Dependencies

Core

FFTW3_OMP, python(2|3), numpy, thurst, boost (preprocessor), g++/clang++ with C++11 and OpenMP support

Cuda

Cuda 8.0 or higher. Compute capacity 3.5.

Doc

Doxygen, Mercurial

Build

First time building

You should first clone the git submodules that are dependencies to tamaas:

git submodule update --init --recursive

If you wish to build Criterion tests, you should compile Criterion (optional, see below):

cd third-party/Criterion
mkdir build
cd build
cmake ..
cmake --build .
cd ../../..

The build system uses SCons. In order to construct the library you should hit:

scons

And to speedup the process you can do:

scons -j 6

In order to clean the build

scons -c

In order to compile in debug

scons build_type=debug

Indeed the default was

scons build_type=release

In order to make the compilation more verbose

scons verbose=true

You can, if you wish, not compile tests that use Criterion. Just deactivate the use_criterion flag (it is on by default):

scons use_criterion=false

For a list of all compilation options:

scons -h

You can customize a few compilation variables:

• CXX changes the compiler
• CXXFLAGS adds flags to the compilation process

For example:

• scons CXX=icpc compiles with the intel C++ compiler
• CXXFLAGS=-mavx2 scons adds the -mavx2 flag (which enables some Intel instructions) to the compilation process

If the CXX variable is defined in your environment, scons will take it as a default compiler.

To edit build configuration, simply edit the build-setup.conf at the repository root.

Build the doxygen

scons build_doc=true

Coding

Python

Tamaas is mainly used through the python interface. An example can be found in examples/new_contact.py

C++

TODO: update the example

Paralellism

Tamaas features shared-memory paralellism with OpenMP. The number of threads can be controlled via the OMP_NUM_THREADS environment variable or the omp_set_num_thread() function in the OpenMP API.

Cuda

To compile Tamaas with Cuda support use the following

scons backend=cuda