original problems we identified. I most enjoy the middle stages of the project involving the strategy about how to tackle the problems (architecture), and then high-level algorithmic design to realize that architecture. I also am appreciating the opportunity as a system’s software researcher to impact the hardware architecture, and feel fortunate that I had the Blue Gene experience where we did that, and am currently involved in such a role

You’ve been working on optimizing software and hardware throughout your career. Now that there is such an emergence of big data/data science, how will the work that you’ve done with large scale systems help with improving data science moving forward?

RW: We should separate two things. In HPC, we have been doing big data for a while. Petabytes of information is nothing that frets an HPC user, though the size of the HPC data does continue to grow, and without the advent of new hybrid memory we would have reached a tipping point. However, the big change is the type of data and therefore what we are trying to do with it. In particular, the amount of unstructured data has massively increased demanding AI-like techniques to help filter and analyze that data, e.g., Watson. Just like big cloud computing and HPC supercomputing have not converged (yet?), I think it remains an open issue how much convergence there will be between unstructured data analytics and classical HPC computing. To be sure though, many of the techniques we have worked on in HPC for large data will be helpful, including the parallel I/O, distributed reduction, and synchronization.

What are your thoughts on careers in industrial research?

RW: Industrial research has changed dramatically in the last two decades. The days of Xerox PARC, AT&T Bell Labs, or IBM Watson providing researchers with funds unattached to their customers or product divisions have past. Instead, what has replaced that model is a keen interest in ensuring researchers are positively impacting customers’ needs. I believe that research in industry is exciting because through that customer focus, we work on tackling problems that will drive technology and society forward. As an aside, I continue to believe in long-term or pure research for science, but I understand the driving forces of corporations. Thus, I remain supportive of foundations like the NSF or NIH in driving academic research. Some of my most enjoyable interactions are meeting with customers to learn what challenges they are facing, and then coming back to the team to brainstorm about what initiatives we could undertake to overcome them. Some are shorter term, say two years, but others are more long term, i.e., more than five years out.

Why the recent move from IBM to Intel?

RW: It was the opportunity to create and drive a software HPC (High Performance Computing) systems effort at Intel. Intel is focusing on the increasingly important area of HPC with a systems-design methodology including system software input, which is my area of expertise. Intel has not historically had a system focus, so while this will be a challenging task, it presents a great opportunity as Intel processors are the dominant ones on the Top 500 list of fastest computers. In fact, on the latest list, which was released June 2013, a Xeon Phi machine was number 1. I am in the Path Finding group, which has a research focus, so I will be able to continue to focus on my research initiatives.

How do you balance work and family?

RW: Since high bandwidth has become widely available, it is easy to work effectively at home. There has also been an increased recognition of the importance of work-family balance. Together these forces allow me to spend time with my family at times when they are available, and work off hours. Also, companies nowadays tend to be international, meaning that meetings occur at all hours and that also drives an understanding that people need to be flexible with their work hours. In short, I still work many hours, but at times when I choose.

You always enjoyed competitive sports and games while at Rochester. How do you satisfy that these days with such a busy work life?

RW: Ah yes, those care-free graduate days. I do continue to play tennis, swim, and ski, but I have to say I have not been in my hockey gear since graduate school. Teach games to your kids, though the nature changes, it can still be competitive. In Boggle, I only get credit for five letter words and up, or four-letter words my daughter doesn’t know the meaning of. Scrabble easily lends itself to a multiplicative factor to equalize abilities. I’ve learned roller blade wheels are a good match for a seven-year old’s strength and bike tire size.

If you could design an app to make your life easier or better, what would it be?

RW: I’m going to take this question a bit broader. Have you ever listened to our parent’s generation saying how good we have it, what with the internet and cell phones J. I wonder to what technology we will say that of our children when they are 30-40. Without a crystal ball it is of course difficult to say, but I think self-driving, (Jetsons style), energy-efficient cars may be one. And while I’m wishing, what about a good green source of energy, did anyone say fusion, it’s only fifteen years away, just like it’s always been.

What are you up to now?

RW: With my move to Intel I am in the process of building a world-class HPC system’s software team and am looking forward to working on the challenges HPC is presenting researchers. The directions I am looking at is first, continuing to understand where technology is driving future hardware architecture and working with the HPC hardware architects to provide capabilities that will best allow software to utilize future machines. Second, I am working on establishing a cohesive system software strategy to move HPC software into the exascale era. Third, our group works on interesting and longer-term government research grants. Fourth, I am spinning up a research effort that I will be personally involved in looking at operating system kernels for future HPC machines. And finally, figuring how we can leverage all the above work to provide the next best class of supercomputers.

I hope everyone who is or was at University of Rochester is doing well, and would be happy to hear from you at