It is not used in current code, it was used only in some legacy benchmarks.

And of course our implementation is faster ;-)

Standard CMAKE_CXX_FLAGS are a space-delimited list, but our
CXX_TESTS_FLAGS are manually passed to CUDA_ADD_EXECUTABLE, which
expects a semicolon-delimited list. Blame cmake for such a crazy
language...

This does not need any code in cmake as the macro is defined by the
compiler itself. See also the table here:
http://nadeausoftware.com/articles/2012/10/c_c_tip_how_detect_compiler_name_and_version_using_compiler_predefined_macros

Přidán parametr build scriptů --have-mic=yes, je podmíněný použitím icpc a vypíná CUDU (nastaví cuda=no než se začna cuda kontrolovat)

We must pass the CUDA_HOST_COMPILER variable manually to use the
compiler configured by the user.

This does not have to be tested and configured manually :)

Plus the associated revamping of the SolverStarter. Also
ODESolverMonitor has been removed as its functionality has been merged
into the IterativeSolverMonitor.