• Number 388  |
• May 13, 2013

LLNL’s Maya Gokhale creates computational sleuthing tools

Computer scientist Maya Gokhale of DOE's Lawrence Livermore National Laboratory enjoys reading mysteries in her leisure time, which is not surprising given her aptitude for computational sleuthing, notably finding the proverbial ‘needle in the haystack,’ the key nugget of information buried in the avalanche of data today’s supercomputers produce.

“Not only is the amount of data being generated growing exponentially,” Gokhale told LLNL’s Science & Technology Review in 2012, “but when the raw data are analyzed, more data — called ‘metadata’ — are generated as well. It’s truly an issue of ‘drowning in data.’”

Recently named a Distinguished Member of Technical Staff by Lab Director Parney Albright, Gokhale leads a group developing data-intensive computing architectures and techniques for addressing the “data overload” problem. This involves the synergy of multiple disciplines including computer science, applied mathematics, and statistics.

Finding ways to better manage enormous sets of data is critical as High Performance Computing (HPC) moves toward next-generation exascale (quintillions of operations per second) computing.

"This is an exciting time for computing at the Lab as we design high performance architectures for data intensive computing and simultaneously face immense challenges to achieve exascale," Gokhale said. "I'm glad to be involved in those challenges with my colleagues at LLNL and the community."

Gokhale, who joined the Laboratory in 2007, is one of the founders of the field of reconfigurable computing and her work has influenced HPC architecture design. She also has been instrumental in developing the hardware and software to make field programmable gate array (FPGA) systems usable and accessible to research community. FPGAs allow researchers to tailor CPUs to their software application.

Her team’s work on a C-to-FPGA compiler earned a R&D 100 Award. In addition, she is co-recipient of three patents related to memory architectures for embedded processors, reconfigurable computing architectures, and cybersecurity.

“I am excited by the challenging scientific and national security problems, by the scale of computing resources needed to address these problems, and by the importance of the work,” Gokhale said in an interview posted in Women@ Energy (http://energy.gov/diversity/articles/women-energy-maya-gokhale). 

Her team has developed a data intensive architecture and algorithm that could compute a large BlueGene size graph on a smaller cluster augmented with flash memory. This innovation achieved strong rankings on the Graph500 list, which measures a computer’s data-intensive computing capabilities. The Graph 500 gets its name from graph-type problems — algorithms — that are a core part of many analytics workloads in applications, such as those for cyber security, medical informatics, and data enrichment.

A fan of Agatha Christie, she said, “I like mystery stories that are puzzles to solve."

Just as Agatha Christie’s Poirot and Marple uses their intellect and intuition to puzzle out mysteries, Gokhale applies her math and computer science skills to extract pieces of evidence from seemingly impenetrable mountains of data.

“I was drawn to mathematics and computer science because I enjoy logical thinking and problem solving to create new ways to build and use computers,” Gokhale said. “I feel satisfaction in using my mind to work on problems that have a direct impact on science and national security.”

Only in Gokhale’s case, the outcomes produce real-world benefits for the research community.

Submitted by DOE's Lawrence Livermore National Laboratory