在 M1 Mac 上的 Conda 环境中安装 Tensorflow 时 Tensorboard 安装失败

Question

I'm trying to install tensorflow on an M1 Mac. I've been trying to follow this tutorial to install the pre release version of tensorflow for Mac . From the pre-release repo:

This pre-release delivers hardware-accelerated TensorFlow and TensorFlow Addons for macOS 11.0+. Native hardware acceleration is supported on M1 Macs and Intel-based Macs through Apple's ML Compute framework.

The pre-release requires Python 3.8, so that's what I set up my conda environment with.

The tutorial above tries to install a previous version of tensorflow for M1 Mac, so I adjusted the commands to use the version of tensorflow for M1 Mac that I've downloaded.

I've been trying to do this install in a virtual environment using Miniforge , as it is what multiple tutorials have recommended for M1 Mac installs. From the Miniforge repo:

This repository holds a minimal installer for Conda specific to conda-forge. It is comparable to Miniconda

---

When I get to the step to install tensorboard, pip install tensorboard , the install gets as far as:

Building wheels for collected packages: grpcio
  Building wheel for grpcio (setup.py) ... error

The errors go on for many pages, but here is the last block:

    Traceback (most recent call last):
      File "/private/var/folders/nq/2mtcpkp51dd149_bnk1g1t340000gn/T/pip-install-ze40tjy3/grpcio_c9426269e87d4673a65bd63bd6af18ee/src/python/grpcio/commands.py", line 282, in build_extensions
        build_ext.build_ext.build_extensions(self)
      File "/Users/bitflow/miniforge3/envs/conda-tf-venv/lib/python3.8/distutils/command/build_ext.py", line 449, in build_extensions
        self._build_extensions_serial()
      File "/Users/bitflow/miniforge3/envs/conda-tf-venv/lib/python3.8/distutils/command/build_ext.py", line 474, in _build_extensions_serial
        self.build_extension(ext)
      File "/Users/bitflow/miniforge3/envs/conda-tf-venv/lib/python3.8/site-packages/setuptools/command/build_ext.py", line 208, in build_extension
        _build_ext.build_extension(self, ext)
      File "/Users/bitflow/miniforge3/envs/conda-tf-venv/lib/python3.8/distutils/command/build_ext.py", line 528, in build_extension
        objects = self.compiler.compile(sources,
      File "/private/var/folders/nq/2mtcpkp51dd149_bnk1g1t340000gn/T/pip-install-ze40tjy3/grpcio_c9426269e87d4673a65bd63bd6af18ee/src/python/grpcio/_parallel_compile_patch.py", line 58, in _parallel_compile
        multiprocessing.pool.ThreadPool(BUILD_EXT_COMPILER_JOBS).map(
      File "/Users/bitflow/miniforge3/envs/conda-tf-venv/lib/python3.8/multiprocessing/pool.py", line 364, in map
        return self._map_async(func, iterable, mapstar, chunksize).get()
      File "/Users/bitflow/miniforge3/envs/conda-tf-venv/lib/python3.8/multiprocessing/pool.py", line 771, in get
        raise self._value
      File "/Users/bitflow/miniforge3/envs/conda-tf-venv/lib/python3.8/multiprocessing/pool.py", line 125, in worker
        result = (True, func(*args, **kwds))
      File "/Users/bitflow/miniforge3/envs/conda-tf-venv/lib/python3.8/multiprocessing/pool.py", line 48, in mapstar
        return list(map(*args))
      File "/private/var/folders/nq/2mtcpkp51dd149_bnk1g1t340000gn/T/pip-install-ze40tjy3/grpcio_c9426269e87d4673a65bd63bd6af18ee/src/python/grpcio/_parallel_compile_patch.py", line 54, in _compile_single_file
        self._compile(obj, src, ext, cc_args, extra_postargs, pp_opts)
      File "/private/var/folders/nq/2mtcpkp51dd149_bnk1g1t340000gn/T/pip-install-ze40tjy3/grpcio_c9426269e87d4673a65bd63bd6af18ee/src/python/grpcio/commands.py", line 265, in new_compile
        return old_compile(obj, src, ext, cc_args, extra_postargs,
      File "/Users/bitflow/miniforge3/envs/conda-tf-venv/lib/python3.8/distutils/unixccompiler.py", line 120, in _compile
        raise CompileError(msg)
    distutils.errors.CompileError: command 'gcc' failed with exit status 1
    
    ----------------------------------------
ERROR: Command errored out with exit status 1: /Users/bitflow/miniforge3/envs/conda-tf-venv/bin/python3.8 -u -c 'import io, os, sys, setuptools, tokenize; sys.argv[0] = '"'"'/private/var/folders/nq/2mtcpkp51dd149_bnk1g1t340000gn/T/pip-install-ze40tjy3/grpcio_c9426269e87d4673a65bd63bd6af18ee/setup.py'"'"'; __file__='"'"'/private/var/folders/nq/2mtcpkp51dd149_bnk1g1t340000gn/T/pip-install-ze40tjy3/grpcio_c9426269e87d4673a65bd63bd6af18ee/setup.py'"'"';f = getattr(tokenize, '"'"'open'"'"', open)(__file__) if os.path.exists(__file__) else io.StringIO('"'"'from setuptools import setup; setup()'"'"');code = f.read().replace('"'"'\r\n'"'"', '"'"'\n'"'"');f.close();exec(compile(code, __file__, '"'"'exec'"'"'))' install --record /private/var/folders/nq/2mtcpkp51dd149_bnk1g1t340000gn/T/pip-record-t11k_i1o/install-record.txt --single-version-externally-managed --compile --install-headers /Users/bitflow/miniforge3/envs/conda-tf-venv/include/python3.8/grpcio Check the logs for full command output.

There was an earlier step to install grpcio, that claimed to work successfully, that used the following command:

pip install --upgrade -t “$env/lib/python3.8/site-packages/” --no-dependencies --force "$libs/grpcio-1.33.2-cp38-cp38-macosx_11_0_arm64.whl"

---

I tried to import tensorflow in python, without having tensorboard installed, and the only error appears to be that tensorboard is not yet installed. This gives me some confidence that the rest of the set up went okay.

>>> import tensorflow
WARNING:root:Limited tf.compat.v2.summary API due to missing TensorBoard installation.
WARNING:root:Limited tf.compat.v2.summary API due to missing TensorBoard installation.
WARNING:root:Limited tf.compat.v2.summary API due to missing TensorBoard installation.
WARNING:root:Limited tf.summary API due to missing TensorBoard installation.

I double checked, and retried, the step around installing grpcio before trying again. While grpcio seemed to install okay again, the tensorboard install still failed in the same way.

I'd really appreciate any help that could be offered to solve this.

Answer 1

I ended up referencing a lot of different materials to try get my solution working. There are two main materials which I can credit with my success:

• The "Troubleshooting and common errors" section of Clayton Pilat's tutorial on how to set up tensorflow on an M1 Mac

grpcio-1.33.2-cp38-cp38-macosx_11_0_arm64.whl is not a supported wheel on this platform.

For some reason the version of Python that you download from the website for MacOS doesnt seem to like these files, so uninstall your current version of Python (if you dont know how to do that click here ). And then install the Python from xcode by pasting this into your terminal.

xcode-select --install

• The environment variables and flags that github user Tenzer used to install grpcio

I had to set a few environment variables to get it to work:

 GRPC_PYTHON_BUILD_SYSTEM_OPENSSL=1 GRPC_PYTHON_BUILD_SYSTEM_ZLIB=1

to make the installation process use the libraries I've already got on the system, and then the following to point it at OpenSSL installed via Homebrew:

 CFLAGS="-I/opt/homebrew/opt/openssl/include" LDFLAGS="-L/opt/homebrew/opt/openssl/lib"

I was focused on trying to fix the issue with the building of grpcio during the tensorboard install. To that end, following the above links, and throwing in a restart for the fun of it, I was finally able to install tensorboard.

Now, if I am using Python in my conda environment, I can import tensorflow and it works just fine!