Preamble
========

PARSEC is the result of the work of many people. First and foremost the authors
of the benchmark programs should be mentioned. They are too numerous to be
listed here, the README files and comments at the beginning of each source code
file provide detailed information about code authorship.

Our goal is to provide a complete list of all contributors. But we are not
impeccable, and it is possible that we did not list all contributors in a way
which sufficiently appreciates their work. In this case we would be grateful
for a notification which would allow us to correct our mistake.


PARSEC 3.0
==========

The following PARSEC users contributed to the next version of the benchmark
suite:

  - Rafael Asenjo, University of Malaga (dedup malloc patch)
  - Joe Devietti, University of Washington (file open mode patch)
  - Chris Fensch, University of Edinburgh (ferret deadlock & TBB patches)
  - Joseph Greathouse, University of Michigan (facesim patches)
  - Darryl Gove, Oracle (streamcluster data race)
  - Paul Keir, University of Glasgow (blackschole patches)
  - Alex Reshetov, Intel (raytrace patch)
  - George Kurian, MIT (provide splash2 package)

Three main changes to PARSEC 3.0 were done by Yungang Bao during whose 
postdoctral research at Princeton: 
  - Add network benchmarks as well as a user-level parallel TCP/IP stack
  - Integrate SPLASH-2 and the inputs-enlarged SPLASH-2x into PARSEC framework.
  - Redesign the framework for supporting external suites.


PARSEC 2.1
==========

The following PARSEC users contributed to the next version of the benchmark
suite:

  - Orestis Agathokleous, Cyprus University (fluidanimate bugfix)
  - Christian Fensch, University of Cambridge (x264 bugfix)
  - Mark Gebhart, University of Texas at Austin (Alpha support)
  - John Henning, Sun Microsystems (fluidanimate bugfix)
  - Thomas Karcher, University of Karlsruhe (several smaller fixes)
  - Grishma Kotecha, Intel (file input of fluidanimate TBB version)
  - Marc de Kruijf, University of Wisconsin-Madison (raytrace bugfix)
  - Michael Voss, Intel (blackscholes TBB improvements)
  - Matthew Watkins, Cornell University (streamcluster bugfix)


PARSEC 2.0
==========

The second release of PARSEC incorporated patches submitted by various users
of the benchmark suite. The following individuals contributed to PARSEC 2.0:

  - Gilberto Contreras, Princeton University (Intel TBB support for four
      workloads)
  - Christian Fensch, University of Cambridge (Solaris 10 / Sparc port of many
      PARSEC workloads)
  - Saugata Ghose, Cornell University (freqmine bugfixes)
  - Wim Heirman, University of Gent (GCC 4.3 port of PARSEC)
  - Nikolay Kurtov, Intel (blackscholes bugfixes, dedup bugfixes, Windows and
      OpenMP port of blackscholes)
  - Jiaqi Zhang, Tsinghua University (dedup bugfixes)

You can find details about the contributions in the CHANGELOG.


PARSEC 1.0
==========

The first version of PARSEC is the outcome of a joint venture between Princeton
University and Intel Corporation. It includes several key applications
developed at Princeton University and significant parts of the original RMS
Benchmark Suite from Intel Corporation.

The initial benchmark selection was decided on by Christian Bienia. He also
sized all input sets, wrote the PARSEC framework and created the final
structure of the PARSEC distribution. He was advised by Prof. Kai Li, who
provided helpful insights and directed the development of the suite.
Prof. Jaswinder Pal Singh assisted with his experience from the creation of
the SPLASH-2 benchmark suite.

At Intel Corporation, Tim Mattson was instrumental at facilitating the public
release of the RMS benchmark programs. He identified many of the programs which
could be released and convinced other people of the value of a public benchmark
suite. The Intel Application Research Lab had a leading role in the development
of the RMS applications. Stanford University allowed the release of parts of
their PhysBAM library.