Innovative Systems Lab Projects

The Innovative Systems Lab addresses two broad areas of research and development: novel computing architectures and cluster computing.

Novel computing architectures

ISL investigates the use of next-generation acceleration systems (GPUs, Cell processors, field-programmable gate arrays, etc.) for advanced science and engineering applications. ISL staff work with multidisciplinary computational science teams on the design and implementation of scientific codes on accelerator-based systems.

• Chemical computations on future high-end computers: ISL staff have been investigating the use of FPGAs and the Cell Broadband Engine to accelerate the execution of two-electron repulsion integral calculations used in direct self-consistent field (SCF) calculations. We are implementing Rys quadrature scheme for two-electron Coulomb repulsion integrals to evaluate primitive integrals [pq|rs] for Gaussian-type orbitals (GTO) basis sets on the SRC-6/7 reconfigurable computers and IBM QS21 Cell/B.E. blade system. The project is funded by National Science Foundation grant #0626354.
• Mapping scientific applications to the Cell Broadband Engine: ISL investigates the use of the Cell Broadband Engine in scientific high-performance computing. Several traditional high-performance scientific computing applications have been implemented on the Cell/B.E. processor, including NAMD (molecular dynamics), MILC (quantum chromodynamics), and DSCF (quantum chemistry) codes.
• Advanced astrophysical algorithms to novel supercomputing hardware: ISL investigates the use of FPGAs and GPUs for the acceleration of data analysis codes used by the cosmology community. The two-point angular correlation algorithm was implemented on several reconfigurable computers, including SRC-6, SGI RC100, and Nallatech H101 as well as NVIDIA GPU G80 and G200 GPU platforms. Current work focuses on the FPGA implementation of an artificial neural network for computing photometric redshift for SDSS datasets. The project is funded by NASA grant #NNG06GH15G.
• Investigating application analysis and design methodologies for computational accelerators: ISL is studying how computational applications can be mapped to newly emerging accelerator-based architectures. ISL staff are developing formal guidelines and recipes that other researchers can adopt when porting their applications to similar accelerator-based architectures and devising methodology for analyzing and characterizing existing applications with respect to novel computing architectures. The project is funded by NSF grant #0810563.
• Acceleration of data-intensive applications on novel hardware architectures: ISL is collaborating with the research division of the National Archives and Records Administration (NARA) on using novel architectures for data ingest processing. All federal records will go digital in 2012, and NARA is preparing to accommodate that data storage. ISL is creating document analyzers to enable single-pass data processing of files being ingested into the archive, to evaluate and index documents on the fly. Analysis areas include rapid text parsing to build word indexes, document type verification, file hashing for later verification, and image analysis for search indexing.

Cluster computing

ISL pushes the boundaries in cluster computing by integrating new technologies, building novel architectures, and developing software to support highly diverse computing resources.

• Heterogeneous clusters: ISL built a prototype 32-node heterogeneous cluster, called QP, that includes multi-core chips as well as two types of application-specific accelerators: graphical processing units (GPUs) and field-programmable gate arrays (FPGAs). The system has been extensively used by several research groups that develop and run applications and system software. The project is co-sponsored by NSF, NVIDIA, Xilinx, and Nallatech.