Application Performance Tools

Activities
(Our team members provide deliverables for the following efforts.)

• ADIOS -- PI Scott Klasky
  (add ADIOS description from Brad here)
  
  
• Cheetah – Cheetah is a framework implemented with the Open MPI code base supporting hierarchical collective communications.  This is a high-performance extensible library that supports arbitrary communication topologies aimed at extreme-scale computer systems.
  
  
• Colony -- PI Terry Jones
The HPC-Colony project is a joint research effort with Oak Ridge National Laboratory, the IBM T.J. Watson Research Center and Haifa Research Center, and the University of Illinois at Urbana-Champaign to create scalable Services and Interfaces that permit easy application porting for high-performance computing (HPC) systems with very large numbers of processors. Funding for the HPC-Colony Project is provided by a grant from the U.S. Department of Energy Office of Science.


• Energy Efficiency for Extreme Scale Systems -- PI Stephen Poole
(add description from Stephen here) * Hercules: The goal of this project is to help scientists improve the performance of applications on new platforms by building a translation tool based on program patterns to specify source-to-source transformations or to control the compilation process of a backend compiler. Hercules provides a rich set of user friendly directives for C/C++/Fortran to specify patterns and to specify transformations. The typical workflow entails the construction of a pattern, aimed at identifying sources exhibiting desired, user-specific, characteristics as well as the specification of transforming actions against the matches sources. Patterns may range from syntax (e.g. intra-procedural loop nests) to performance (e.g. instruction throughput). The current transformations include: loop transformations, inter-procedural data transformations and parallelization support. We envision Hercules working with several production open-source compilers, the primary focus being the Open64 compiler.


• Institute for Advanced Architectures and Algorithms (IAA) -- IAA Co-Director Jeffrey Nichols
In the next few years, tremendous increases in supercomputer capability will revolutionize the way science is done, and predictive computer simulations will play a critical role in national security, energy, scientific discovery, and national competitiveness. To meet these challenges, Sandia and Oak Ridge have established the Institute for Advanced Architectures and Algorithms (IAA). Sandia and Oak Ridge will build upon their long history of collaborating together, strong ties to universities, and successful industry collaborations (e.g., with Intel, Cray, AMD, Micron, SUN and IBM).


• Scalable Storage Systems -- PI Stephen Poole
Our goal is to enhance the scalability of existing parallel file systems, including the National Center for Computational Science's center-wide file system Spider. As the world's largest Lustre file system, spider has exhibited several scalability challenges that were absent at smaller scales. In particular, we are developing adaptive file system interfaces that allow users to easily access the highest available performance modes of the file system without modifying user applications.


• Minimizing System Noise Effects For Extreme Scale Scientific Simulation through Function Delegation -- PI Richard Graham
The scalability of collective communication algorithms limits the scalability of many parallel scientific simulations.  Our goal is to advance the use of asynchronous communication support for collective communications, to minimize the effects of system noise on application scalability and to provide opportunities for computation and communication overlap.  As part of this project we are working with Mellanox Technologies to provide network offload communication library requirements for asynchronous collective communication support.


• Scalable Tools Communications Infrastructure -- PI Richard Graham
The Scalable Tools Communication Infrastructure (STCI) is a multi-platform standalone library supporting the full process life-cycle management.  The focus of this project is on portability (both system-level and application-level), scalability, and fault-tolerance.


• Open MPI -- ORNL Contact Richard Graham
The Open MPI Project is an open source MPI implementation that is developed and maintained by a consortium of academic, research, and industry partners. Open MPI is therefore able to combine the expertise, technologies, and resources from all across the High Performance Computing community in order to build the best MPI library available. Open MPI offers advantages for system and software vendors, application developers and computer science researchers.