[TensorFlow] [package list]

TensorFlow/tensorflow-2.16.1-rocm-6.2.0-python-3.10-horovod-0.28.1-singularity-20241007 (tensorflow-2.16.1-rocm-6.2.0-python-3.10-horovod-0.28.1-singularity-20241007.eb)

Install with the EasyBuild-user module in partition/container:

module load LUMI partition/container EasyBuild-user
eb tensorflow-2.16.1-rocm-6.2.0-python-3.10-horovod-0.28.1-singularity-20241007.eb

The module will be available in all versions of the LUMI stack and in the CrayEnv stack.

To access module help after installation use module spider TensorFlow/tensorflow-2.16.1-rocm-6.2.0-python-3.10-horovod-0.28.1-singularity-20241007.

EasyConfig:

easyblock = 'MakeCp'

local_c_rocm_version =       '6.2.0'
local_c_python_mm =          '3.10' # major.minor only
local_c_Tensorflow_version = '2.16.1'
local_c_horovod_version =    '0.28.1'
local_c_dockerhash =         '8c75f24cfffc'
local_c_date =               '20241007'

name =          'TensorFlow'
version =       local_c_Tensorflow_version
versionsuffix = f'-rocm-{local_c_rocm_version}-python-{local_c_python_mm}-horovod-{local_c_horovod_version}-singularity-{local_c_date}'

local_sif = f'lumi-tensorflow-rocm-{local_c_rocm_version}-python-{local_c_python_mm}-tensorflow-{local_c_Tensorflow_version}-horovod-{local_c_horovod_version}-dockerhash-{local_c_dockerhash}.sif'
#local_docker = 'lumi-tensorflow-rocm-5.5.1-python-3.10-tensorflow-2.11.1-horovod-0.28.1.docker'

homepage = 'https://www.tensorflow.org/'

whatis = [
    'Description: TensorFlow, a machine learning package'
]

description = """
This module provides a container with TensorFlow %(version)s. 

The module defines a number of environment variables:
*   SIF and SIFTENSORFLOW: The full path and name of the Singularity SIF file 
    to use with singularity exec etc.
*   SINGULARITY_BIND: Mounts the necessary directories from the system,
    including /users, /project, /scratch and /flash so that you should be
    able to use your regular directories in the container.
*   RUNSCRIPTS and RUNSCRIPTSTENSORFLOW: The directory with some sample
    runscripts.

Note that this container uses a Conda environment internally. When in
the container, the command to activate the container is contained in the
environment variable WITH_CONDA.
"""

toolchain = SYSTEM

sources = [
    {
        'filename':    local_sif,
        'extract_cmd': '/bin/cp %s .'
    },
#    {
#        'filename':    local_docker,
#        'extract_cmd': '/bin/cp %s .'
#    },
]

skipsteps = ['build']

files_to_copy = [
    ([local_sif],    '.'),
#    ([local_docker], 'share/docker-defs/')    
]

local_bin_start_shell="""
#!/bin/bash -e

# Run application
if [[ -f "/.singularity.d/Singularity" ]] 
then
    # In a singularity container, just in case a user would add this to the path.
    if [ "\$#" -eq 0 ]
    then
        \$WITH_CONDA ; exec bash
    else
        \$WITH_CONDA ; exec bash "\$@"
    fi
else
    # Not yet in the container
    if [ "\$#" -eq 0 ]
    then
        singularity exec \$SIFTENSORFLOW bash -c '\\$WITH_CONDA ; exec bash'
    else
        singularity exec \$SIFTENSORFLOW bash "\$@"
    fi
fi

"""

local_runscript_python_simple="""
#!/bin/bash -e

# Start conda environment inside the container
\$WITH_CONDA

# Run application
python "\$@"

"""

postinstallcmds = [
    'mkdir -p %(installdir)s/bin',
    f'cat >%(installdir)s/bin/start-shell <<EOF {local_bin_start_shell}EOF',
    'chmod a+x %(installdir)s/bin/start-shell',
    'mkdir -p %(installdir)s/runscripts',
    f'cat >%(installdir)s/runscripts/conda-python-simple <<EOF {local_runscript_python_simple}EOF',
    'chmod a+x %(installdir)s/runscripts/conda-python-simple'
]

sanity_check_paths = {
    # We deliberately don't check for local_sif as the user is allowed to remove that file
    # but may still want to regenerate the module which would then fail in the sanity check.
    #'files': [f'share/docker-defs/{local_docker}'],
    'files': ['bin/start-shell', 'runscripts/conda-python-simple'],
    'dirs':  [],
}

modextravars = {
    # SIF variables currently set by a function via modluafooter.
    #'SIF':                 '%(installdir)s/' + local_sif,
    #'SIFTENSORFLOW':       '%(installdir)s/' + local_sif,
    'RUNSCRIPTS':           '%(installdir)s/runscripts',
    'RUNSCRIPTSTENSORFLOW': '%(installdir)s/runscripts',
    'SINGULARITY_BIND':     '/var/spool/slurmd,/opt/cray,/usr/lib64/libcxi.so.1,' +
                            '%(installdir)s/runscripts:/runscripts,' + 
                            '/pfs,/scratch,/projappl,/project,/flash,/appl'
}

modluafooter = f"""
-- Call a routine to set the various environment variables.
create_container_vars( '{local_sif}', '%(name)s', '%(installdir)s' )
"""

moduleclass = 'devel'

[TensorFlow] [package list]