Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Windows installations fail #216

Open
folterj opened this issue Jan 11, 2024 · 20 comments
Open

Windows installations fail #216

folterj opened this issue Jan 11, 2024 · 20 comments
Assignees
Labels
bug Something isn't working

Comments

@folterj
Copy link

folterj commented Jan 11, 2024

All Windows installations as described appear to fail for me.
OS: Windows 10 64-bit
Python: 3.x 64-bit
GPU: Nvidia GTX 1050 Ti

I've tried the approaches as documented without success:

  • The easy way (tensorflow with and without -gpu)
  • Using Beta Installers
  • Installing via Conda (/pip)
  • Compile it yourself
    • creating a local conda channel, and installing from there
    • running CMake/build manually with customized options

The main error appears to be that a suitable version of tensorflow cannot be found.
As a side note, according to the tensorflow documentation, Windows GPU is supported up to TF version 2.10, is currently preferred installed via PIP, not using the 'tensorflow-gpu' package as the regular package automatically detects GPU. I have been able to install tensorflow (and pytorch) separately with GPU support.
Installing tensorflow separately before installing TREX, results in TREX unable to find tensorflow, even if the expected version is installed (2.8). I've attempted installing various versions of Python 3 (all 64 bit) and tensorflow (with and without -gpu).
The Windows beta installer fails on env/version conflicts (see attached file install.log).
For completeness (though this is my least favourite method), the compile using CMake approach fails as bellow (I note it appears to use python 3.12):

  test_python.cpp
C:\Users\Y\PycharmProjects\trex\Application\build\_deps\pybind11-src\include\pybind11\detail\type_cas
ter_base.h(482,21): error C2027: use of undefined type '_frame' [C:\Users\Y\PycharmProjects\trex\Appl
ication\build\src\commons\examples\test_python.vcxproj]
  (compiling source file '../../../../src/commons/examples/test_python.cpp')
  C:\Program Files\Python312\include\pytypedefs.h(22,16):
  see declaration of '_frame'

Another note is that the python and tensorflow versions do not appear to be pinned in the configurations.

@folterj folterj added the bug Something isn't working label Jan 11, 2024
@roaldarbol
Copy link

I'm running into the same issue, just to confirm that it indeed is an issue.

@mooch443
Copy link
Owner

Hey, I assume this is to be expected - did you try the normal versions first? Or what is your use case?

I have deleted the beta installers since they weren't supposed to be fully public yet - I had just given them to some people and they become outdated pretty quickly :-) I assume they all fail, although I haven't tested in a while.

@roaldarbol
Copy link

Hey! The issue is that the normal version doesn't install, I think that's why we've both tried using the beta installer too. Here's the log from right now, trying a few different ways to install the non-beta, first with tensorflow-gpu which is the one showing on trex.run:

(base) C:\>conda create -n tracking -c trexing trex numpy=1.23 tensorflow-gpu
Retrieving notices: ...working... done
Collecting package metadata (current_repodata.json): done
Solving environment: failed with repodata from current_repodata.json, will retry with next repodata source.
Collecting package metadata (repodata.json): done
Solving environment: failed

PackagesNotFoundError: The following packages are not available from current channels:

  - tensorflow-gpu

Current channels:

  - https://conda.anaconda.org/trexing/win-64
  - https://conda.anaconda.org/trexing/noarch
  - https://conda.anaconda.org/conda-forge/win-64
  - https://conda.anaconda.org/conda-forge/noarch

To search for alternate channels that may provide the conda package you're
looking for, navigate to

    https://anaconda.org

and use the search bar at the top of the page.

and here with just tensorflow:

(base) C:\>conda create -n tracking -c trexing trex numpy=1.23 tensorflow
Collecting package metadata (current_repodata.json): done
Solving environment: failed with repodata from current_repodata.json, will retry with next repodata source.
Collecting package metadata (repodata.json): done
Solving environment: /
Found conflicts! Looking for incompatible packages.
This can take several minutes.  Press CTRL-C to abort.
failed

UnsatisfiableError: The following specifications were found to be incompatible with each other:

Output in format: Requested package -> Available versions

Package tk conflicts for:
numpy=1.23 -> pypy3.9[version='>=7.3.9'] -> tk[version='>=8.6.11,<8.7.0a0|>=8.6.12,<8.7.0a0|>=8.6.13,<8.7.0a0']
trex -> python[version='>=3.9,<3.10.0a0'] -> tk[version='>=8.6.11,<8.7.0a0|>=8.6.12,<8.7.0a0|>=8.6.13,<8.7.0a0']

Package zlib conflicts for:
tensorflow -> grpcio[version='>=1.8.6'] -> zlib[version='1.2.*|1.2.11|>=1.2.11,<1.3.0a0|>=1.2.12,<1.3.0a0|1.2.8']
numpy=1.23 -> pypy3.9[version='>=7.3.9'] -> zlib[version='>=1.2.11,<1.3.0a0|>=1.2.12,<1.3.0a0']

Package numpy conflicts for:
numpy=1.23
tensorflow -> numpy[version='>=1.11.0|>=1.12.1|>=1.13.3']
trex -> scikit-learn -> numpy[version='>=1.11.3,<2.0a0|>=1.14.6,<2.0a0|>=1.16.5,<2.0a0|>=1.16.6,<2.0a0|>=1.17.5,<2.0a0|>=1.18.5,<2.0a0|>=1.19.5,<2.0a0|>=1.20.3,<2.0a0|>=1.21.6,<2.0a0|>=1.22.4,<2.0a0|>=1.23.5,<2.0a0|>=1.26.3,<2.0a0|>=1.26.2,<2.0a0|>=1.26.0,<2.0a0|>=1.23.4,<2.0a0|>=1.21.5,<2.0a0|>=1.21.4,<2.0a0|>=1.19.4,<2.0a0|>=1.19.2,<2.0a0|>=1.22.4,<1.28|>=1.26.3,<1.28|>=1.23.5,<1.28|>=1.26.0,<1.28|>=1.21.6,<1.28|>=1.21.6,<1.27|>=1.23.5,<1.27|>=1.20.3,<1.27|>=1.20.3,<1.26|>=1.23.4,<1.26|>=1.21.6,<1.26|>=1.20.3,<1.25|>=1.21.6,<1.25|>=1.21.6,<1.23|>=1.20.3,<1.23|>=1.12.0|>=1.9.1']
trex -> numpy==1.18.5
tensorflow -> tensorflow-base==1.14.0=py36h9f0ad1d_0 -> numpy[version='>=1.12.0|>=1.16.1|>=1.16.1,<2.0.0a0|>=1.9.1']

Package tensorflow conflicts for:
trex -> keras -> tensorflow[version='>=2.2']
trex -> tensorflow=2.6
tensorflow

without numpy=1.23:

(base) C:\>conda create -n tracking -c trexing trex tensorflow
Collecting package metadata (current_repodata.json): done
Solving environment: failed with repodata from current_repodata.json, will retry with next repodata source.
Collecting package metadata (repodata.json): done
Solving environment: /
Found conflicts! Looking for incompatible packages.
This can take several minutes.  Press CTRL-C to abort.
failed

UnsatisfiableError: The following specifications were found to be incompatible with each other:

Output in format: Requested package -> Available versions

Package tensorflow conflicts for:
trex -> tensorflow=2.6
tensorflow
trex -> keras -> tensorflow[version='>=2.2']

and finally with the command from the latest Github release:

(base) C:\>conda create -n track -c trexing trex
Collecting package metadata (current_repodata.json): done
Solving environment: failed with repodata from current_repodata.json, will retry with next repodata source.
Collecting package metadata (repodata.json): done
Solving environment: |
Found conflicts! Looking for incompatible packages.
This can take several minutes.  Press CTRL-C to abort.
failed

UnsatisfiableError:

@roaldarbol
Copy link

Hey @mooch443! Just to prioritise my tracking in the coming weeks - do you reckon you will have time to look at/sort this one out? I still have the option of running the tracking on my Mac, but it's also my writing/analysis workhorse, so will take me longer to get through the tracking (it then runs nights only). :-)

@mooch443
Copy link
Owner

mooch443 commented Mar 2, 2024

conda create -n trex_normal -c trexing trex tensorflow-gpu numpy=1.23.1
this works for me. in general the trouble seems to be that the available software in conda repositories has evolved, but unfortunately i hadn't specified the exact numpy version in this older version of the software. specifying 1.23.1 worked for me (on windows).

@roaldarbol
Copy link

Still doesn't work for me:

(base) C:\>conda create -n trex_normal -c trexing trex tensorflow-gpu numpy=1.23.1
Collecting package metadata (current_repodata.json): done
Solving environment: failed with repodata from current_repodata.json, will retry with next repodata source.
Collecting package metadata (repodata.json): done
Solving environment: failed

PackagesNotFoundError: The following packages are not available from current channels:

  - tensorflow-gpu

Current channels:

  - https://conda.anaconda.org/trexing/win-64
  - https://conda.anaconda.org/trexing/noarch
  - https://conda.anaconda.org/conda-forge/win-64
  - https://conda.anaconda.org/conda-forge/noarch

To search for alternate channels that may provide the conda package you're
looking for, navigate to

    https://anaconda.org

and use the search bar at the top of the page.

@roaldarbol
Copy link

roaldarbol commented Mar 21, 2024

Ah, I think I found the issue!
How have you installed conda @mooch443 - with Anaconda? When you create a conda env, does it use a different channel by default? I think if you've installed conda with Anaconda, the default channel is anaconda, whereas installed with miniforge, the default is conda-forge.
Now it seems that tensorflow-gpu is on anaconda, bu NOT on conda-forge, which could explain the discrepancies. So maybe we should be more explicit and add the anaconda channel into the one-liner to ensure it works across all conda installations.

This works for me:

conda create -n trexenv -c trexing -c anaconda trex tensorflow-gpu

(and much much faster solving by using pixi)

So seems like numpy may not have been the culprit after all 👍

@mooch443
Copy link
Owner

What version of conda do you have? Maybe you should update that? Because in newer versions of conda you'll get a summary of the current channels being used before the error messages. And it still works for me with the exact command. So within the base environment I recommend doing:

conda update --override-channels -c defaults --all

This is good practice anyway. They may change how information is retrieved from channels in new versions. Make sure it actually updates something and then try again. You can also try adding the defaults to the previous command:

conda create -n trex_normal --override-channels -c trexing -c defaults trex tensorflow-gpu numpy=1.23.1

This is just "conda debugging", which is the most boring part of software development. I just merged my current version into the main dev branch btw, trying to find the last couple missing pieces ;) Looking forward to adding actual new features afterward. You can try this, but you might get a version thats bugged depending on when you download:

# windows or linux
conda create -n betarex --override-channels -c trex-beta -c pytorch -c nvidia -c defaults trex

@mooch443
Copy link
Owner

mooch443 commented Mar 21, 2024

Ah, I think I found the issue! How have you installed conda @mooch443 - with Anaconda? When you create a conda env, does it use a different channel by default? I think if you've installed conda with Anaconda, the default channel is anaconda, whereas installed with miniforge, the default is conda-forge. Now it seems that tensorflow-gpu is on anaconda, bu NOT on conda-forge, which could explain the discrepancies. So maybe we should be more explicit and add the anaconda channel into the one-liner to ensure it works across all conda installations.

No I have not installed Anaconda - I am using miniconda, so its just the main/defaults channel. But the solution is to override-channels as in my previous message then. Also updating conda might be a good idea, and getting rid of the anaconda part. Nobody needs the GUI :-D

This works for me:

conda create -n trexenv -c trexing -c anaconda trex tensorflow-gpu

This seems wrong. Can you try my command I suggested above (after updating)?

@roaldarbol
Copy link

roaldarbol commented Mar 21, 2024

What is it precisely you dislike about the solution I used? I'm not sure I like the idea about using defaults channel as that seems to vary between installations, and it's terribly implicit - even when I can now see which channels are being used, I have no idea which channel repository is being used as default (you may know this much better than me, am I understanding this correctly?)

I ran the updating, now none of the solutions work anymore:

(base) C:\>conda create -n trex_normal --override-channels -c trexing -c defaults trex tensorflow-gpu numpy=1.23.1
Channels:
 - trexing
 - defaults
Platform: win-64
Collecting package metadata (repodata.json): done
Solving environment: done

## Package Plan ##

  environment location: C:\Users\mr630\AppData\Local\mambaforge\envs\trex_normal

  added / updated specs:
    - numpy=1.23.1
    - tensorflow-gpu
    - trex


The following packages will be downloaded:

    package                    |            build
    ---------------------------|-----------------
    _tflow_select-2.1.0        |              gpu           3 KB
    abseil-cpp-20210324.2      |       hd77b12b_0         1.6 MB
    absl-py-1.4.0              |   py39haa95532_0         180 KB
    aiohttp-3.9.3              |   py39h2bbff1b_0         660 KB
    aiosignal-1.2.0            |     pyhd3eb1b0_0          12 KB
    astor-0.8.1                |   py39haa95532_0          47 KB
    astunparse-1.6.3           |             py_0          17 KB
    async-timeout-4.0.3        |   py39haa95532_0          13 KB
    attrs-23.1.0               |   py39haa95532_0         143 KB
    blas-1.0                   |              mkl           6 KB
    blinker-1.6.2              |   py39haa95532_0          29 KB
    brotli-python-1.0.9        |   py39hd77b12b_7         309 KB
    cachetools-4.2.2           |     pyhd3eb1b0_0          13 KB
    certifi-2024.2.2           |   py39haa95532_0         160 KB
    cffi-1.16.0                |   py39h2bbff1b_0         242 KB
    click-8.1.7                |   py39haa95532_0         164 KB
    colorama-0.4.6             |   py39haa95532_0          32 KB
    cryptography-41.0.3        |   py39h3438e0d_0         1.1 MB
    cudatoolkit-11.3.1         |       h59b6b97_2       545.3 MB
    cudnn-8.2.1                |       cuda11.3_0       428.9 MB
    ffmpeg-4.2.2               |       he774522_0        17.6 MB
    flatbuffers-2.0.0          |       h6c2663c_0         1.4 MB
    frozenlist-1.4.0           |   py39h2bbff1b_0          46 KB
    gast-0.4.0                 |     pyhd3eb1b0_0          13 KB
    giflib-5.2.1               |       h8cc25b3_3          88 KB
    google-auth-2.6.0          |     pyhd3eb1b0_0          83 KB
    google-auth-oauthlib-0.4.1 |             py_2          20 KB
    google-pasta-0.2.0         |     pyhd3eb1b0_0          46 KB
    grpcio-1.42.0              |   py39hc60d5dd_0         1.9 MB
    h5py-3.9.0                 |   py39hfc34f40_0         914 KB
    hdf5-1.12.1                |       h51c971a_3        11.9 MB
    icc_rt-2022.1.0            |       h6049295_2         6.5 MB
    icu-68.1                   |       h6c2663c_0        11.0 MB
    idna-3.4                   |   py39haa95532_0          93 KB
    importlib-metadata-7.0.1   |   py39haa95532_0          41 KB
    intel-openmp-2021.4.0      |    haa95532_3556         2.2 MB
    joblib-1.2.0               |   py39haa95532_0         388 KB
    jpeg-9e                    |       h2bbff1b_1         320 KB
    keras-preprocessing-1.1.2  |     pyhd3eb1b0_0          35 KB
    libcurl-8.5.0              |       h86230a5_0         343 KB
    libpng-1.6.39              |       h8cc25b3_0         369 KB
    libprotobuf-3.17.2         |       h23ce68f_1         1.9 MB
    libssh2-1.10.0             |       hcd4344a_2         236 KB
    markdown-3.4.1             |   py39haa95532_0         148 KB
    markupsafe-2.1.3           |   py39h2bbff1b_0          25 KB
    mkl-2021.4.0               |     haa95532_640       114.9 MB
    mkl-service-2.4.0          |   py39h2bbff1b_0          51 KB
    mkl_fft-1.3.1              |   py39h277e83a_0         139 KB
    mkl_random-1.2.2           |   py39hf11a4ad_0         225 KB
    multidict-6.0.4            |   py39h2bbff1b_0          50 KB
    numpy-1.23.1               |   py39h7a0a035_0          10 KB
    numpy-base-1.23.1          |   py39hca35cd5_0         5.0 MB
    oauthlib-3.2.2             |   py39haa95532_0         209 KB
    openssl-1.1.1w             |       h2bbff1b_0         5.5 MB
    opt_einsum-3.3.0           |     pyhd3eb1b0_1          57 KB
    packaging-23.2             |   py39haa95532_0         148 KB
    pip-23.3.1                 |   py39haa95532_0         2.8 MB
    platformdirs-3.10.0        |   py39haa95532_0          35 KB
    pooch-1.7.0                |   py39haa95532_0          84 KB
    protobuf-3.17.2            |   py39hd77b12b_0         254 KB
    pyasn1-0.4.8               |     pyhd3eb1b0_0          54 KB
    pyasn1-modules-0.2.8       |             py_0          72 KB
    pyjwt-2.4.0                |   py39haa95532_0          38 KB
    pyopenssl-23.2.0           |   py39haa95532_0          96 KB
    pysocks-1.7.1              |   py39haa95532_0          55 KB
    python-3.9.18              |       h6244533_0        19.4 MB
    python-flatbuffers-1.12    |     pyhd3eb1b0_0          24 KB
    requests-2.31.0            |   py39haa95532_1          98 KB
    requests-oauthlib-1.3.0    |             py_0          23 KB
    rsa-4.7.2                  |     pyhd3eb1b0_1          28 KB
    scikit-learn-1.3.0         |   py39h4ed8f06_1         7.0 MB
    scipy-1.10.1               |   py39h321e85e_0        18.7 MB
    setuptools-68.2.2          |   py39haa95532_0         933 KB
    six-1.16.0                 |     pyhd3eb1b0_1          18 KB
    snappy-1.1.10              |       h6c2663c_1          92 KB
    sqlite-3.41.2              |       h2bbff1b_0         894 KB
    tensorboard-2.6.0          |             py_1         4.9 MB
    tensorboard-data-server-0.6.1|   py39haa95532_0          17 KB
    tensorboard-plugin-wit-1.8.1|   py39haa95532_0         671 KB
    tensorflow-2.6.0           |gpu_py39he88c5ba_0           4 KB
    tensorflow-base-2.6.0      |gpu_py39hb3da07e_0       203.8 MB
    tensorflow-estimator-2.6.0 |     pyh7b7c402_0         267 KB
    tensorflow-gpu-2.6.0       |       h17022bd_0           3 KB
    termcolor-2.1.0            |   py39haa95532_0          12 KB
    threadpoolctl-2.2.0        |     pyh0d69192_0          16 KB
    typing_extensions-4.9.0    |   py39haa95532_1          54 KB
    urllib3-2.1.0              |   py39haa95532_1         154 KB
    vs2015_runtime-14.27.29016 |       h5e58377_2        1007 KB
    werkzeug-2.3.8             |   py39haa95532_0         345 KB
    wheel-0.35.1               |     pyhd3eb1b0_0          38 KB
    win_inet_pton-1.1.0        |   py39haa95532_0          35 KB
    wrapt-1.14.1               |   py39h2bbff1b_0          49 KB
    yarl-1.9.3                 |   py39h2bbff1b_0         109 KB
    zipp-3.17.0                |   py39haa95532_0          23 KB
    zlib-1.2.13                |       h8cc25b3_0         113 KB
    ------------------------------------------------------------
                                           Total:        1.39 GB

The following NEW packages will be INSTALLED:

  _tflow_select      pkgs/main/win-64::_tflow_select-2.1.0-gpu
  abseil-cpp         pkgs/main/win-64::abseil-cpp-20210324.2-hd77b12b_0
  absl-py            pkgs/main/win-64::absl-py-1.4.0-py39haa95532_0
  aiohttp            pkgs/main/win-64::aiohttp-3.9.3-py39h2bbff1b_0
  aiosignal          pkgs/main/noarch::aiosignal-1.2.0-pyhd3eb1b0_0
  astor              pkgs/main/win-64::astor-0.8.1-py39haa95532_0
  astunparse         pkgs/main/noarch::astunparse-1.6.3-py_0
  async-timeout      pkgs/main/win-64::async-timeout-4.0.3-py39haa95532_0
  attrs              pkgs/main/win-64::attrs-23.1.0-py39haa95532_0
  blas               pkgs/main/win-64::blas-1.0-mkl
  blinker            pkgs/main/win-64::blinker-1.6.2-py39haa95532_0
  brotli-python      pkgs/main/win-64::brotli-python-1.0.9-py39hd77b12b_7
  ca-certificates    pkgs/main/win-64::ca-certificates-2024.3.11-haa95532_0
  cachetools         pkgs/main/noarch::cachetools-4.2.2-pyhd3eb1b0_0
  certifi            pkgs/main/win-64::certifi-2024.2.2-py39haa95532_0
  cffi               pkgs/main/win-64::cffi-1.16.0-py39h2bbff1b_0
  charset-normalizer pkgs/main/noarch::charset-normalizer-2.0.4-pyhd3eb1b0_0
  click              pkgs/main/win-64::click-8.1.7-py39haa95532_0
  colorama           pkgs/main/win-64::colorama-0.4.6-py39haa95532_0
  cryptography       pkgs/main/win-64::cryptography-41.0.3-py39h3438e0d_0
  cudatoolkit        pkgs/main/win-64::cudatoolkit-11.3.1-h59b6b97_2
  cudnn              pkgs/main/win-64::cudnn-8.2.1-cuda11.3_0
  ffmpeg             pkgs/main/win-64::ffmpeg-4.2.2-he774522_0
  flatbuffers        pkgs/main/win-64::flatbuffers-2.0.0-h6c2663c_0
  frozenlist         pkgs/main/win-64::frozenlist-1.4.0-py39h2bbff1b_0
  gast               pkgs/main/noarch::gast-0.4.0-pyhd3eb1b0_0
  giflib             pkgs/main/win-64::giflib-5.2.1-h8cc25b3_3
  google-auth        pkgs/main/noarch::google-auth-2.6.0-pyhd3eb1b0_0
  google-auth-oauth~ pkgs/main/noarch::google-auth-oauthlib-0.4.1-py_2
  google-pasta       pkgs/main/noarch::google-pasta-0.2.0-pyhd3eb1b0_0
  grpcio             pkgs/main/win-64::grpcio-1.42.0-py39hc60d5dd_0
  h5py               pkgs/main/win-64::h5py-3.9.0-py39hfc34f40_0
  hdf5               pkgs/main/win-64::hdf5-1.12.1-h51c971a_3
  icc_rt             pkgs/main/win-64::icc_rt-2022.1.0-h6049295_2
  icu                pkgs/main/win-64::icu-68.1-h6c2663c_0
  idna               pkgs/main/win-64::idna-3.4-py39haa95532_0
  importlib-metadata pkgs/main/win-64::importlib-metadata-7.0.1-py39haa95532_0
  intel-openmp       pkgs/main/win-64::intel-openmp-2021.4.0-haa95532_3556
  joblib             pkgs/main/win-64::joblib-1.2.0-py39haa95532_0
  jpeg               pkgs/main/win-64::jpeg-9e-h2bbff1b_1
  keras-preprocessi~ pkgs/main/noarch::keras-preprocessing-1.1.2-pyhd3eb1b0_0
  libcurl            pkgs/main/win-64::libcurl-8.5.0-h86230a5_0
  libpng             pkgs/main/win-64::libpng-1.6.39-h8cc25b3_0
  libprotobuf        pkgs/main/win-64::libprotobuf-3.17.2-h23ce68f_1
  libssh2            pkgs/main/win-64::libssh2-1.10.0-hcd4344a_2
  markdown           pkgs/main/win-64::markdown-3.4.1-py39haa95532_0
  markupsafe         pkgs/main/win-64::markupsafe-2.1.3-py39h2bbff1b_0
  mkl                pkgs/main/win-64::mkl-2021.4.0-haa95532_640
  mkl-service        pkgs/main/win-64::mkl-service-2.4.0-py39h2bbff1b_0
  mkl_fft            pkgs/main/win-64::mkl_fft-1.3.1-py39h277e83a_0
  mkl_random         pkgs/main/win-64::mkl_random-1.2.2-py39hf11a4ad_0
  multidict          pkgs/main/win-64::multidict-6.0.4-py39h2bbff1b_0
  numpy              pkgs/main/win-64::numpy-1.23.1-py39h7a0a035_0
  numpy-base         pkgs/main/win-64::numpy-base-1.23.1-py39hca35cd5_0
  oauthlib           pkgs/main/win-64::oauthlib-3.2.2-py39haa95532_0
  openssl            pkgs/main/win-64::openssl-1.1.1w-h2bbff1b_0
  opt_einsum         pkgs/main/noarch::opt_einsum-3.3.0-pyhd3eb1b0_1
  packaging          pkgs/main/win-64::packaging-23.2-py39haa95532_0
  pip                pkgs/main/win-64::pip-23.3.1-py39haa95532_0
  platformdirs       pkgs/main/win-64::platformdirs-3.10.0-py39haa95532_0
  pooch              pkgs/main/win-64::pooch-1.7.0-py39haa95532_0
  protobuf           pkgs/main/win-64::protobuf-3.17.2-py39hd77b12b_0
  pyasn1             pkgs/main/noarch::pyasn1-0.4.8-pyhd3eb1b0_0
  pyasn1-modules     pkgs/main/noarch::pyasn1-modules-0.2.8-py_0
  pycparser          pkgs/main/noarch::pycparser-2.21-pyhd3eb1b0_0
  pyjwt              pkgs/main/win-64::pyjwt-2.4.0-py39haa95532_0
  pyopenssl          pkgs/main/win-64::pyopenssl-23.2.0-py39haa95532_0
  pysocks            pkgs/main/win-64::pysocks-1.7.1-py39haa95532_0
  python             pkgs/main/win-64::python-3.9.18-h6244533_0
  python-flatbuffers pkgs/main/noarch::python-flatbuffers-1.12-pyhd3eb1b0_0
  requests           pkgs/main/win-64::requests-2.31.0-py39haa95532_1
  requests-oauthlib  pkgs/main/noarch::requests-oauthlib-1.3.0-py_0
  rsa                pkgs/main/noarch::rsa-4.7.2-pyhd3eb1b0_1
  scikit-learn       pkgs/main/win-64::scikit-learn-1.3.0-py39h4ed8f06_1
  scipy              pkgs/main/win-64::scipy-1.10.1-py39h321e85e_0
  setuptools         pkgs/main/win-64::setuptools-68.2.2-py39haa95532_0
  six                pkgs/main/noarch::six-1.16.0-pyhd3eb1b0_1
  snappy             pkgs/main/win-64::snappy-1.1.10-h6c2663c_1
  sqlite             pkgs/main/win-64::sqlite-3.41.2-h2bbff1b_0
  tensorboard        pkgs/main/noarch::tensorboard-2.6.0-py_1
  tensorboard-data-~ pkgs/main/win-64::tensorboard-data-server-0.6.1-py39haa95532_0
  tensorboard-plugi~ pkgs/main/win-64::tensorboard-plugin-wit-1.8.1-py39haa95532_0
  tensorflow         pkgs/main/win-64::tensorflow-2.6.0-gpu_py39he88c5ba_0
  tensorflow-base    pkgs/main/win-64::tensorflow-base-2.6.0-gpu_py39hb3da07e_0
  tensorflow-estima~ pkgs/main/noarch::tensorflow-estimator-2.6.0-pyh7b7c402_0
  tensorflow-gpu     pkgs/main/win-64::tensorflow-gpu-2.6.0-h17022bd_0
  termcolor          pkgs/main/win-64::termcolor-2.1.0-py39haa95532_0
  threadpoolctl      pkgs/main/noarch::threadpoolctl-2.2.0-pyh0d69192_0
  trex               trexing/win-64::trex-1.1.9-g4ce4be1_0
  typing_extensions  pkgs/main/win-64::typing_extensions-4.9.0-py39haa95532_1
  tzdata             pkgs/main/noarch::tzdata-2024a-h04d1e81_0
  urllib3            pkgs/main/win-64::urllib3-2.1.0-py39haa95532_1
  vc                 pkgs/main/win-64::vc-14.2-h21ff451_1
  vs2015_runtime     pkgs/main/win-64::vs2015_runtime-14.27.29016-h5e58377_2
  werkzeug           pkgs/main/win-64::werkzeug-2.3.8-py39haa95532_0
  wheel              pkgs/main/noarch::wheel-0.35.1-pyhd3eb1b0_0
  win_inet_pton      pkgs/main/win-64::win_inet_pton-1.1.0-py39haa95532_0
  wrapt              pkgs/main/win-64::wrapt-1.14.1-py39h2bbff1b_0
  yarl               pkgs/main/win-64::yarl-1.9.3-py39h2bbff1b_0
  zipp               pkgs/main/win-64::zipp-3.17.0-py39haa95532_0
  zlib               pkgs/main/win-64::zlib-1.2.13-h8cc25b3_0


Proceed ([y]/n)? y

cudatoolkit-11.3.1   | 545.3 MB  | #############6                                                               |  18%
cudnn-8.2.1          | 428.9 MB  | ###############7                                                             |  21%
tensorflow-base-2.6. | 203.8 MB  | #############################################                                |  59%
mkl-2021.4.0         | 114.9 MB  | ########################################################7                    |  75%
tensorflow-base-2.6. | 203.8 MB  | ###########################################9                                 |  58%
mkl-2021.4.0         | 114.9 MB  | ######################################################9                      |  72%
python-3.9.18        | 19.4 MB   | ############################################################################ | 100%
scipy-1.10.1         | 18.7 MB   | ############################################################################ | 100%
ffmpeg-4.2.2         | 17.6 MB   | ############################################################################ | 100%
hdf5-1.12.1          | 11.9 MB   | ############################################################################ | 100%
icu-68.1             | 11.0 MB   | ############################################################################ | 100%
openssl-1.1.1w       | 5.5 MB    | ########################################################################1    |  95%


InvalidArchiveError("Error with archive C:\\Users\\mr630\\AppData\\Local\\mambaforge\\pkgs\\tensorflow-base-2.6.0-gpu_py39hb3da07e_0.conda.  You probably need to delete and re-download or re-create this file.  Message was:\n\nfailed with error: [Errno 2] No such file or directory: 'C:\\\\Users\\\\mr630\\\\AppData\\\\Local\\\\mambaforge\\\\pkgs\\\\tensorflow-base-2.6.0-gpu_py39hb3da07e_0\\\\Lib\\\\site-packages\\\\tensorflow\\\\include\\\\external\\\\cudnn_frontend_archive\\\\_virtual_includes\\\\cudnn_frontend\\\\third_party\\\\cudnn_frontend\\\\include\\\\cudnn_frontend_EngineConfigGenerator.h'")

When I follow along in the pkgs folder during the installation, tensorflow-base-2.6.0 is being created, but then it suddenly disappears - no clue why. I'm continuing to try some debugging. (Still works with pixi without any issues).

I think no matter what, it would be good to add a section to the docs about this sort of issue with some general things to attempt. Conda channels is a black box to lots of folks, so having some things to try is good I think.

@roaldarbol
Copy link

roaldarbol commented Mar 21, 2024

Oh, you're changing from Tensorflow to PyTorch in the new version?! AMAZING!!! So many headaches from that stuff! (that installation works just fine!)

@mooch443
Copy link
Owner

I see, you should clean your caches then I reckon. conda clean -a could do the trick - but this is probably just because you haven't done this in a while / they did something illegal, like uploading under the same name?

In any case, as far as I understand defaults is a combination of the largest and best maintained channels. It is a specific set of channels, not just anaconda or whatever is your "default" setting. I cannot find specific documentation on that, but afaik it includes main (https://anaconda.org/main), r and free. The output you provided seems to suggest that it works as intended. Just that the packages in your cache are broken.

@mooch443
Copy link
Owner

Oh, you're changing from Tensorflow to PyTorch in the new version?! AMAZING!!! So many headaches from that stuff! (that installation works just fine!)

I am using both in fact. Not an easy feat. I would like to port all the previous code to pytorch, but why change a running system. The difference is installing it via pip instead of conda (this process is hidden to the end user). It also means you have to use a fix numpy version.

@roaldarbol
Copy link

roaldarbol commented Mar 21, 2024

I see, you should clean your caches then I reckon. conda clean -a could do the trick - but this is probably just because you haven't done this in a while / they did something illegal, like uploading under the same name?

Absolutely right, I've never done that in fact! Didn't even know that was a thing. However, the issue remains the same afterwards. Maybe tensorflow-base-2.6.0-gpu is being overridden by tensorflow-gpu?

EDIT: Installing without tensorflow-gpu works, then subsequently installing tensorflow-gpu reveals that it modifies some of the dependencies. I can't quite figure out what's going on, and unfortunately don't have more time for debugging today, but can try again tomorrow. Here's the output of attempting to install tensorflow-gpu afterwards:

(trex_normal_2) C:\>conda install tensorflow-gpu
Channels:
 - conda-forge
 - defaults
 - trexing
Platform: win-64
Collecting package metadata (repodata.json): done
Solving environment: done

## Package Plan ##

  environment location: C:\Users\mr630\AppData\Local\mambaforge\envs\trex_normal_2

  added / updated specs:
    - tensorflow-gpu


The following packages will be downloaded:

    package                    |            build
    ---------------------------|-----------------
    certifi-2024.2.2           |     pyhd8ed1ab_0         157 KB  conda-forge
    cudatoolkit-11.3.1         |      hf2f0253_13       610.8 MB  conda-forge
    libprotobuf-3.17.2         |       h7755175_1         2.3 MB  conda-forge
    protobuf-3.17.2            |   py39h415ef7b_0         263 KB  conda-forge
    python_abi-3.9             |           2_cp39           4 KB  conda-forge
    tensorflow-base-2.6.0      |gpu_py39hb3da07e_0       203.8 MB
    vs2015_runtime-14.38.33130 |      hcb4865c_18          17 KB  conda-forge
    ------------------------------------------------------------
                                           Total:       817.4 MB

The following NEW packages will be INSTALLED:

  python_abi         conda-forge/win-64::python_abi-3.9-2_cp39
  tensorflow-gpu     pkgs/main/win-64::tensorflow-gpu-2.6.0-h17022bd_0
  ucrt               conda-forge/win-64::ucrt-10.0.22621.0-h57928b3_0
  vc14_runtime       conda-forge/win-64::vc14_runtime-14.38.33130-h82b7239_18

The following packages will be UPDATED:

  libprotobuf        pkgs/main::libprotobuf-3.14.0-h23ce68~ --> conda-forge::libprotobuf-3.17.2-h7755175_1
  protobuf           pkgs/main::protobuf-3.14.0-py39hd77b1~ --> conda-forge::protobuf-3.17.2-py39h415ef7b_0
  vs2015_runtime     pkgs/main::vs2015_runtime-14.27.29016~ --> conda-forge::vs2015_runtime-14.38.33130-hcb4865c_18

The following packages will be SUPERSEDED by a higher-priority channel:

  certifi            pkgs/main/win-64::certifi-2024.2.2-py~ --> conda-forge/noarch::certifi-2024.2.2-pyhd8ed1ab_0
  cudatoolkit        pkgs/main::cudatoolkit-11.8.0-hd77b12~ --> conda-forge::cudatoolkit-11.3.1-hf2f0253_13
  cudnn                  pkgs/main::cudnn-8.9.2.26-cuda11_0 --> conda-forge::cudnn-8.2.1.32-h754d62a_0
  openssl              pkgs/main::openssl-1.1.1w-h2bbff1b_0 --> conda-forge::openssl-1.1.1w-hcfcfb64_0

The following packages will be DOWNGRADED:

  _tflow_select                                   2.3.0-mkl --> 2.1.0-gpu
  tensorflow                       2.6.0-mkl_py39h31650da_0 --> 2.6.0-gpu_py39he88c5ba_0
  tensorflow-base                  2.6.0-mkl_py39h9201259_0 --> 2.6.0-gpu_py39hb3da07e_0


Proceed ([y]/n)? y


Downloading and Extracting Packages:


InvalidArchiveError("Error with archive C:\\Users\\mr630\\AppData\\Local\\mambaforge\\pkgs\\tensorflow-base-2.6.0-gpu_py39hb3da07e_0.conda.  You probably need to delete and re-download or re-create this file.  Message was:\n\nfailed with error: [Errno 2] No such file or directory: 'C:\\\\Users\\\\mr630\\\\AppData\\\\Local\\\\mambaforge\\\\pkgs\\\\tensorflow-base-2.6.0-gpu_py39hb3da07e_0\\\\Lib\\\\site-packages\\\\tensorflow\\\\include\\\\external\\\\cudnn_frontend_archive\\\\_virtual_includes\\\\cudnn_frontend\\\\third_party\\\\cudnn_frontend\\\\include\\\\cudnn_frontend_EngineConfigGenerator.h'")

@roaldarbol
Copy link

In any case, as far as I understand defaults is a combination of the largest and best maintained channels. It is a specific set of channels, not just anaconda or whatever is your "default" setting. I cannot find specific documentation on that, but afaik it includes main (https://anaconda.org/main), r and free. The output you provided seems to suggest that it works as intended. Just that the packages in your cache are broken.

I think at least the r channel is no longer getting support - I can't remember where I read it, but it also seems the packages are no longer being updated there. I think it just confuses having implicit channels. It's also one of the reasons prefix-dev (who developed mamba and now pixi) have decided on only having explicit channel names. In pixi they default to conda-forge I think, but every channel has to be named.

@roaldarbol
Copy link

Just reiterating, I think some of this information would be really good to place in docs - maybe have a Troubleshooting section inside of Installation.

@mooch443
Copy link
Owner

(trex_normal_2) C:>conda install tensorflow-gpu
Channels:

  • conda-forge
  • defaults
  • trexing

As you can see, conda-forge is injected here again. This is the reason it tries some weird stuff. You shouldn't mix main/conda-forge channels, this usually breaks stuff!

Regarding the tensorflow-gpu, maybe there is generally some stuff going on with your channels. If you can, reinstall conda and use miniconda - or just install an additional miniconda on your system in a different location. I suspect this might resolve issues.

@mooch443
Copy link
Owner

Just reiterating, I think some of this information would be really good to place in docs - maybe have a Troubleshooting section inside of Installation.

I also think this is mainly a conda issue right now, but if there will be similar issues with the new version then I'll add them.

@mooch443
Copy link
Owner

In any case, as far as I understand defaults is a combination of the largest and best maintained channels. It is a specific set of channels, not just anaconda or whatever is your "default" setting. I cannot find specific documentation on that, but afaik it includes main (https://anaconda.org/main), r and free. The output you provided seems to suggest that it works as intended. Just that the packages in your cache are broken.

I think at least the r channel is no longer getting support - I can't remember where I read it, but it also seems the packages are no longer being updated there. I think it just confuses having implicit channels. It's also one of the reasons prefix-dev (who developed mamba and now pixi) have decided on only having explicit channel names. In pixi they default to conda-forge I think, but every channel has to be named.

Okay, I was just mentioning that for completeness. The r channel does not really do anything here. I compiled the software using the defaults channels, so if you switch to conda-forge when installing it - even if you get the packages that it wants and it installs "successfully" - it won't work properly in my experience. On Windows it might be more tolerant, but on Linux you won't get it working at all in fact.

@roaldarbol
Copy link

(trex_normal_2) C:>conda install tensorflow-gpu
Channels:

  • conda-forge
  • defaults
  • trexing

As you can see, conda-forge is injected here again. This is the reason it tries some weird stuff. You shouldn't mix main/conda-forge channels, this usually breaks stuff!

Regarding the tensorflow-gpu, maybe there is generally some stuff going on with your channels. If you can, reinstall conda and use miniconda - or just install an additional miniconda on your system in a different location. I suspect this might resolve issues.

This env was installed with your line, so it's injecting conda-forge even with override-channels. And conda-forge is not part of my defaults:

(base) C:\WINDOWS\system32>conda config --show default_channels
default_channels:
  - https://repo.anaconda.com/pkgs/main
  - https://repo.anaconda.com/pkgs/r
  - https://repo.anaconda.com/pkgs/msys2

But even when conda-forge is not included, the result is the same:

(base) C:\WINDOWS\system32>conda create -n trex_normal --override-channels -c trexing -c defaults trex numpy=1.23.1
Channels:
 - trexing
 - defaults
Platform: win-64
Collecting package metadata (repodata.json): done
Solving environment: done

## Package Plan ##

  environment location: C:\Users\mr630\AppData\Local\mambaforge\envs\trex_normal

  added / updated specs:
    - numpy=1.23.1
    - trex


The following NEW packages will be INSTALLED:

  _tflow_select      pkgs/main/win-64::_tflow_select-2.3.0-mkl
  abseil-cpp         pkgs/main/win-64::abseil-cpp-20210324.2-hd77b12b_0
  absl-py            pkgs/main/win-64::absl-py-1.4.0-py39haa95532_0
  aiohttp            pkgs/main/win-64::aiohttp-3.9.3-py39h2bbff1b_0
  aiosignal          pkgs/main/noarch::aiosignal-1.2.0-pyhd3eb1b0_0
  astor              pkgs/main/win-64::astor-0.8.1-py39haa95532_0
  astunparse         pkgs/main/noarch::astunparse-1.6.3-py_0
  async-timeout      pkgs/main/win-64::async-timeout-4.0.3-py39haa95532_0
  attrs              pkgs/main/win-64::attrs-23.1.0-py39haa95532_0
  blas               pkgs/main/win-64::blas-1.0-mkl
  blinker            pkgs/main/win-64::blinker-1.6.2-py39haa95532_0
  brotli-python      pkgs/main/win-64::brotli-python-1.0.9-py39hd77b12b_7
  ca-certificates    pkgs/main/win-64::ca-certificates-2024.3.11-haa95532_0
  cachetools         pkgs/main/noarch::cachetools-4.2.2-pyhd3eb1b0_0
  certifi            pkgs/main/win-64::certifi-2024.2.2-py39haa95532_0
  cffi               pkgs/main/win-64::cffi-1.16.0-py39h2bbff1b_0
  charset-normalizer pkgs/main/noarch::charset-normalizer-2.0.4-pyhd3eb1b0_0
  click              pkgs/main/win-64::click-8.1.7-py39haa95532_0
  colorama           pkgs/main/win-64::colorama-0.4.6-py39haa95532_0
  cryptography       pkgs/main/win-64::cryptography-41.0.3-py39h3438e0d_0
  cudatoolkit        pkgs/main/win-64::cudatoolkit-11.8.0-hd77b12b_0
  cudnn              pkgs/main/win-64::cudnn-8.9.2.26-cuda11_0
  ffmpeg             pkgs/main/win-64::ffmpeg-4.2.2-he774522_0
  flatbuffers        pkgs/main/win-64::flatbuffers-2.0.0-h6c2663c_0
  frozenlist         pkgs/main/win-64::frozenlist-1.4.0-py39h2bbff1b_0
  gast               pkgs/main/noarch::gast-0.4.0-pyhd3eb1b0_0
  giflib             pkgs/main/win-64::giflib-5.2.1-h8cc25b3_3
  google-auth        pkgs/main/noarch::google-auth-2.6.0-pyhd3eb1b0_0
  google-auth-oauth~ pkgs/main/noarch::google-auth-oauthlib-0.4.1-py_2
  google-pasta       pkgs/main/noarch::google-pasta-0.2.0-pyhd3eb1b0_0
  grpcio             pkgs/main/win-64::grpcio-1.42.0-py39hc60d5dd_0
  h5py               pkgs/main/win-64::h5py-3.9.0-py39hfc34f40_0
  hdf5               pkgs/main/win-64::hdf5-1.12.1-h51c971a_3
  icc_rt             pkgs/main/win-64::icc_rt-2022.1.0-h6049295_2
  icu                pkgs/main/win-64::icu-68.1-h6c2663c_0
  idna               pkgs/main/win-64::idna-3.4-py39haa95532_0
  importlib-metadata pkgs/main/win-64::importlib-metadata-7.0.1-py39haa95532_0
  intel-openmp       pkgs/main/win-64::intel-openmp-2021.4.0-haa95532_3556
  joblib             pkgs/main/win-64::joblib-1.2.0-py39haa95532_0
  jpeg               pkgs/main/win-64::jpeg-9e-h2bbff1b_1
  keras-preprocessi~ pkgs/main/noarch::keras-preprocessing-1.1.2-pyhd3eb1b0_0
  libcurl            pkgs/main/win-64::libcurl-8.5.0-h86230a5_0
  libpng             pkgs/main/win-64::libpng-1.6.39-h8cc25b3_0
  libprotobuf        pkgs/main/win-64::libprotobuf-3.14.0-h23ce68f_0
  libssh2            pkgs/main/win-64::libssh2-1.10.0-hcd4344a_2
  markdown           pkgs/main/win-64::markdown-3.4.1-py39haa95532_0
  markupsafe         pkgs/main/win-64::markupsafe-2.1.3-py39h2bbff1b_0
  mkl                pkgs/main/win-64::mkl-2021.4.0-haa95532_640
  mkl-service        pkgs/main/win-64::mkl-service-2.4.0-py39h2bbff1b_0
  mkl_fft            pkgs/main/win-64::mkl_fft-1.3.1-py39h277e83a_0
  mkl_random         pkgs/main/win-64::mkl_random-1.2.2-py39hf11a4ad_0
  multidict          pkgs/main/win-64::multidict-6.0.4-py39h2bbff1b_0
  numpy              pkgs/main/win-64::numpy-1.23.1-py39h7a0a035_0
  numpy-base         pkgs/main/win-64::numpy-base-1.23.1-py39hca35cd5_0
  oauthlib           pkgs/main/win-64::oauthlib-3.2.2-py39haa95532_0
  openssl            pkgs/main/win-64::openssl-1.1.1w-h2bbff1b_0
  opt_einsum         pkgs/main/noarch::opt_einsum-3.3.0-pyhd3eb1b0_1
  packaging          pkgs/main/win-64::packaging-23.2-py39haa95532_0
  pip                pkgs/main/win-64::pip-23.3.1-py39haa95532_0
  platformdirs       pkgs/main/win-64::platformdirs-3.10.0-py39haa95532_0
  pooch              pkgs/main/win-64::pooch-1.7.0-py39haa95532_0
  protobuf           pkgs/main/win-64::protobuf-3.14.0-py39hd77b12b_1
  pyasn1             pkgs/main/noarch::pyasn1-0.4.8-pyhd3eb1b0_0
  pyasn1-modules     pkgs/main/noarch::pyasn1-modules-0.2.8-py_0
  pycparser          pkgs/main/noarch::pycparser-2.21-pyhd3eb1b0_0
  pyjwt              pkgs/main/win-64::pyjwt-2.4.0-py39haa95532_0
  pyopenssl          pkgs/main/win-64::pyopenssl-23.2.0-py39haa95532_0
  pysocks            pkgs/main/win-64::pysocks-1.7.1-py39haa95532_0
  python             pkgs/main/win-64::python-3.9.18-h6244533_0
  python-flatbuffers pkgs/main/noarch::python-flatbuffers-1.12-pyhd3eb1b0_0
  requests           pkgs/main/win-64::requests-2.31.0-py39haa95532_1
  requests-oauthlib  pkgs/main/noarch::requests-oauthlib-1.3.0-py_0
  rsa                pkgs/main/noarch::rsa-4.7.2-pyhd3eb1b0_1
  scikit-learn       pkgs/main/win-64::scikit-learn-1.3.0-py39h4ed8f06_1
  scipy              pkgs/main/win-64::scipy-1.10.1-py39h321e85e_0
  setuptools         pkgs/main/win-64::setuptools-68.2.2-py39haa95532_0
  six                pkgs/main/noarch::six-1.16.0-pyhd3eb1b0_1
  snappy             pkgs/main/win-64::snappy-1.1.10-h6c2663c_1
  sqlite             pkgs/main/win-64::sqlite-3.41.2-h2bbff1b_0
  tensorboard        pkgs/main/noarch::tensorboard-2.6.0-py_1
  tensorboard-data-~ pkgs/main/win-64::tensorboard-data-server-0.6.1-py39haa95532_0
  tensorboard-plugi~ pkgs/main/win-64::tensorboard-plugin-wit-1.8.1-py39haa95532_0
  tensorflow         pkgs/main/win-64::tensorflow-2.6.0-mkl_py39h31650da_0
  tensorflow-base    pkgs/main/win-64::tensorflow-base-2.6.0-mkl_py39h9201259_0
  tensorflow-estima~ pkgs/main/noarch::tensorflow-estimator-2.6.0-pyh7b7c402_0
  termcolor          pkgs/main/win-64::termcolor-2.1.0-py39haa95532_0
  threadpoolctl      pkgs/main/noarch::threadpoolctl-2.2.0-pyh0d69192_0
  trex               trexing/win-64::trex-1.1.9-g4ce4be1_0
  typing_extensions  pkgs/main/win-64::typing_extensions-4.9.0-py39haa95532_1
  tzdata             pkgs/main/noarch::tzdata-2024a-h04d1e81_0
  urllib3            pkgs/main/win-64::urllib3-2.1.0-py39haa95532_1
  vc                 pkgs/main/win-64::vc-14.2-h21ff451_1
  vs2015_runtime     pkgs/main/win-64::vs2015_runtime-14.27.29016-h5e58377_2
  werkzeug           pkgs/main/win-64::werkzeug-2.3.8-py39haa95532_0
  wheel              pkgs/main/noarch::wheel-0.35.1-pyhd3eb1b0_0
  win_inet_pton      pkgs/main/win-64::win_inet_pton-1.1.0-py39haa95532_0
  wrapt              pkgs/main/win-64::wrapt-1.14.1-py39h2bbff1b_0
  yarl               pkgs/main/win-64::yarl-1.9.3-py39h2bbff1b_0
  zipp               pkgs/main/win-64::zipp-3.17.0-py39haa95532_0
  zlib               pkgs/main/win-64::zlib-1.2.13-h8cc25b3_0


Proceed ([y]/n)? y


Downloading and Extracting Packages:

Preparing transaction: done
Verifying transaction: done
Executing transaction: done
#
# To activate this environment, use
#
#     $ conda activate trex_normal
#
# To deactivate an active environment, use
#
#     $ conda deactivate


(base) C:\WINDOWS\system32>conda activate trex_normal

(trex_normal) C:\WINDOWS\system32>conda install tensorflow-gpu
Channels:
 - conda-forge
 - defaults
 - trexing
Platform: win-64
Collecting package metadata (repodata.json): failed

CondaError: KeyboardInterrupt

Terminate batch job (Y/N)? y

(trex_normal) C:\WINDOWS\system32>conda install --override-channels -c trexing -c defaults tensorflow-gpu
Channels:
 - trexing
 - defaults
Platform: win-64
Collecting package metadata (repodata.json): done
Solving environment: done

## Package Plan ##

  environment location: C:\Users\mr630\AppData\Local\mambaforge\envs\trex_normal

  added / updated specs:
    - tensorflow-gpu


The following packages will be downloaded:

    package                    |            build
    ---------------------------|-----------------
    tensorflow-base-2.6.0      |gpu_py39hb3da07e_0       203.8 MB
    ------------------------------------------------------------
                                           Total:       203.8 MB

The following NEW packages will be INSTALLED:

  tensorflow-gpu     pkgs/main/win-64::tensorflow-gpu-2.6.0-h17022bd_0

The following packages will be UPDATED:

  libprotobuf                             3.14.0-h23ce68f_0 --> 3.17.2-h23ce68f_1
  protobuf                            3.14.0-py39hd77b12b_1 --> 3.17.2-py39hd77b12b_0

The following packages will be DOWNGRADED:

  _tflow_select                                   2.3.0-mkl --> 2.1.0-gpu
  cudatoolkit                             11.8.0-hd77b12b_0 --> 11.3.1-h59b6b97_2
  cudnn                                   8.9.2.26-cuda11_0 --> 8.2.1-cuda11.3_0
  tensorflow                       2.6.0-mkl_py39h31650da_0 --> 2.6.0-gpu_py39he88c5ba_0
  tensorflow-base                  2.6.0-mkl_py39h9201259_0 --> 2.6.0-gpu_py39hb3da07e_0


Proceed ([y]/n)? y


Downloading and Extracting Packages:


InvalidArchiveError("Error with archive C:\\Users\\mr630\\AppData\\Local\\mambaforge\\pkgs\\tensorflow-base-2.6.0-gpu_py39hb3da07e_0.conda.  You probably need to delete and re-download or re-create this file.  Message was:\n\nfailed with error: [Errno 2] No such file or directory: 'C:\\\\Users\\\\mr630\\\\AppData\\\\Local\\\\mambaforge\\\\pkgs\\\\tensorflow-base-2.6.0-gpu_py39hb3da07e_0\\\\Lib\\\\site-packages\\\\tensorflow\\\\include\\\\external\\\\cudnn_frontend_archive\\\\_virtual_includes\\\\cudnn_frontend\\\\third_party\\\\cudnn_frontend\\\\include\\\\cudnn_frontend_EngineConfigGenerator.h'")

I'll try uninstalling conda tomorrow and try again. Thanks for the ping-pong, I'm sure we'll get to the bottom of it!

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
bug Something isn't working
Projects
None yet
Development

No branches or pull requests

3 participants