Former Research Projects

Sequoia

Sequoia is a programming language designed to facilitate the development of memory hierarchy aware parallel programs that remain portable across modern machines with different memory hierarchy configurations. Sequoia abstractly exposes hierarchical memory in the programming model and provides language mechanisms to describe communication vertically through the machine and to localize computation to particular memory locations within the machine. We have implemented a complete programming system, including a compiler and runtime systems for Cell processor-based blade systems and distributed memory clusters, and have demonstrated efficient performance running Sequoia programs on both of these platforms.

Folding@Home - GPGPU

I've been lucky enough to help the Folding@Home folks get the first GPU based client out. This has been a long time coming and I'm glad they had the patience to keep at it and keep working with Brook and GPGPU programming. I especially want to thank Vishal for the hard work adapting algorithms to get the code running well. I also want to thank ATI for all the support getting things up and running. The current estimate is that a R580 is 20-40X a single core 2.8GHz Pentium4. At these rates, we can now do in close to a month what used to take years of simulation. See current stats here.

BrookGPU

BrookGPU is a compiler and runtime implementation of the Brook stream programming language which provides an easy, C-like programming environment for today's GPU. As the programmability and performance of modern GPUs continues to increase, many researchers are looking to graphics hardware to solve problems previously performed on general purpose CPUs. In many cases, performing general purpose computation on graphics hardware can provide a significant advantage over implementations on traditional CPUs.

GPUBench

GPUBench is a benchmark suite designed to analyze the performance of programmable graphics processors in areas of particular importance to general purpose computation. More so than being a measuring stick for GPU performance, GPUBench is intended to be a tool for developers of GPU-accelerated applications, since a more complete understanding of machine capabilities will dictate better software design decisions. Information included in GPUBench reports is useful in accurately estimating the performance of shader programs as well as determining if a particular GPU is a viable target platform for a particular computation.

Chromium

Chromium is a system for interactive rendering on clusters of workstations. It is a completely extensible architecture, so that parallel rendering algorithms can be implemented on clusters with ease. Various parallel rendering techniques such as sort-first and sort-last may be implemented with Chromium. Furthermore, Chromium allows filtering and manipulation of OpenGL command streams for non-invasive rendering algorithms. Chromium runs on Microsoft Windows and several types of Unix such as Linux and IRIX.

Graphics Clusters

I've spent a great deal of time researching, developing, and building graphics clusters, including helping design several research and production clusters. I have also given many talks on building graphics clusters.

Raptor

Raptor is a parallel volume renderer written for use with Chromium. Raptor provides an example of a sort-last parallel application utilizing Chromium's distributed event model, CRUT, and compositing systems, such as the binaryswap SPU.

NPACI Vistools

While at SDSC, I worked on the NPACI VisTools. NPACI's VisTools are a collection of software libraries and applications that allow users to create high resolution visualizations of large complex scientific data. Current applications let users interactively navigate large volumes on their desktop and do full color perspective volume rendering. The toolkits are designed to work with data sets that are too large to fit in memory using out-of-core techniques.