mdolab / adflow

ADflow is a finite volume RANS solver tailored for gradient-based aerodynamic design optimization.
Other
229 stars 100 forks source link

Installation failed on Centos 7-based cluster #363

Open helyamany opened 2 months ago

helyamany commented 2 months ago

I attempted to install ADflow on a homogeneous HPC supercomputer (all nodes share the same architecture) using Intel Composer 2017. Initially, I encountered the following error: 'for' loop initial declarations are only allowed in C99 mode. I resolved this by adding -std=c99 to the C flags in the Makefile. However, upon rerunning the build, I encountered a new error: (mpicc -fPIC -O -std=c99 -I../c_defines -I../metis-4.0 -I/path/to/petsc/build/include -c ../adjoint/ADFirstAidKit/adStack.c -o adStack.o

    --- Compiled metisInterface.c successfully ---

../adjoint/ADFirstAidKit/adStack.c:65:1: error: unknown type name ‘u_int64_t’ static u_int64_t pushPopTraffic = 0 ;).

I haven't been able to resolve this issue. Could you please provide assistance or any tips to address this? (PS. I installed the CGNS and PETSc from the source using the same compiler)

Moreover, I attempted another way to install it by creating a conda environment having the latest versions of all recommended packages on this website (https://mdolab-mach-aero.readthedocs-hosted.com/en/latest/installInstructions/install3rdPartyPackages.html) using anaconda 2024 but failed and gave the following error: (../adjoint/outputReverseFast/utils_fast_b.f90 /path/to/adflow/src/build/fort_depend.py:86: SyntaxWarning: invalid escape sequence '\s')

Code versions for the first attempt

Code versions for the second attempt

I really appreciate any help you can provide.

A-CGray commented 2 months ago

Can you try doing the installation with the package versions we currently use on our "latest" docker images? I think your first attempt is using too old a version of python and the second attempt is using too new a version. You can find the details on the MPI versions we use here and other package versions here

helyamany commented 2 months ago

Hi Ali, Thank you for your quick reply. Unfortunately, I could not access the two links you sent. Both took me to empty pages. Can you please resend it again? Also, what is the best approach if I want to install the software without using dockers? Thank you so much,

marcomangano commented 2 months ago

Check the table at the top of this page with the tested dependencies: https://mdolab-mach-aero.readthedocs-hosted.com/en/latest/installInstructions/install3rdPartyPackages.html I believe both the Python and PETSc versions might be too new, type errors are very common in this case.

helyamany commented 2 months ago

Hi Marco, Thank you for your comment. I'll give it another try using the recommended versions and hopefully resolve all the errors. Thank you,

gawng commented 1 month ago

@helyamany any progress on this? Are there still errors?