README

This directory includes code used for experiments presented in the SC'12 
paper. These generated codes were tuned for an Intel Xeon based on the 
Westmere-EP architecture, although they should perform well on any other 
multi-core.  It is recommended that these be compiled with Intel's C/C++ 
compiler (14.0 or higher) for best performance.

Use Pluto to regenerate for a different tile size or tune for a different 
architecture; use Intel C/C++ compiler 14.0 or higher so that the generated 
codes are vectorized well.