LAMMPS-23Jun2022-foss-2021b-kokkos-CUDA-11.4.1.eb

name = 'LAMMPS'
version = '23Jun2022'
versionsuffix = '-kokkos'
_cuda_suffix = '-CUDA-%(cudaver)s'
versionsuffix += _cuda_suffix

homepage = 'https://www.lammps.org'
description = """LAMMPS is a classical molecular dynamics code, and an acronym
for Large-scale Atomic/Molecular Massively Parallel Simulator. LAMMPS has
potentials for solid-state materials (metals, semiconductors) and soft matter
(biomolecules, polymers) and coarse-grained or mesoscopic systems. It can be
used to model atoms or, more generically, as a parallel particle simulator at
the atomic, meso, or continuum scale. LAMMPS runs on single processors or in
parallel using message-passing techniques and a spatial-decomposition of the
simulation domain. The code is designed to be easy to modify or extend with new
functionality.
"""

toolchain = {'name': 'foss', 'version': '2021b'}
toolchainopts = {'openmp': True, 'usempi': True}

# 'https://github.com/lammps/lammps/archive/'
source_urls = [GITHUB_LOWER_SOURCE]
sources = [
    'stable_%(version)s_update1.tar.gz',
]
patches = ['LAMMPS-23June2022_fix-timestep-balance-example.patch']
checksums = [
    {'stable_23Jun2022_update1.tar.gz': '58e3b2b984f8935bb0db5631e143be2826c45ffd48844f7c394f36624a3e17a2'},
    {'LAMMPS-23June2022_fix-timestep-balance-example.patch':
     '0ad6c953f08b0d539b189b7aa7cef76774aebb9d633b07d5b6001ac0fcea3081'},
]

builddependencies = [
    ('CMake', '3.22.1'),
    ('pkg-config', '0.29.2'),
    ('archspec', '0.1.3'),
]
dependencies = [
    ('CUDA', '11.4.1', '', SYSTEM),
    ('UCX-CUDA', '1.11.2', _cuda_suffix),
    ('NCCL', '2.10.3', _cuda_suffix),
    ('Python', '3.9.6'),
    ('libpng', '1.6.37'),
    ('libjpeg-turbo', '2.0.6'),
    ('netCDF', '4.8.1'),
    ('GSL', '2.7'),
    ('zlib', '1.2.11'),
    ('gzip', '1.10'),
    ('cURL', '7.78.0'),
    ('HDF5', '1.12.1'),
    # ('tbb', '2020.3'),  # Optionally required by Intel package but has a link problem when using nvlink
    ('PCRE', '8.45'),
    ('libxml2', '2.9.10'),
    ('FFmpeg', '4.3.2'),
    ('Voro++', '0.4.6'),
    ('kim-api', '2.3.0'),
    ('Eigen', '3.4.0'),
    ('PLUMED', '2.7.3'),
    ('ScaFaCoS', '1.0.1'),
    ('SciPy-bundle', '2021.10'),
    # VTK package is auto-disabled if this dep is not available
    ('VTK', '9.1.0'),
]

# To use additional custom configuration options, use the 'configopts' easyconfig parameter
# See docs and lammps easyblock for more information.
# https://github.com/lammps/lammps/blob/master/cmake/README.md#lammps-configuration-options

# OpenMP-Kokkos build is default in the current easyblock. One can switch to serial backend of Kokkos,
# which is claimed to be faster in pure MPI calculations
# configopts = "-DKokkos_ENABLE_SERIAL=yes "

# packages auto-enabled by easyblock
# 'GPU'    - if cuda package is present and kokkos is disabled
# 'KOKKOS' - if kokkos is enabled (by default)

# GCC 11/12 defaults to C++17 standard, need to be consistent
configopts = "-DKokkos_CXX_STANDARD=17 "

# include the following extra packages into the build
general_packages = [
    'ASPHERE',
    'ATC',
    'AWPMD',
    'BOCS',
    'BODY',
    'BPM',
    'BROWNIAN',
    'CG-DNA',
    'CG-SDK',
    'CLASS2',
    'COLLOID',
    'COLVARS',
    'COMPRESS',
    'CORESHELL',
    'DIELECTRIC',
    'DIFFRACTION',
    'DIPOLE',
    'DPD-BASIC',
    'DPD-MESO',
    'DPD-REACT',
    'DPD-SMOOTH',
    'DRUDE',
    'EFF',
    'ELECTRODE',
    'EXTRA-COMPUTE',
    'EXTRA-DUMP',
    'EXTRA-FIX',
    'EXTRA-MOLECULE',
    'EXTRA-PAIR',
    'FEP',
    'GRANULAR',
    'H5MD',
    'INTERLAYER',
    'KIM',
    'KSPACE',
    'LATBOLTZ',
    'MANIFOLD',
    'MACHDYN',
    'MANYBODY',
    'MC',
    'MEAM',
    'MGPT',
    'MISC',
    'ML-IAP',
    'ML-PACE',
    'ML-RANN',
    'ML-SNAP',
    'MOFFF',
    'MOLECULE',
    'MOLFILE',
    'MPIIO',
    'NETCDF',
    'OPENMP',
    'ORIENT',
    'PERI',
    'PHONON',
    'PLUGIN',
    'PLUMED',
    'POEMS',
    'PTM',
    'PYTHON',
    'QEQ',
    'QTB',
    'REACTION',
    'REAXFF',
    'REPLICA',
    'RIGID',
    'SCAFACOS',
    'SHOCK',
    'SMTBQ',
    'SPH',
    'SPIN',
    'SRD',
    'TALLY',
    'UEF',
    'VORONOI',
    'VTK',
    'YAFF',
]

# Excluded packages due to requiring additional (non-trivial) deps
# - ADIOS
# - LATTE
# - MESONT (requires very large files downloaded during build)
# - ML-HDNNP (requires N2P2)
# - ML-QUIP
# - MSCG
# - QMMM (setup seems complex)

# hardware-specific option
# cuda_compute_capabilities = ['8.0']

moduleclass = 'chem'