Rajeev Thakur

Introduction

My curriculum vitae: pdf

Education

• Ph.D., Computer Engineering, Syracuse University, 1995
• M.S., Computer Engineering, Syracuse University, 1992
• B.E., Computer Engineering, University of Bombay, India, 1990

Book

• Using MPI-2: Advanced Features of the Message-Passing Interface, by William Gropp, Ewing Lusk, and Rajeev Thakur, MIT Press, 1999.

  A Japanese translation of the book is available here. The translation was done by Takao Hatazaki. Here is a link to the cover of the Japanese version.

Publications

Info about Parallel I/O Benchmarks

Current Projects

• MPICH2

  I am a member of the MPICH group at Argonne. MPICH is a freely available, portable implementation of the MPI (Message Passing Interface) standard. Our current efforts are focussed on the design and implementation of MPICH2, the next-generation MPICH. The goal of MPICH2 is to develop a full implementation of MPI-2 that runs everywhere and also delivers high performance. "Everywhere" means on all machine architectures (shared memory, distributed memory, clusters of SMPs), all operating systems, and all communication technologies (TCP, shared memory, Myrinet, InfiniBand). MPI-2 includes all of MPI-1 message passing plus one-sided communication, dynamic process management, parallel I/O, and some other features. My efforts focus on implementing efficient collective communication algorithms, one-sided communication operations, dynamic process management, and thread safety. This project won a 2005 R&D 100 Award (photo).

• Formal Verification of MPI Programs

  Bill Gropp and I are collaborating with Prof. Ganesh Gopalakrishnan and Prof. Mike Kirby from the University of Utah on formal verification of MPI programs. Our initial efforts focussed on MPI one-sided communication. Our model checker revealed a bug in an algorithm for byte-range locking. See this paper for details.

• MPI-IO

  I participated actively in the definition of the parallel I/O interface in MPI-2, commonly known as MPI-IO. I also developed a freely available, portable implementation of MPI-IO, called ROMIO. Several computer-system vendors (HP, SGI, NEC, Cray, and IBM for BG/L) have used ROMIO as the basis for their own MPI-IO implementations. Our current work in this area focuses on parallel I/O for Linux clusters by using the PVFS parallel file system (with Rob Ross and others).

• SciDAC

  I am involved in the SciDAC project on Scientific Data Management, where among other things we are developing a parallel version of the NetCDF I/O library (in collaboration with Prof. Alok Choudhary's group at Northwestern University).

• UPC-IO

  Along Prof. Tarek El-Ghazawi's group at George Washington University and Dan Bonachea from UC Berkeley, I am involved in the definition and implementation of a parallel I/O interface for UPC, called UPC-IO.

Past Projects

• High-Performance I/O in Java

  Dan Bonachea (UC Berkeley), Phil Dickens (Illinois Institute of Technology), and I evaluated the I/O capabilities of Java for high-performance computing. We found that the main deficiency in Java was the lack of direct support for bulk (array) I/O for data types other than bytes, which affected performance in real applications. We proposed extensions to the Java I/O library that address these deficiencies. Performance results show that the proposed methods improve performance by a factor of 4 to 5. See these publications for details.

• Scientific Data Management

  This project aimed to simplify the problem of management, storage, efficient access, and analysis of the large data sets generated by scientific experiments and simulations. This work was in collaboration with Alok Choudhary (Northwestern University) and Jaechun No (formerly Argonne, now at Sejong University in Korea). We developed a software system for scientific data management that combines the good features of parallel file I/O (namely, high performance) and databases (namely, convenience and flexibility). This work is reported in the following publications.

Professional Activities

• Associate Editor, IEEE Transactions on Parallel and Distributed Systems, (2004-2007)
• Program Committee, ICPP 2010
• Program Committee, P2S2 2010
• Program Committee, Euro MPI 2010
• Program Committee, NAS 2010
• Program Committee, ICS 2010
• Program Committee, SNAPI 2010
• Co-Chair, Student Research Symposium, HiPC 2009
• Program Committee, SC09
• Tutorials Co-Chair, SC09
• Program Committee, PGAS 2009
• Program Committee, Cluster 2009
• Program Committee, Euro PVM/MPI 2009
• Program Committee, ICPP 2009
• Program Committee, P2S2 2009
• Program Committee, PADTAD 2009
• Co-Chair, Student Research Symposium, HiPC 2008
• Program Committee, SC08
• Program Committee, ICPADS 2008
• Program Committee, P2S2 2008
• Program Committee, ICPP 2008
• Program Committee, ACSAC 2008
• Poster/Presentation Chair, HiPC '07
• Program Committee, IISWC 2007
• Program Committee, Cluster 2007
• Program Committee, ACSAC 2007
• Program Committee, SMTPS Workshop at IPDPS 2007
• Poster/Presentation Chair, HiPC '06
• Program Committee, ICPADS 2006
• Program Committee, Eleventh Asia-Pacific Computer Systems Architecture Conference (ACSAC 2006)
• Program Committee, SMTPS Workshop at IPDPS 2006
• Poster/Presentation Chair, HiPC '05
• Program Committee, Tenth Asia-Pacific Computer Systems Architecture Conference (ACSAC'2005)
• Program Committee, SMTPS Workshop at IPDPS 2005
• Program Committee, Cluster 2004
• Program Committee, Ninth Asia-Pacific Computer Systems Architecture Conference (ACSAC'2004)
• Program Committee, ICPP 2004
• Program Committee, Cluster 2003
• Program Committee, Workshop on Parallel I/O in Cluster Computing and Computational Grids at CCGrid 2003
• Vice Program Chair, Cluster 2002
• Editorial Board of a Special Issue of Calculateurs Paralleles on Parallel I/O for Cluster Computing, 2001
• Program Committee, CCGrid2001
• Global Chair, Parallel I/O and Storage Technology, Euro-Par 2000
• BOFs (Birds-of-a-Feather sessions) Chair, SC2000
• Proposal Review Panels at the National Science Foundation, 1999
• General Co-Chair, IOPADS '99
• Ph.D. thesis committees of Rakesh Krishnaiyer (1998), Chutimet Srinilta (1998), and Jaechun No (1999) at Syracuse University; Murali Vilayannur at Penn State University (2005); Sachin More (2000), Avery Ching (2007), and Kenin Coloma (2007) at Northwestern University
• Program Committee, HiPC '99
• Program Committee, Heterogeneous Computing Workshop at IPPS '99
• Vice-Chair, Parallel I/O Workshop at Euro-Par '98
• Program Committee, PDCS-98
• Poster Exhibits Chair, SC98
• Co-Guest Editor, Special Issue of IJHPCA on I/O in Parallel Applications
• Tutorials Vice-Chair, SC97
• Program Committee, HiPC '97
• Publicity Chair, IOPADS '97
• Tutorials Vice-Chair, Supercomputing '96

Address

Tel: (630) 252-7847
email: thakur at mcs.anl.gov