README

   LU Decomposition with Optimization for the Intel MIC Architecture
   Copyright 2015, Colfax International

   Author:  andrey@colfax-intl.com  Andrey Vladimirov
            phi@colfax-intl.com     General inquiries


DESCRIPTION: The code in this archive supplements the publication
	     "Fine-Tuning Vectorization and Memory Traffic
              on Intel Xeon Phi Coprocessors:	
              LU Decomposition of Small Matrices"
             (A. Vladimirov, 2015 -- Colfax Research 
              http://colfaxresearch.com/fine-tuning-vectorization-and-memory-traffic-on-intel-xeon-phi-coprocessors-lu-decomposition-of-small-matrices/ )

	     Directories step-00/ through step-05/ contain
             the LU decomposition code at different stages of optimization,
             with step-05/ being the most optimized.
             Directory step-mkl/ contains the code used for Intel MKL benchmarks.

REQUIREMENTS:
  - Intel C++ compiler version 15.0.1.133 or greater;
  - Multi-core processor based on Intel architecture;
  - 8 GB of RAM or more;
  - An Intel Xeon Phi coprocessor with passwordless SSH 
    authentication configured
  - Linux operating system in order to use the included
    Makefile and benchmark script.

EXAMPLES OF USAGE:
  - To compile the code in one of the steps, run "make"
  - To execute the code on the CPU, run "make run-cpu"
  - To execute the code on the coprocessor, run "make run-mic"