Note: At present CUDA is not requires as the GPU component of the library has not yet been implemented.
You can get the nano library from here -- nano -- it is a header only library.
By default it is installed to usr/include. If you install it somewhere else make sure that it is on your $PATH
so that tensor can find it.
You can get the Boost test library from here -- Boost.Test -- and follow the Getting Started Guide to install the libraries.
Note: You only need the test library from Boost, but it is dynamically linked in the Makefile provided with the tensor library, so you should install it to allow dynamic linking (i.e the library needs to be installed and not just the header files).
The following compilers have been tested:
g++ (version 4.9.1) clang++ (version 3.6.1)
Since tensor is (currently) a header-only library, there is nothing to install if you would just like to use it in your own application. However, tests are provided with tensor so ensure that everything is qorking as expected as well as to provide examples of the usage of tensor.
To compile the tests, cd into tests/
, at which point you are provided with a few options:
To make all the tests, simply issue
make all (or make -jN all -- to make in parallel with N processes)
The following individual test components are provided
To make an individual tests, issuse
make <test_name>
with the appropriate test name, for example make container
To clean the tests, issue
make clean