Closed MaximeVdB closed 2 years ago
The windows executable takes 12 sec and that seems very reasonable to me for the testcase. I need to say that I have no idea how much the official windows executable (only available one is v6, we can ask Dirk for v7) would take for the same task.
By default, our binaries (executable and (shared) library) are dynamically linked to e.g. the
libgfortran
runtime library. This PR adds the option for static builds (with theSTATIC=1
make option). So these are self-contained binaries which can then be released as precompiled standalone versions (at least theaquacrop
executable; there is probably less interest in releasing alibaquacrop.a
static library).Performed tests:
STATIC=0
) are still passing withFORTRAN_EXE=0/1
andDEBUG=0/1
with both the Singularity image and thefoss-2018a
toolchain.STATIC=1
on Genius and running the Perennial testcase with it works on a Ubuntu WSL2 on my local Windows laptop. The run time was about 3 seconds, which is similar to what I get on a Genius login node.STATIC=1
andFC=x86_64-w64-mingw32-gfortran
on a Ubuntu WSL2 on my local Windows laptop (MinGW v7.3) works. I can then execute the same Perennial test in a Windows Powershell, and the resulting outputs are identical to the reference. The run time was about 12 seconds, which is clearly slower than usual (but still acceptable?).