oneAPI Registration, Download, Licensing and Installation
Support for Getting Started questions related to download, Installation and licensing for Intel oneAPI Toolkits and software development tools.
1453 Discussions

PyPI manylinux wheels do not reflect minimum system requirements

mayeut
Beginner
573 Views

All Intel oneAPI linux wheels published on PyPI are claiming to be supported on glibc 2.5+ / CentOS 5+ distros by using a manylinux1 platform tag.

This does not reflect the system requirements mentioned in https://www.intel.com/content/www/us/en/developer/articles/system-requirements/intel-oneapi-base-toolkit-system-requirements.html and allows to install such wheel on unsupported platforms with hard to understand error messages when running those. They should fail to install in the first place.

Please update the platform tags used per PEP600: https://peps.python.org/pep-0600/

examples of invalid oneAPI wheels:

should be manylinux2014_x86_64.manylinux_2_17_x86_64:

> python3.9 -m pip download --no-deps intel_cmplr_lib_rt -d wheels
> auditwheel show ./wheels/intel_cmplr_lib_rt-2022.1.0-py2.py3-none-manylinux1_x86_64.whl

intel_cmplr_lib_rt-2022.1.0-py2.py3-none-manylinux1_x86_64.whl is
consistent with the following platform tag: "linux_x86_64".

The wheel references external versioned symbols in these
system-provided shared libraries: librt.so.1 with versions
{'GLIBC_2.2.5'}, libdl.so.2 with versions {'GLIBC_2.2.5'}, libz.so.1
with versions {'ZLIB_1.2.0'}, libgcc_s.so.1 with versions {'GCC_3.0'},
libpthread.so.0 with versions {'GLIBC_2.3.2', 'GLIBC_2.12',
'GLIBC_2.2.5'}, libc.so.6 with versions {'GLIBC_2.8', 'GLIBC_2.6',
'GLIBC_2.7', 'GLIBC_2.11', 'GLIBC_2.15', 'GLIBC_2.3.4', 'GLIBC_2.4',
'GLIBC_2.14', 'GLIBC_2.3.3', 'GLIBC_2.2.5', 'GLIBC_2.3'},
libstdc++.so.6 with versions {'GLIBCXX_3.4'}, libOpenCL.so.1 with
versions {'OPENCL_1.0', 'OPENCL_1.1', 'OPENCL_1.2', 'OPENCL_2.0',
'OPENCL_2.1'}

This constrains the platform tag to "manylinux_2_17_x86_64". In order
to achieve a more compatible tag, you would need to recompile a new
wheel from source on a system with earlier versions of these
libraries, such as a recent manylinux image.

python3.9 -m pip download --no-deps intel_openmp -d wheels
auditwheel show ./wheels/intel_openmp-2022.1.0-py2.py3-none-manylinux1_x86_64.whl

intel_openmp-2022.1.0-py2.py3-none-manylinux1_x86_64.whl is consistent
with the following platform tag: "linux_x86_64".

The wheel references external versioned symbols in these
system-provided shared libraries: libdl.so.2 with versions
{'GLIBC_2.2.5'}, librt.so.1 with versions {'GLIBC_2.2.5'},
libpthread.so.0 with versions {'GLIBC_2.3.2', 'GLIBC_2.12',
'GLIBC_2.3.4', 'GLIBC_2.2.5'}, libc.so.6 with versions {'GLIBC_2.3.2',
'GLIBC_2.3.4', 'GLIBC_2.4', 'GLIBC_2.11', 'GLIBC_2.2.5', 'GLIBC_2.14',
'GLIBC_2.3', 'GLIBC_2.7', 'GLIBC_2.15', 'GLIBC_2.8', 'GLIBC_2.6',
'GLIBC_2.3.3'}, libgcc_s.so.1 with versions {'GCC_3.3', 'GCC_3.0'},
libstdc++.so.6 with versions {'CXXABI_1.3', 'GLIBCXX_3.4'},
libiomp5.so with versions {'VERSION'}, libOpenCL.so.1 with versions
{'OPENCL_1.1', 'OPENCL_1.2', 'OPENCL_2.0', 'OPENCL_1.0',
'OPENCL_2.1'}, libelf.so.1 with versions {'ELFUTILS_1.0',
'ELFUTILS_1.5'}

This constrains the platform tag to "manylinux_2_17_x86_64". In order
to achieve a more compatible tag, you would need to recompile a new
wheel from source on a system with earlier versions of these
libraries, such as a recent manylinux image.

 

should be manylinux2010_x86_64.manylinux_2_12_x86_64:

> python3.9 -m pip download --no-deps intel_fortran_rt -d wheels
> auditwheel show ./wheels/intel_fortran_rt-2022.1.0-py2.py3-none-manylinux1_x86_64.whl

intel_fortran_rt-2022.1.0-py2.py3-none-manylinux1_x86_64.whl is
consistent with the following platform tag: "linux_x86_64".

The wheel references external versioned symbols in these
system-provided shared libraries: libc.so.6 with versions
{'GLIBC_2.2.5', 'GLIBC_2.3', 'GLIBC_2.7', 'GLIBC_2.4', 'GLIBC_2.3.2',
'GLIBC_2.3.4', 'GLIBC_2.8'}

This constrains the platform tag to "manylinux_2_12_x86_64". In order
to achieve a more compatible tag, you would need to recompile a new
wheel from source on a system with earlier versions of these
libraries, such as a recent manylinux image.

 

should be manylinux_2_27_x86_64:

python3.9 -m pip download --no-deps intel_opencl_rt -d wheels
auditwheel show ./wheels/intel_opencl_rt-2022.1.0-py2.py3-none-manylinux1_x86_64.whl

intel_opencl_rt-2022.1.0-py2.py3-none-manylinux1_x86_64.whl is
consistent with the following platform tag: "linux_x86_64".

The wheel references external versioned symbols in these
system-provided shared libraries: libpthread.so.0 with versions
{'GLIBC_2.2.5', 'GLIBC_2.12', 'GLIBC_2.3.2'}, libdl.so.2 with versions
{'GLIBC_2.2.5'}, librt.so.1 with versions {'GLIBC_2.2.5'}, libz.so.1
with versions {'ZLIB_1.2.0'}, libstdc++.so.6 with versions
{'GLIBCXX_3.4.19', 'CXXABI_1.3', 'GLIBCXX_3.4.11', 'GLIBCXX_3.4.14',
'CXXABI_1.3.8', 'CXXABI_1.3.9', 'GLIBCXX_3.4.18', 'CXXABI_1.3.11',
'GLIBCXX_3.4.22', 'GLIBCXX_3.4.9', 'GLIBCXX_3.4.5', 'GLIBCXX_3.4',
'CXXABI_1.3.5', 'GLIBCXX_3.4.17', 'GLIBCXX_3.4.15', 'GLIBCXX_3.4.21',
'CXXABI_1.3.3', 'GLIBCXX_3.4.20'}, libgcc_s.so.1 with versions
{'GCC_3.0', 'GCC_3.3'}, libc.so.6 with versions {'GLIBC_2.3.4',
'GLIBC_2.11', 'GLIBC_2.3', 'GLIBC_2.6', 'GLIBC_2.15', 'GLIBC_2.14',
'GLIBC_2.2.5', 'GLIBC_2.17', 'GLIBC_2.4'}

This constrains the platform tag to "manylinux_2_27_x86_64". In order
to achieve a more compatible tag, you would need to recompile a new
wheel from source on a system with earlier versions of these
libraries, such as a recent manylinux image.

 

0 Kudos
2 Replies
ArpanB_Intel
Moderator
530 Views

Hi Matthieu, the toolkits and the tools within are tested on a list of OS platforms that make up the Supported OS platforms list.


Please note that the product may or may not work on the OS platforms that are not listed in the Supported OS list. However, if it does not work, Intel® will be unable to support it.


0 Kudos
ArpanB_Intel
Moderator
491 Views

Matthieu, unfortunately we were unable to hear back from you.


If you have any further queries, please post a new question as this thread will no longer be monitored by Intel®.


0 Kudos
Reply