Examples in the integration test reported segmentation fault on my machine

[xuan112358]

Describe the bug

When running ABACUS with code after commit #4622, I found that I can't run examples in integrate test.
They all reported segmentation fault.
Take “tests/integrate/201_NO_15_pseudopots” for example, the log.txt was:

Expected behavior

No response

To Reproduce

No response

Environment

compiler: -DMPI_CXX_COMPILER=mpiicpc -DCMAKE_CXX_COMPILER=icpc
build with deepks and libri

Additional Context

No response

Task list for Issue attackers (only for developers)

• Verify the issue is not a duplicate.
• Describe the bug.
• Steps to reproduce.
• Expected behavior.
• Error message.
• Environment details.
• Additional context.
• Assign a priority level (low, medium, high, urgent).
• Assign the issue to a team member.
• Label the issue with relevant tags.
• Identify possible related issues.
• Create a unit test or automated test to reproduce the bug (if applicable).
• Fix the bug.
• Test the fix.
• Update documentation (if necessary).
• Close the issue and inform the reporter (if applicable).

[dyzheng]

What is your compiling command?

[xuan112358]

@dyzheng
cmake -B build -DCMAKE_INSTALL_PREFIX=./
-DELPA_DIR=/home/xuan/03_library/elpa/elpa-2021.05.002 -DELPA_INCLUDE_DIRS=/home/xuan/03_library/elpa/elpa-2021.05.002/include/elpa-2021.05.002
-DCEREAL_INCLUDE_DIR=/home/xuan/03_library/cereal/include
-DMPI_CXX_COMPILER=mpiicpc -DCMAKE_CXX_COMPILER=icpc
-DTorch_DIR=/home/xuan/03_library/libtorch/share/cmake/Torch/
-Dlibnpy_INCLUDE_DIR=/home/xuan/03_library/libnpy-new/include -DENABLE_DEEPKS=1 -DLibxc_DIR=/home/xuan/03_library/libxc/libxc-5.2.3 -DENABLE_LIBRI=1
&& cmake --build build -j30 && cmake --install build

[Cstandardlib]

I will check with deepks and libri on my machine right away.

[dyzheng]

I have tested both in GNU and Intel environment and cannot rerun your error.

[dyzheng]

Intel environment can rerun your error with LibRI and DeePKS

[dyzheng]

Error comes from DeePKS with Intel compiler.

[dyzheng]

Can you try the newest commit, I test it with no error.

[xuan112358]

@dyzheng
I can't compile with the command above after commit #4613
The error is :

[dyzheng]

@xuan112358 can you rerun the toolchain to rebuild libtorch in your machine?

[xuan112358]

@dyzheng It seems like I can just download the LibTorch zip file but not need to build it? I downloaded the new LibTorch v2.3.1 in https://download.pytorch.org/libtorch/cpu/libtorch-cxx11-abi-shared-with-deps-2.3.1%2Bcpu.zip and tried again. The error was the same.

[caic99]

Duplicated with #4636. Let's track the problem there.

[Cstandardlib]

Build #4622 with IntelLLVM 2024.1.0 and encountered a compilation error

Build latest with IntelLLVM 2024.1.0
Warnings are generated by cmake as follows:

and build failed with error

Building command:
CXX=mpiicpx cmake -B build -DBUILD_TESTING=1 -DENABLE_LIBRI=ON -DENABLE_DEEPKS=1 -DTorch_DIR=~/develop/libtorch/share/cmake/Torch/ -Dlibnpy_INCLUDE_DIR=~/develop/libnpy/include
with https://download.pytorch.org/libtorch/cpu/libtorch-cxx11-abi-shared-with-deps-2.3.1%2Bcpu.zip

[caic99]

@Cstandardlib Thanks for testing, and would you please apply the modifications in #4644 ?

[Cstandardlib]

@caic99 Thanks! I'll update my test results in #4636.

[caic99]

Hi @xuan112358 ,
I've updated the building process. Would you try using the latest codes to compile and run it again? Thanks.

[xuan112358]

@caic99 I‘ve tried it and all worked well now ! Also , I've tried to comile and run with Intel oneAPI toolkit 2024 and it worked well too. Thanks very much!