Installing the DeePMD-kit’s python interface failure

[benshpiro]

Hi,
I managed to install the C++ interface, however, I am struggling with installing DeePMD-kit's python interface.
I am following the instructions on:
https://docs.deepmodeling.com/projects/deepmd/en/master/install/install-from-source.html

After the "pip install ." command (in the deepmd source directory) I am getting the following error:

---

CMake Error: Error: generator : Ninja
Does not match the generator used previously: Unix Makefiles
Either remove the CMakeCache.txt file and CMakeFiles directory or choose a different binary directory.
Traceback (most recent call last):
File "/tmp/pip-build-env-99826ywr/overlay/lib/python3.9/site-packages/skbuild/setuptools_wrap.py", line 632, in setup
env = cmkr.configure(
File "/tmp/pip-build-env-99826ywr/overlay/lib/python3.9/site-packages/skbuild/cmaker.py", line 334, in configure
raise SKBuildError(

An error occurred while configuring with CMake.
...
Please see CMake's output for more information.
[end of output]

note: This error originates from a subprocess, and is likely not a problem with pip.
ERROR: Failed building wheel for deepmd-kit
Failed to build deepmd-kit
ERROR: Could not build wheels for deepmd-kit, which is required to install pyproject.toml-based projects

---

Has anyone dealt with this error?

Thank you!

[njzjz]

The solution has been printed on the screen:

Either remove the CMakeCache.txt file and CMakeFiles directory or choose a different binary directory.

[njzjz]

I don't get it. Can you check if it exists in /home/bensh/miniconda3/envs/deePMD/lib/python3.9/site-packages/deepmd/pkg_config/run_config.ini? If not, did you see it copied (or generated) when you use pip install -v?

[benshpiro]

No it didn't exist.
I tried to start from the beginning and cloned deepmd-kit again. This time I see pkg_config/run_config.ini when I use pip install -v, but dp -h still gives an error (a different one this time...):

File "/home/bensh/miniconda3/envs/tf/lib/python3.9/site-packages/deepmd/env.py", line 269, in get_module
    module = tf.load_op_library(str(module_file))
  File "/home/bensh/miniconda3/envs/tf/lib/python3.9/site-packages/tensorflow/python/framework/load_library.py", line 54, in load_op_library
    lib_handle = py_tf.TF_LoadLibrary(library_filename)
tensorflow.python.framework.errors_impl.NotFoundError: /home/bensh/miniconda3/envs/tf/lib/python3.9/site-packages/deepmd/op/libop_abi.so: undefined symbol: _ZNK10tensorflow8OpKernel11TraceStringERKNS_15OpKernelContextEb

The above exception was the direct cause of the following exception:

Traceback (most recent call last):
  File "/home/bensh/miniconda3/envs/tf/bin/dp", line 5, in <module>
    from deepmd.entrypoints.main import main
  File "/home/bensh/miniconda3/envs/tf/lib/python3.9/site-packages/deepmd/__init__.py", line 7, in <module>
    import deepmd.utils.network as network
  File "/home/bensh/miniconda3/envs/tf/lib/python3.9/site-packages/deepmd/utils/__init__.py", line 2, in <module>
    from .data import DeepmdData
  File "/home/bensh/miniconda3/envs/tf/lib/python3.9/site-packages/deepmd/utils/data.py", line 10, in <module>
    from deepmd.env import GLOBAL_NP_FLOAT_PRECISION
  File "/home/bensh/miniconda3/envs/tf/lib/python3.9/site-packages/deepmd/env.py", line 354, in <module>
    op_module = get_module("op_abi")
  File "/home/bensh/miniconda3/envs/tf/lib/python3.9/site-packages/deepmd/env.py", line 325, in get_module
    raise RuntimeError(error_message) from e
RuntimeError: This deepmd-kit package is inconsitent with TensorFlow Runtime, thus an error is raised when loading op_abi. You need to rebuild deepmd-kit against this TensorFlow runtime.

How do I "rebuild deepmd-kit against this TensorFlow runtime" ?

[njzjz]

I try to reproduce the error but see nothing wrong: https://gist.github.com/njzjz/01fad54f6161e876cac5d1c46555fdef
I guess it's the same issue as #2023. Please provide more information about what your compiler is and how you install TensorFlow.

[benshpiro]

my gcc is 8.3.1 20190311 (Red Hat 8.3.1-3)
I installed tensorflow with:
pip install --upgrade tensorflow-cpu
I also tried using
pip install --upgrade tensorflow#

but I get the same error message as before:
RuntimeError: This deepmd-kit package is inconsitent with TensorFlow Runtime, thus an error is raised when loading op_abi. You need to rebuild deepmd-kit against this TensorFlow runtime.

[njzjz]

As explained in #2023, this is a bug from the Red Hat compilers. You can build a compiler by yourself.