README

: BUILD PREREQUISITES:

Update create symbolic links for the correct python version
and CUDA hardware SM compute level.

:: Python version dependent cudalt.py script
Link, or write and link, example:
ln -s cudalt-2.7.py cudalt.py
        or
ln -s cudalt-3.2.py cudalt.py


:: CUDA Makefile include:
Various Nvidia cards have differing SM compute levels.  For this
reason, a cuda.<sm>.mk file is provided for each relevant SM.

You must create a symlink to the proper Make include file or you
will get the error:
----------------------------------------------------------------------
automake-1.13: error: cannot open < cuda.mk: No such file or directory
autoreconf-2.68: automake failed with exit status: 1
----------------------------------------------------------------------

For example, on a Kepler 
ln -s cuda.20.mk cuda.mk

For Fermi, use 
ln -s cuda.30.mk cuda.mk

: AUTOTOOLS BUILD & INSTALL
autoreconf -ivf && ./configure && make
sudo make install


THANKS:

Autotools + CUDA derived from ClusterChimps
www.clusterchimps.org