University Collaboration


Cornell University
       Completed   Keshav Pingali
   Calin Cascaval
   
Continuous Program Optimization
  ·  Project Page

Dartmouth College
       Completed   Thomas H. Cormen
   Calin Cascaval
   
Out-of-core computation
  ·  Project Page

Northwestern University
       Ongoing   Prof. Alok Choudhary
   Hao Yu
Gheorghe Almasi
Manish Gupta
   
Runtime System Optimizations for Highly Scalable Parallel I/O Prof. Alok Choudhary's group has been involved in developing scalable I/O techniques for large-scale systems. They have developed techniques that capture and use access pattern information at runtime to reorganize access patterns which are highly optimized for performing I/O; perform runtime level caching to reduce number of I/O accesses; and reduce synchronization and locking requirements (which are used by conservative techniques to account for worst case) to minimize overheads that hinder performance. We have established a collaboration with this group to do further research on scalable parallel I/O, and apply these results to Blue Gene/L.
  ·  Blue Gene

SUNY Stony Brook
       Ongoing   Brian Colle
Arie Kaufman
   Lloyd Treinish
   
Weather modelling, dispersion modelling, visualization, web-based interfaces
  ·  Deep Thunder

Towson University
       Ongoing   Michael Felton
   Lloyd Treinish
   
Weather modelling applications, GIS, user interfaces, visualization, web-based interfaces
  ·  Deep Thunder

Universitat Politecnica de Catalunya (Barcelona)
       Ongoing   Jesus Labarta
   Jose Moreira
Manish Gupta
   
Optimization of MPI and development of performance tools for the Blue Gene/L supercomputer
  ·  Project Page

University of Edinburgh
       Ongoing    Lorna Smith
   John Gunnels
   
High-performance mathematical kernels. Specifically, BLAS, LAPACK, and ScaLAPACK-related Research and Development for Blue Gene/L.

University of Illinois at Urbana-Champaign
       Completed   Josep Torrellas
David Padua
   Calin Cascaval
   
Programming languages, Continuous Program Optimization, Thread level speculation
  ·  Project Page

University of Maryland
       Ongoing   Jeff Hollingsworth
   Simone Sbaraglia
   
Performance Measurement and Modeling of Deep Hierarchy Systems