Many of the supercomputers we run at SDSC are designed to make it easier for web developers to run portals (science gateways) so that researchers can use the power of a supercomputer from their web browser. My job is figuring out how to put together the pieces of technology, like the servers, storage, and networking, that can do these jobs well.
Like I wrote earlier, I started out doing research in physics, and before I worked on supercomputers, I was using them. My area of research was the kind of turbulence that occurs out in space where stars form. Simulating those regions of space takes a serious amount of computing power and generates a serious amount of data. My research required almost as computing power for data analysis as it took to run the simulations. One cool thing about the data from the turbulence simulations is the kinds of extremely large and engaging images we can generate. This led me to collaborating with projects developing innovative visualization capabilities and building display walls that could display the images across multiple LCD panels.
One of the projects I have enjoyed the most was building Meteor, a little supercomputer made out of 15 Raspberry Pis. Each Raspberry Pi was connected to its own display, so we could show what the Pi was doing, and all of the Pis were connected to a shared network. I worked with a team of students to create video games that run across all of the Pis at the same time so we could use all 15 displays to see the game. This is exactly the kind of programming we do on supercomputers to break up big problems into smaller pieces for each computer to work on.
I’ve also built much larger supercomputers. Building larger systems means working with more components.
