README

This is the COmmon Data Exchange and Rconstruction Engine (codeare).
This software is a framework for live communication with MRI sequence and 
reconstruction software. It is distributed under der LGPL v3.0

BUILD & INSTALL
For example:
# cd codeare
# mkdir build
# cd build
# cmake ..
# make 
# make test
# sudo make install

Linux optimised compilation examples:
(64bit Linux, Intel Composer XE 2015, MKL, MATLAB 2015a)
# . /opt/intel/bin/compilervars.sh intel64
# export LD_LIBRARY_PATH=/opt/intel/lib/intel64:/opt/intel/mkl/lib/intel64
# cd codeare
# mkdir build
# cd build
# cmake -DCMAKE_C_COMPILER=icc \
        -DCMAKE_CXX_COMPILER=icpc \
        -DBLA_VENDOR=Intel10_64lp \
        -DMATLAB_ROOT=/usr/local/MATLAB/R2015a ..

(64bit Linux, ACML 6.1, MATLAB 2015a)
# cd codeare
# mkdir build
# cd build
# cmake -DBLA_VENDOR=ACML_MP \ 
        -DBLAS_acml_mv_LIBRARY=/opt/acml/gfortran64_mp/lib/libacml_mp.so \
        -DMATLAB_ROOT=/usr/local/MATLAB/R2015a ..

Test runs:

Mac OS X, Quad-core Intel Core i7 @ 2.5GHz
Accelerate framework: 17.8s

Ubuntu Linux, 12-core Intel Xeon E5-2680v3 @ 2.50GHz
Test run On Intel Xeon E5-2680v3 @ 2.50GHz: 
MKL:              26.0s
ACML:             28.0s
OpenBLAS + ATLAS: 29.2s

Ubuntu Linux, 16-core AMD Opteron 6300 @ 2.4GHz
ACML:             42.7s
MKL:              43.2s
OpenBLAS + ATLAS: 46.6s

Windows 8.1, 4-core Core i5 @ 1.9GHz
MKL 15.0:         42.7s

For more information refer to http://codeare.org