A publication of the Advanced Simulation & Computing Division, NA-121.2, NNSA Defense Programs
September 2008
NA-ASC-500-08—Issue 8
Return to this issue’s stories
ASC Salutes Matthew L. Leininger
As a deputy for Advanced Technology Projects within the ASC Program at Lawrence Livermore National Laboratory, Matt Leininger works at the center of advances in supercomputing that are enabling predictive scientific simulations.
“It’s exciting to see the direct impact advanced supercomputing architectures and predictive models are having on both scientific discovery and real-world problems such as stockpile stewardship,” said Matt.
The new 14–20 petaFLOPS supercomputer, Sequoia, is one of the projects on which Matt works. Due for deployment in late 2011, Sequoia will be preceded by a 500-teraFLOPS early development system called Dawn, due for deployment in early 2009.
“Sequoia is a prime example of how LLNL is pushing the frontiers of supercomputing,” said Matt. “It’s a huge technology leap forward.” Sequoia will have on the order of one million processor cores and deliver a 20-time improvement over BlueGene/L for science codes while simultaneously delivering a 24-time improvement over ASC Purple for the integrated design codes. While serving as a stepping stone to petascale computing, Sequoia will also put the ASC Program on the path to reaching exascale computing within the next ten years.
This past year, Matt has also been part of the team developing a new lab collaboration where LLNL computer scientists are working with industry to field a large-scale Linux cluster testbed. This cluster, called Hyperion, is a dedicated development and testing environment for critical enabling of Linux cluster technologies and an evaluation testbed for new hardware and software technologies. The Hyperion partnership will help ensure the success of future multi-petascale platforms, such as Sequoia, and prepare commodity Linux cluster technologies for the petascale era.
The Hyperion partnership involves LLNL, Dell, Intel, Cisco, Sun, Mellanox, QLogic, RedHat Data Direct Networks, LSI, and Supermicro. The first phase of Hyperion is currently being integrated onsite at LLNL and is made up of 576 nodes. The second phase will be delivered in January and will double the size of the Hyperion cluster to 1,152 nodes. Partnership personnel will be using Hyperion to develop and scale new hardware technologies and system software such as the Lustre parallel file system, the ASC Tri-lab Operating System Software (TOSS) based on RedHat Linux, the common parallel programming models such as MPI and OpenMP, the OpenFabrics high performance networking software for InfiniBand, and low latency ethernet.
Prior to becoming a LLNL scientific staff member in 2007, Matt received a PhD in theoretical and computational chemistry from the University of Georgia in 1999, joined Sandia National Laboratories (Livermore) as a post-doctoral candidate, and later became a Sandia staff member in the High Performance Computing and Networking group. Matt led the recent ASC PathForward project, which funded the development of the OpenFabrics InfiniBand software stack now run on 25 percent of the machines on the Top 500, including the ASC tri-lab Linux capacity clusters. Matt’s areas of expertise include parallel quantum chemistry methods, application middleware, and high-performance networks and storage. His current interests include scalable system architectures, commodity-based and high-performance computing, and enabling predictive simulation.
DOE Privacy Disclaimer | Sandia Privacy Disclaimer | 2008-6301 W
Developed and maintained by Sandia National Laboratories for NA-121.2