August 2014

Where are you from, originally?

I was born in Paris, France. Paris is an old and large city, known for its sightseeing among other things. As a kid, it probably looks even larger than it really is. The public transportation facilities can, however, rapidly take you from one place to another. I greatly enjoyed growing up there, and after a few years abroad, this is where I live today with my wife Isabelle and my daughter Hortense.

Can you summarize your educational background?

For three years in the late eighties, I went from Paris to Toulouse in the south west of France. There, I earned a computer science degree from the ENSEEIHT engineering school. I spent the last year working at CERFACS, a European research center, in the parallel algorithms group led by Iain Duff. In January 1993, I started as a graduate research assistant at the Computer Science Department of the University of Tennessee, Knoxville, within Jack’s group. I earned a PhD from UTK in 1996 and stayed on as a postdoctoral researcher at ICL until 2001.

How did you get introduced to ICL?

The first time I heard about ICL was at CERFACS. I was working there on a blocked implementation of the Level 3 BLAS for MIMD vector processors, AKA, “shared-memory super-computing in practice.” The Alliant FX/80 and the Convex C220 were my playground. One day Jack came for a visit. During the afternoon, Iain and Jack were having a lively discussion in an office next to mine. The first distributed-memory supercomputers, such as the Intel Hypercubes iPSC/2 and iPSC/860, were just coming out. I did not know at that time that I would have the chance to work on such systems (at ORNL) just a few years later as a UTK/ICL student.

What did you work on during your time at ICL?

During my stay at ICL as a student, I worked on my degree, but I guess this is not the purpose of your question. 😉 I first contributed to the ScaLAPACK project. I then worked on the ATLAS, HPL, and GrADS projects. For various reasons, ScaLAPACK, ATLAS, and HPL will always have a special place in mind. I am certainly very glad to see them still heavily used today by the high-performance computing community. I am also very grateful to have had the chance to contribute to them.

What are some of your favorite memories from your time at ICL?

I certainly have kept great memories of my time at ICL. They are all definitely related to the great and talented people I had a chance to meet while I was there. I certainly remember the conviviality of the Friday lunches. I also remember quite well the trips we took to the various Supercomputing editions and other conferences that we attended. I enjoyed visiting other states and locations all over the country. Even if these trips were quite demanding in terms of preparation, they have always been a source of fun and discovery.

Tell us where you are and what you’re doing now.

I eventually moved back to France in 2001. I worked for Sun Microsystems (Oracle nowadays) for a few years as a benchmark engineer. I then took a job at ESI-Group within the product engineering department. ESI-Group produces manufacturing software tools that help with designing virtual prototypes, i.e., better products, such as safer and lighter cars. The finite element method is often used and thus one needs to solve large linear sparse systems and generalized symmetric eigenvalue problems. Distributed-memory computers are commonly used by our industrial customers, so my job also has a large “high-performance computing” aspect. Nowadays, getting industrial applications—like the ones ESI-Group produces—to efficiently use GPUs or Intel Xeon Phi co-processors is certainly a significant part of my daily job.

In what ways did working at ICL prepare you for what you do now, if at all?

Working at ICL prepared me very well for what I am doing now. In fact, I do use the software I worked on when I was at ICL on a daily basis. Knowing parts of those packages inside and out is a great advantage when building something on top of them. When working for the software industry, having released some freely available software in source form is certainly a highly valuable and useful experience.

Tell us something about yourself that might surprise people.

A couple of months before officially joining UTK/ICL, I came for a “private” visit to Knoxville. It was important for me to see with my own eyes where I had imagined spending the next few years of my life. I certainly remember meeting Jack and the “linear algebra” group in Ayres Hall. Clint Whaley took some time to show me around as well as some useful practical details of UT’s student life. At that point, I was convinced that I would enjoy living there and the years that have followed certainly confirmed this impression.