This directory contains implementations of Polybench parallel
benchmarks as originally completed for the Modulo Unrolling in Chapel
work at University of Maryland by Aroon Sharma, Darren Smith, Joshua
Koehler, and Rajeev Barua.
This work was published in the paper "Affine Loop Optimization Based on
Modulo Unrolling in Chapel" in PGAS 2014.


These benchmarks originally compared different distributions, including
a custom versions of Cyclic and Block Cyclic. These remain in this
directory as a historical artifact but have been disabled from the
tests.
They are CyclicZipOpt.chpl and MyBlockCyclic.chpl. See the paper
for the results of using these optimizations.


For more efficient parallel testing, most tests have been grouped
into subdirectories:

LinearAlgebra/
 2mm
 cholesky
 lu
 mvt
 syr2k
 syrk
 trmm

Stencil/
 fdtd-2d
 fdtd-apml
 jacobi-1d
 jacobi-2d
 stencil9

./
 correlation
 covariance
 floyd-warshall
 folding
 pascal
