Merge branch 'it4i-karolina'

a/ANSYS/ANSYS-2023R2-intel-2022.12-palanti.eb

@@ -0,0 +1,37 @@
+# IT4Innovations
+# LK 2024
+
+name = 'ANSYS'
+version = '2023R2'
+versionsuffix = '-palanti'
+
+homepage = 'http://www.ansys.com'
+description = """ANSYS simulation software enables organizations to confidently predict
+ how their products will operate in the real world. We believe that every product is
+ a promise of something greater. """
+
+toolchain = {'name': 'intel', 'version': '2022.12'}
+
+sources = [
+    {'filename': '%(name)s%(version)s_LINX64_DISK1.iso', 'extract_cmd': '7z x -aos %s'},
+    {'filename': '%(name)s%(version)s_LINX64_DISK2.iso', 'extract_cmd': '7z x -aos %s'},
+    {'filename': '%(name)s%(version)s_LINX64_DISK3.iso', 'extract_cmd': '7z x -aos %s'},
+]
+checksums = [
+    {'ANSYS2023R2_LINX64_DISK1.iso': '4f952592f1f95a6f3715dbaeb6e56b2625f6826a80c5842234406c7741f29e1f'},
+    {'ANSYS2023R2_LINX64_DISK2.iso': 'd02fd9e4db1a08dc867b771979f5d9ffa4d69fd94740e60444e856911fb89ce0'},
+    {'ANSYS2023R2_LINX64_DISK3.iso': 'd167f30a3cf632a234111e9a398dae6f824cb0eae5a7cb7afa1453474726de4a'},
+]
+
+#dependencies = [
+#   ('libGLU', '9.0.2'), # mame v systemu, pro jistotu zkusime bez protoze stary CentOS
+#   ('libnsl', '2.0.0'), # moc nove pro CentOS 7!!!
+#]
+
+osdependencies = [('p7zip', 'p7zip-plugins')]
+
+import os
+license_server = os.getenv('EB_ANSYS_LICENSE_SERVER', '185.215.147.62')
+license_server_port = os.getenv('EB_ANSYS_LICENSE_SERVER_PORT', '2325:1055')
+
+moduleclass = 'tools'

c/CP2K/CP2K-2024.1-foss-2022a.eb

@@ -0,0 +1,40 @@
+# IT4Innovations
+# LK 2024
+
+name = 'CP2K'
+version = '2024.1'
+
+homepage = 'https://www.cp2k.org/'
+description = """CP2K is a freely available (GPL) program, written in Fortran 95, to perform atomistic and molecular
+ simulations of solid state, liquid, molecular and biological systems. It provides a general framework for different
+ methods such as e.g. density functional theory (DFT) using a mixed Gaussian and plane waves approach (GPW), and
+ classical pair and many-body potentials. """
+
+toolchain = {'name': 'foss', 'version': '2022a'}
+toolchainopts = {'pic': True, 'openmp': True}
+
+source_urls = ['https://github.com/cp2k/cp2k/releases/download/v%(version)s/']
+sources = [SOURCELOWER_TAR_BZ2]
+checksums = ['a7abf149a278dfd5283dc592a2c4ae803b37d040df25d62a5e35af5c4557668f']
+
+#                https://github.com/cp2k/cp2k/releases/download/v2022.1/cp2k-2022.1.tar.bz2
+dependencies = [
+    ('Libint', '2.7.2', '-lmax-6-cp2k'),
+    ('libxc', '5.2.3'),
+    ('libxsmm', '1.17'),
+    ('FFTW', '3.3.10'),
+    ('PLUMED', '2.8.0'),
+]
+
+builddependencies = [
+    ('flex', '2.6.4'),
+    ('Bison', '3.8.2'),
+]
+
+type = 'psmp'
+
+# regression test reports handful of failures,
+# we're assuming those are OK to ignore...
+ignore_regtest_fails = True
+
+moduleclass = 'chem'

h/HDF5/HDF5-1.14.0-nvompi-2022.07.eb

@@ -0,0 +1,27 @@
+# IT4Innovations
+# LK 2024
+
+name = 'HDF5'
+version = '1.14.0'
+
+homepage = 'https://portal.hdfgroup.org/display/support'
+description = """HDF5 is a data model, library, and file format for storing and managing data.
+ It supports an unlimited variety of datatypes, and is designed for flexible
+ and efficient I/O and for high volume and complex data."""
+
+toolchain = {'name': 'nvompi', 'version': '2022.07'}
+toolchainopts = {'pic': True, 'usempi': True}
+
+source_urls = ['https://support.hdfgroup.org/ftp/HDF5/releases/hdf5-%(version_major_minor)s/hdf5-%(version)s/src']
+sources = [SOURCELOWER_TAR_GZ]
+checksums = ['a571cc83efda62e1a51a0a912dd916d01895801c5025af91669484a1575a6ef4']
+
+configopts = '--enable-fortran --enable-fortran 2003 --enable-cxx --enable-parallel --enable-unsupported --enable-shared'
+
+dependencies = [
+    ('OpenMPI', '4.1.4', '-CUDA-11.7.0', ('NVHPC', '22.7')),
+    ('zlib', '1.2.12'),
+    ('Szip', '2.1.1'),
+]
+
+moduleclass = 'data'

h/HDF5/HDF5-1.14.0-nvompi-2023.03.eb

@@ -1,5 +1,5 @@
 # IT4Innovations
-# LK 2023
+# LK 2024
 
 name = 'HDF5'
 version = '1.14.0'
@@ -9,8 +9,8 @@ description = """HDF5 is a data model, library, and file format for storing and
  It supports an unlimited variety of datatypes, and is designed for flexible
  and efficient I/O and for high volume and complex data."""
 
-toolchain = {'name': 'NVHPC', 'version': '23.5'}
+toolchain = {'name': 'nvompi', 'version': '2023.03'}
-toolchainopts = {'pic': True}
+toolchainopts = {'pic': True, 'usempi': True}
 
 source_urls = ['https://support.hdfgroup.org/ftp/HDF5/releases/hdf5-%(version_major_minor)s/hdf5-%(version)s/src']
 sources = [SOURCELOWER_TAR_GZ]
@@ -19,7 +19,7 @@ checksums = ['a571cc83efda62e1a51a0a912dd916d01895801c5025af91669484a1575a6ef4']
 configopts = '--enable-fortran --enable-fortran 2003 --enable-cxx --enable-parallel --enable-unsupported --enable-shared'
 
 dependencies = [
-    ('OpenMPI', '4.1.5', '-CUDA-12.2.0'),
+    ('OpenMPI', '4.1.6', '-CUDA-12.2.0'),
     ('zlib', '1.2.12'),
     ('Szip', '2.1.1'),
 ]

h/HDF5/HDF5-1.14.0-nvompi-2023.05.eb

@@ -0,0 +1,27 @@
+# IT4Innovations
+# LK 2024
+
+name = 'HDF5'
+version = '1.14.0'
+
+homepage = 'https://portal.hdfgroup.org/display/support'
+description = """HDF5 is a data model, library, and file format for storing and managing data.
+ It supports an unlimited variety of datatypes, and is designed for flexible
+ and efficient I/O and for high volume and complex data."""
+
+toolchain = {'name': 'nvompi', 'version': '2023.05'}
+toolchainopts = {'pic': True, 'usempi': True}
+
+source_urls = ['https://support.hdfgroup.org/ftp/HDF5/releases/hdf5-%(version_major_minor)s/hdf5-%(version)s/src']
+sources = [SOURCELOWER_TAR_GZ]
+checksums = ['a571cc83efda62e1a51a0a912dd916d01895801c5025af91669484a1575a6ef4']
+
+configopts = '--enable-fortran --enable-fortran 2003 --enable-cxx --enable-parallel --enable-unsupported --enable-shared'
+
+dependencies = [
+    ('OpenMPI', '4.1.6', '-CUDA-12.2.0'),
+    ('zlib', '1.2.12'),
+    ('Szip', '2.1.1'),
+]
+
+moduleclass = 'data'

h/HDF5/HDF5-1.14.0-nvompi-2023.11.eb

@@ -0,0 +1,27 @@
+# IT4Innovations
+# LK 2024
+
+name = 'HDF5'
+version = '1.14.0'
+
+homepage = 'https://portal.hdfgroup.org/display/support'
+description = """HDF5 is a data model, library, and file format for storing and managing data.
+ It supports an unlimited variety of datatypes, and is designed for flexible
+ and efficient I/O and for high volume and complex data."""
+
+toolchain = {'name': 'nvompi', 'version': '2023.11'}
+toolchainopts = {'pic': True, 'usempi': True}
+
+source_urls = ['https://support.hdfgroup.org/ftp/HDF5/releases/hdf5-%(version_major_minor)s/hdf5-%(version)s/src']
+sources = [SOURCELOWER_TAR_GZ]
+checksums = ['a571cc83efda62e1a51a0a912dd916d01895801c5025af91669484a1575a6ef4']
+
+configopts = '--enable-fortran --enable-fortran 2003 --enable-cxx --enable-parallel --enable-unsupported --enable-shared'
+
+dependencies = [
+    ('OpenMPI', '4.1.6', '-CUDA-12.2.0'),
+    ('zlib', '1.2.12'),
+    ('Szip', '2.1.1'),
+]
+
+moduleclass = 'data'

n/NCCL/NCCL-2.19.4-GCCcore-13.2.0-CUDA-12.2.0.eb

@@ -0,0 +1,29 @@
+# IT4Innovations
+# LK 2024
+
+name = 'NCCL'
+version = '2.19.4'
+versionsuffix = '-CUDA-%(cudaver)s'
+
+homepage = 'https://developer.nvidia.com/nccl'
+description = """The NVIDIA Collective Communications Library (NCCL) implements multi-GPU and multi-node collective
+communication primitives that are performance optimized for NVIDIA GPUs."""
+
+toolchain = {'name': 'GCCcore', 'version': '13.2.0'}
+
+github_account = 'NVIDIA'
+source_urls = [GITHUB_SOURCE]
+sources = ['v%(version)s-1.tar.gz']
+checksums = ['a3948ade5d83a036dde7ca6d0fee3a960a4de5c7915a9be903a28a747b5babc4']
+
+builddependencies = [('binutils', '2.40')]
+
+dependencies = [
+    ('CUDA', '12.2.0', '', SYSTEM),
+    ('UCX-CUDA', '1.15.0', versionsuffix),
+]
+
+# default CUDA compute capabilities to use (override via --cuda-compute-capabilities)
+cuda_compute_capabilities = ['8.0']
+
+moduleclass = 'lib'

n/NVHPC/NVHPC-23.11.eb

@@ -0,0 +1,34 @@
+# IT4Innovations
+# LK 2024
+
+name = 'NVHPC'
+version = '23.11'
+
+homepage = 'https://developer.nvidia.com/hpc-sdk/'
+description = """C, C++ and Fortran compilers included with the NVIDIA HPC SDK (previously: PGI)"""
+
+toolchain = SYSTEM
+
+# By downloading, you accept the HPC SDK Software License Agreement (https://docs.nvidia.com/hpc-sdk/eula/index.html)
+accept_eula = True
+source_urls = ['https://developer.download.nvidia.com/hpc-sdk/%(version)s/']
+local_tarball_tmpl = 'nvhpc_2023_%%(version_major)s%%(version_minor)s_Linux_%s_cuda_multi.tar.gz'
+sources = [local_tarball_tmpl % '%(arch)s']
+checksums = ['33483a069a911f9309cd53859ab90d2778fb176df906e9e8d2bd55f45eeec400']
+
+local_gccver = '12.2.0'
+dependencies = [
+    ('GCCcore', local_gccver),
+    ('binutils', '2.39', '', ('GCCcore', local_gccver)),
+    ('numactl', '2.0.16', '', ('GCCcore', local_gccver))
+]
+
+default_cuda_version = '12.2'
+
+import os
+if os.environ.get("CLUSTERNAME") in ["KAROLINA"]:
+   cuda_compute_capabilities = "8.0"
+else:
+   cuda_compute_capabilities = "7.0"
+
+moduleclass = 'compiler'

n/NVHPC/NVHPC-23.7.eb

@@ -16,7 +16,11 @@ local_tarball_tmpl = 'nvhpc_2023_%%(version_major)s%%(version_minor)s_Linux_%s_c
 sources = [local_tarball_tmpl % '%(arch)s']
 checksums = ['fea91d95ff18bca1ce7afde50371caa02001ade8bed6ddfc5ff70862ccbebece']
 
+
+local_gccver = '13.2.0'
 dependencies = [
+    ('GCCcore', local_gccver),
+    ('binutils', '2.40', '', ('GCCcore', local_gccver)),
     ('numactl', '2.0.16')
 ]
 

n/nvompi/nvompi-2023.03.eb

@@ -0,0 +1,22 @@
+# IT4Innovations
+# LK 2024
+
+easyblock = 'Toolchain'
+
+name = 'nvompi'
+version = '2023.03'
+
+homepage = '(none)'
+description = 'NVHPC based compiler toolchain, including OpenMPI for MPI support.'
+
+toolchain = SYSTEM
+
+local_compiler = ('NVHPC', '23.3')
+
+dependencies = [
+    local_compiler,
+    ('OpenMPI', '4.1.6', '-CUDA-12.2.0', local_compiler),
+    ('CUDA', '12.2.0', '', SYSTEM),
+]
+
+moduleclass = 'toolchain'

n/nvompi/nvompi-2023.05.eb

@@ -0,0 +1,22 @@
+# IT4Innovations
+# LK 2024
+
+easyblock = 'Toolchain'
+
+name = 'nvompi'
+version = '2023.05'
+
+homepage = '(none)'
+description = 'NVHPC based compiler toolchain, including OpenMPI for MPI support.'
+
+toolchain = SYSTEM
+
+local_compiler = ('NVHPC', '23.5')
+
+dependencies = [
+    local_compiler,
+    ('OpenMPI', '4.1.6', '-CUDA-12.2.0', local_compiler),
+    ('CUDA', '12.2.0', '', SYSTEM),
+]
+
+moduleclass = 'toolchain'

n/nvompi/nvompi-2023.11.eb

@@ -0,0 +1,22 @@
+# IT4Innovations
+# LK 2024
+
+easyblock = 'Toolchain'
+
+name = 'nvompi'
+version = '2023.11'
+
+homepage = '(none)'
+description = 'NVHPC based compiler toolchain, including OpenMPI for MPI support.'
+
+toolchain = SYSTEM
+
+local_compiler = ('NVHPC', '23.11')
+
+dependencies = [
+    local_compiler,
+    ('OpenMPI', '4.1.6', '-CUDA-12.2.0', local_compiler),
+    ('CUDA', '12.2.0', '', SYSTEM),
+]
+
+moduleclass = 'toolchain'

o/OpenMPI/OpenMPI-4.1.4-NVHPC-22.7-CUDA-11.7.0.eb

@@ -90,7 +90,7 @@ sanity_check_paths = {
                     x for x in [
                         "mpi-ext", "mpif-config", "mpif", "mpi", "mpi_portable_platform"]], 'dirs': [], }
 
-mport os
+import os
 if os.environ.get("CLUSTERNAME") in ["BARBORA"]:
    modextravars = {'OMPI_MCA_btl_openib_if_include': 'mlx5_0',
                 'OMPI_MCA_btl_tcp_if_include': '10.33.4.0/24',

o/OpenMPI/OpenMPI-4.1.5-NVHPC-23.11-CUDA-12.2.0.eb

@@ -0,0 +1,115 @@
+# IT4Innovations
+# LK 2023
+
+name = 'OpenMPI'
+version = '4.1.5'
+versionsuffix = '-CUDA-12.2.0'
+
+homepage = 'https://www.open-mpi.org/'
+description = """The Open MPI Project is an open source MPI-3 implementation."""
+
+toolchain = {'name': 'NVHPC', 'version': '23.11'}
+
+source_urls = ['https://www.open-mpi.org/software/ompi/v%(version_major_minor)s/downloads']
+sources = [SOURCELOWER_TAR_BZ2]
+patches = [
+    'OpenMPI-4.1.1_build-with-internal-cuda-header.patch',
+    'OpenMPI-4.1.1_opal-datatype-cuda-performance.patch',
+]
+checksums = [
+    'a640986bc257389dd379886fdae6264c8cfa56bc98b71ce3ae3dfbd8ce61dbe3',  # openmpi-4.1.5.tar.bz2
+    # OpenMPI-4.1.1_build-with-internal-cuda-header.patch
+    '63eac52736bdf7644c480362440a7f1f0ae7c7cae47b7565f5635c41793f8c83',
+    # OpenMPI-4.1.1_opal-datatype-cuda-performance.patch
+    'b767c7166cf0b32906132d58de5439c735193c9fd09ec3c5c11db8d5fa68750e',
+]
+
+builddependencies = [
+    ('pkgconf', '1.9.3'),
+    ('Perl', '5.36.0'),
+    ('Autotools', '20220317'),
+]
+
+dependencies = [
+    ('zlib', '1.2.12'),
+    ('CUDA', '12.2.0', '', True),
+    ('hwloc', '2.8.0'),
+    ('libevent', '2.1.12'),
+    ('UCX', '1.14.1'),
+    ('UCX-CUDA', '1.14.1', '-CUDA-%(cudaver)s'),
+    ('libfabric', '1.16.1'),
+    ('PMIx', '4.2.2'),
+    ('UCC', '1.1.0'),
+    ('UCC-CUDA', '1.1.0', '-CUDA-%(cudaver)s'),
+]
+
+# Update configure to include changes from the "internal-cuda" patch
+# by running a subset of autogen.pl sufficient to achieve this
+# without doing the full, long-running regeneration.
+preconfigopts = ' && '.join([
+    'cd config',
+    'autom4te --language=m4sh opal_get_version.m4sh -o opal_get_version.sh',
+    'cd ..',
+    'autoconf',
+    'autoheader',
+    'aclocal',
+    'automake',
+    ''
+])
+
+# CUDA related patches and custom configure option can be removed if CUDA support isn't wanted.
+configopts = ' --with-cuda=internal'
+configopts += ' CC=pgcc CXX=pgc++ FC=pgfortran'
+configopts += ' CXXFLAGS="-fPIC"'
+
+# IT4I-specific settings
+
+#configopts += '--enable-shared '
+configopts += ' --enable-mpi-thread-multiple'
+configopts += ' --with-verbs'
+configopts += ' --enable-mpirun-prefix-by-default'
+configopts += ' --with-hwloc=$EBROOTHWLOC'  # hwloc support
+#configopts += ' --with-tm=/opt/pbs '  # Enable PBS
+configopts += ' --with-slurm' # Enable slurm
+configopts += ' --enable-mpi-cxx'  # Enable building the C++ MPI bindings
+configopts += ' --with-ucx=$EBROOTUCX'
+configopts += ' --with-knem=/opt/knem-1.1.4.90mlnx1'
+
+osdependencies = [('libibverbs-dev', 'libibverbs-devel', 'rdma-core-devel')]
+
+postinstallcmds = [
+    'echo "# By default, for Open MPI 4.0 and later, infiniband ports on a device are not used by default." >> %(installdir)s/etc/openmpi-mca-params.conf',
+    'echo "btl_openib_allow_ib = true" >> %(installdir)s/etc/openmpi-mca-params.conf',
+]
+
+local_libs = ["mpi_mpifh", "mpi", "ompitrace", "open-pal", "open-rte"]
+sanity_check_paths = {
+    'files': [
+        "bin/%s" %
+        binfile for binfile in [
+            "ompi_info", "opal_wrapper", "orterun"]] + [
+                "lib/lib%s.%s" %
+                (libfile, SHLIB_EXT) for libfile in local_libs] + [
+                    "include/%s.h" %
+                    x for x in [
+                        "mpi-ext", "mpif-config", "mpif", "mpi", "mpi_portable_platform"]], 'dirs': [], }
+
+import os
+if os.environ.get("CLUSTERNAME") in ["BARBORA"]:
+   modextravars = {'OMPI_MCA_btl_openib_if_include': 'mlx5_0',
+                'OMPI_MCA_btl_tcp_if_include': '10.33.4.0/24',
+                'OMPI_MCA_orte_base_help_aggregate': '0',
+                'SLURM_MPI_TYPE': 'pmix_v4',
+   }
+elif os.environ.get("CLUSTERNAME") in ["KAROLINA"]:
+   modextravars = {'OMPI_MCA_btl_openib_if_include': 'mlx5_0',
+                'OMPI_MCA_orte_base_help_aggregate': '0',
+                'SLURM_MPI_TYPE': 'pmix_v4',
+   }
+else:
+   modextravars = {'OMPI_MCA_btl_openib_if_include': 'mlx4_0',
+                'OMPI_MCA_oob_tcp_if_include': '10.0.0.0/8',
+                'SLURM_MPI_TYPE': 'pmix_v4',
+   }
+
+moduleclass = 'mpi'

o/OpenMPI/OpenMPI-4.1.6-NVHPC-23.11-CUDA-12.2.0.eb

@@ -0,0 +1,115 @@
+# IT4Innovations
+# LK 2023
+
+name = 'OpenMPI'
+version = '4.1.6'
+versionsuffix = '-CUDA-12.2.0'
+
+homepage = 'https://www.open-mpi.org/'
+description = """The Open MPI Project is an open source MPI-3 implementation."""
+
+toolchain = {'name': 'NVHPC', 'version': '23.11'}
+
+source_urls = ['https://www.open-mpi.org/software/ompi/v%(version_major_minor)s/downloads']
+sources = [SOURCELOWER_TAR_BZ2]
+patches = [
+    'OpenMPI-4.1.1_build-with-internal-cuda-header.patch',
+    'OpenMPI-4.1.1_opal-datatype-cuda-performance.patch',
+]
+checksums = [
+    'f740994485516deb63b5311af122c265179f5328a0d857a567b85db00b11e415',  # openmpi-4.1.6.tar.bz2
+    # OpenMPI-4.1.1_build-with-internal-cuda-header.patch
+    '63eac52736bdf7644c480362440a7f1f0ae7c7cae47b7565f5635c41793f8c83',
+    # OpenMPI-4.1.1_opal-datatype-cuda-performance.patch
+    'b767c7166cf0b32906132d58de5439c735193c9fd09ec3c5c11db8d5fa68750e',
+]
+
+builddependencies = [
+    ('pkgconf', '1.9.3'),
+    ('Perl', '5.36.0'),
+    ('Autotools', '20220317'),
+]
+
+dependencies = [
+    ('zlib', '1.2.12'),
+    ('CUDA', '12.2.0', '', True),
+    ('hwloc', '2.8.0'),
+    ('libevent', '2.1.12'),
+    ('UCX', '1.14.1'),
+    ('UCX-CUDA', '1.14.1', '-CUDA-%(cudaver)s'),
+    ('libfabric', '1.16.1'),
+    ('PMIx', '4.2.2'),
+    ('UCC', '1.1.0'),
+    ('UCC-CUDA', '1.1.0', '-CUDA-%(cudaver)s'),
+]
+
+# Update configure to include changes from the "internal-cuda" patch
+# by running a subset of autogen.pl sufficient to achieve this
+# without doing the full, long-running regeneration.
+preconfigopts = ' && '.join([
+    'cd config',
+    'autom4te --language=m4sh opal_get_version.m4sh -o opal_get_version.sh',
+    'cd ..',
+    'autoconf',
+    'autoheader',
+    'aclocal',
+    'automake',
+    ''
+])
+
+# CUDA related patches and custom configure option can be removed if CUDA support isn't wanted.
+configopts = ' --with-cuda=internal'
+configopts += ' CC=pgcc CXX=pgc++ FC=pgfortran'
+configopts += ' CXXFLAGS="-fPIC"'
+
+# IT4I-specific settings
+
+#configopts += '--enable-shared '
+configopts += ' --enable-mpi-thread-multiple'
+configopts += ' --with-verbs'
+configopts += ' --enable-mpirun-prefix-by-default'
+configopts += ' --with-hwloc=$EBROOTHWLOC'  # hwloc support
+#configopts += ' --with-tm=/opt/pbs '  # Enable PBS
+configopts += ' --with-slurm' # Enable slurm
+configopts += ' --enable-mpi-cxx'  # Enable building the C++ MPI bindings
+configopts += ' --with-ucx=$EBROOTUCX'
+configopts += ' --with-knem=/opt/knem-1.1.4.90mlnx1'
+
+osdependencies = [('libibverbs-dev', 'libibverbs-devel', 'rdma-core-devel')]
+
+postinstallcmds = [
+    'echo "# By default, for Open MPI 4.0 and later, infiniband ports on a device are not used by default." >> %(installdir)s/etc/openmpi-mca-params.conf',
+    'echo "btl_openib_allow_ib = true" >> %(installdir)s/etc/openmpi-mca-params.conf',
+]
+
+local_libs = ["mpi_mpifh", "mpi", "ompitrace", "open-pal", "open-rte"]
+sanity_check_paths = {
+    'files': [
+        "bin/%s" %
+        binfile for binfile in [
+            "ompi_info", "opal_wrapper", "orterun"]] + [
+                "lib/lib%s.%s" %
+                (libfile, SHLIB_EXT) for libfile in local_libs] + [
+                    "include/%s.h" %
+                    x for x in [
+                        "mpi-ext", "mpif-config", "mpif", "mpi", "mpi_portable_platform"]], 'dirs': [], }
+
+import os
+if os.environ.get("CLUSTERNAME") in ["BARBORA"]:
+   modextravars = {'OMPI_MCA_btl_openib_if_include': 'mlx5_0',
+                'OMPI_MCA_btl_tcp_if_include': '10.33.4.0/24',
+                'OMPI_MCA_orte_base_help_aggregate': '0',
+                'SLURM_MPI_TYPE': 'pmix_v4',
+   }
+elif os.environ.get("CLUSTERNAME") in ["KAROLINA"]:
+   modextravars = {'OMPI_MCA_btl_openib_if_include': 'mlx5_0',
+                'OMPI_MCA_orte_base_help_aggregate': '0',
+                'SLURM_MPI_TYPE': 'pmix_v4',
+   }
+else:
+   modextravars = {'OMPI_MCA_btl_openib_if_include': 'mlx4_0',
+                'OMPI_MCA_oob_tcp_if_include': '10.0.0.0/8',
+                'SLURM_MPI_TYPE': 'pmix_v4',
+   }
+
+moduleclass = 'mpi'

o/OpenMPI/OpenMPI-4.1.6-NVHPC-23.3-CUDA-12.2.0.eb

@@ -0,0 +1,115 @@
+# IT4Innovations
+# LK 2024
+
+name = 'OpenMPI'
+version = '4.1.6'
+versionsuffix = '-CUDA-12.2.0'
+
+homepage = 'https://www.open-mpi.org/'
+description = """The Open MPI Project is an open source MPI-3 implementation."""
+
+toolchain = {'name': 'NVHPC', 'version': '23.3'}
+
+source_urls = ['https://www.open-mpi.org/software/ompi/v%(version_major_minor)s/downloads']
+sources = [SOURCELOWER_TAR_BZ2]
+patches = [
+    'OpenMPI-4.1.1_build-with-internal-cuda-header.patch',
+    'OpenMPI-4.1.1_opal-datatype-cuda-performance.patch',
+]
+checksums = [
+    'f740994485516deb63b5311af122c265179f5328a0d857a567b85db00b11e415',  # openmpi-4.1.6.tar.bz2
+    # OpenMPI-4.1.1_build-with-internal-cuda-header.patch
+    '63eac52736bdf7644c480362440a7f1f0ae7c7cae47b7565f5635c41793f8c83',
+    # OpenMPI-4.1.1_opal-datatype-cuda-performance.patch
+    'b767c7166cf0b32906132d58de5439c735193c9fd09ec3c5c11db8d5fa68750e',
+]
+
+builddependencies = [
+    ('pkgconf', '1.9.3'),
+    ('Perl', '5.36.0'),
+    ('Autotools', '20220317'),
+]
+
+dependencies = [
+    ('zlib', '1.2.12'),
+    ('CUDA', '12.2.0', '', True),
+    ('hwloc', '2.8.0'),
+    ('libevent', '2.1.12'),
+    ('UCX', '1.14.1'),
+    ('UCX-CUDA', '1.14.1', '-CUDA-%(cudaver)s'),
+    ('libfabric', '1.16.1'),
+    ('PMIx', '4.2.2'),
+    ('UCC', '1.1.0'),
+    ('UCC-CUDA', '1.1.0', '-CUDA-%(cudaver)s'),
+]
+
+# Update configure to include changes from the "internal-cuda" patch
+# by running a subset of autogen.pl sufficient to achieve this
+# without doing the full, long-running regeneration.
+preconfigopts = ' && '.join([
+    'cd config',
+    'autom4te --language=m4sh opal_get_version.m4sh -o opal_get_version.sh',
+    'cd ..',
+    'autoconf',
+    'autoheader',
+    'aclocal',
+    'automake',
+    ''
+])
+
+# CUDA related patches and custom configure option can be removed if CUDA support isn't wanted.
+configopts = ' --with-cuda=internal'
+configopts += ' CC=pgcc CXX=pgc++ FC=pgfortran'
+configopts += ' CXXFLAGS="-fPIC"'
+
+# IT4I-specific settings
+
+#configopts += '--enable-shared '
+configopts += ' --enable-mpi-thread-multiple'
+configopts += ' --with-verbs'
+configopts += ' --enable-mpirun-prefix-by-default'
+configopts += ' --with-hwloc=$EBROOTHWLOC'  # hwloc support
+#configopts += ' --with-tm=/opt/pbs '  # Enable PBS
+configopts += ' --with-slurm' # Enable slurm
+configopts += ' --enable-mpi-cxx'  # Enable building the C++ MPI bindings
+configopts += ' --with-ucx=$EBROOTUCX'
+configopts += ' --with-knem=/opt/knem-1.1.4.90mlnx1'
+
+osdependencies = [('libibverbs-dev', 'libibverbs-devel', 'rdma-core-devel')]
+
+postinstallcmds = [
+    'echo "# By default, for Open MPI 4.0 and later, infiniband ports on a device are not used by default." >> %(installdir)s/etc/openmpi-mca-params.conf',
+    'echo "btl_openib_allow_ib = true" >> %(installdir)s/etc/openmpi-mca-params.conf',
+]
+
+local_libs = ["mpi_mpifh", "mpi", "ompitrace", "open-pal", "open-rte"]
+sanity_check_paths = {
+    'files': [
+        "bin/%s" %
+        binfile for binfile in [
+            "ompi_info", "opal_wrapper", "orterun"]] + [
+                "lib/lib%s.%s" %
+                (libfile, SHLIB_EXT) for libfile in local_libs] + [
+                    "include/%s.h" %
+                    x for x in [
+                        "mpi-ext", "mpif-config", "mpif", "mpi", "mpi_portable_platform"]], 'dirs': [], }
+
+import os
+if os.environ.get("CLUSTERNAME") in ["BARBORA"]:
+   modextravars = {'OMPI_MCA_btl_openib_if_include': 'mlx5_0',
+                'OMPI_MCA_btl_tcp_if_include': '10.33.4.0/24',
+                'OMPI_MCA_orte_base_help_aggregate': '0',
+                'SLURM_MPI_TYPE': 'pmix_v4',
+   }
+elif os.environ.get("CLUSTERNAME") in ["KAROLINA"]:
+   modextravars = {'OMPI_MCA_btl_openib_if_include': 'mlx5_0',
+                'OMPI_MCA_orte_base_help_aggregate': '0',
+                'SLURM_MPI_TYPE': 'pmix_v4',
+   }
+else:
+   modextravars = {'OMPI_MCA_btl_openib_if_include': 'mlx4_0',
+                'OMPI_MCA_oob_tcp_if_include': '10.0.0.0/8',
+                'SLURM_MPI_TYPE': 'pmix_v4',
+   }
+
+moduleclass = 'mpi'

o/OpenMPI/OpenMPI-4.1.6-NVHPC-23.5-CUDA-12.2.0.eb

@@ -0,0 +1,115 @@
+# IT4Innovations
+# LK 2023
+
+name = 'OpenMPI'
+version = '4.1.6'
+versionsuffix = '-CUDA-12.2.0'
+
+homepage = 'https://www.open-mpi.org/'
+description = """The Open MPI Project is an open source MPI-3 implementation."""
+
+toolchain = {'name': 'NVHPC', 'version': '23.5'}
+
+source_urls = ['https://www.open-mpi.org/software/ompi/v%(version_major_minor)s/downloads']
+sources = [SOURCELOWER_TAR_BZ2]
+patches = [
+    'OpenMPI-4.1.1_build-with-internal-cuda-header.patch',
+    'OpenMPI-4.1.1_opal-datatype-cuda-performance.patch',
+]
+checksums = [
+    'f740994485516deb63b5311af122c265179f5328a0d857a567b85db00b11e415',  # openmpi-4.1.6.tar.bz2
+    # OpenMPI-4.1.1_build-with-internal-cuda-header.patch
+    '63eac52736bdf7644c480362440a7f1f0ae7c7cae47b7565f5635c41793f8c83',
+    # OpenMPI-4.1.1_opal-datatype-cuda-performance.patch
+    'b767c7166cf0b32906132d58de5439c735193c9fd09ec3c5c11db8d5fa68750e',
+]
+
+builddependencies = [
+    ('pkgconf', '1.9.3'),
+    ('Perl', '5.36.0'),
+    ('Autotools', '20220317'),
+]
+
+dependencies = [
+    ('zlib', '1.2.12'),
+    ('CUDA', '12.2.0', '', True),
+    ('hwloc', '2.8.0'),
+    ('libevent', '2.1.12'),
+    ('UCX', '1.14.1'),
+    ('UCX-CUDA', '1.14.1', '-CUDA-%(cudaver)s'),
+    ('libfabric', '1.16.1'),
+    ('PMIx', '4.2.2'),
+    ('UCC', '1.1.0'),
+    ('UCC-CUDA', '1.1.0', '-CUDA-%(cudaver)s'),
+]
+
+# Update configure to include changes from the "internal-cuda" patch
+# by running a subset of autogen.pl sufficient to achieve this
+# without doing the full, long-running regeneration.
+preconfigopts = ' && '.join([
+    'cd config',
+    'autom4te --language=m4sh opal_get_version.m4sh -o opal_get_version.sh',
+    'cd ..',
+    'autoconf',
+    'autoheader',
+    'aclocal',
+    'automake',
+    ''
+])
+
+# CUDA related patches and custom configure option can be removed if CUDA support isn't wanted.
+configopts = ' --with-cuda=internal'
+configopts += ' CC=pgcc CXX=pgc++ FC=pgfortran'
+configopts += ' CXXFLAGS="-fPIC"'
+
+# IT4I-specific settings
+
+#configopts += '--enable-shared '
+configopts += ' --enable-mpi-thread-multiple'
+configopts += ' --with-verbs'
+configopts += ' --enable-mpirun-prefix-by-default'
+configopts += ' --with-hwloc=$EBROOTHWLOC'  # hwloc support
+#configopts += ' --with-tm=/opt/pbs '  # Enable PBS
+configopts += ' --with-slurm' # Enable slurm
+configopts += ' --enable-mpi-cxx'  # Enable building the C++ MPI bindings
+configopts += ' --with-ucx=$EBROOTUCX'
+configopts += ' --with-knem=/opt/knem-1.1.4.90mlnx1'
+
+osdependencies = [('libibverbs-dev', 'libibverbs-devel', 'rdma-core-devel')]
+
+postinstallcmds = [
+    'echo "# By default, for Open MPI 4.0 and later, infiniband ports on a device are not used by default." >> %(installdir)s/etc/openmpi-mca-params.conf',
+    'echo "btl_openib_allow_ib = true" >> %(installdir)s/etc/openmpi-mca-params.conf',
+]
+
+local_libs = ["mpi_mpifh", "mpi", "ompitrace", "open-pal", "open-rte"]
+sanity_check_paths = {
+    'files': [
+        "bin/%s" %
+        binfile for binfile in [
+            "ompi_info", "opal_wrapper", "orterun"]] + [
+                "lib/lib%s.%s" %
+                (libfile, SHLIB_EXT) for libfile in local_libs] + [
+                    "include/%s.h" %
+                    x for x in [
+                        "mpi-ext", "mpif-config", "mpif", "mpi", "mpi_portable_platform"]], 'dirs': [], }
+
+import os
+if os.environ.get("CLUSTERNAME") in ["BARBORA"]:
+   modextravars = {'OMPI_MCA_btl_openib_if_include': 'mlx5_0',
+                'OMPI_MCA_btl_tcp_if_include': '10.33.4.0/24',
+                'OMPI_MCA_orte_base_help_aggregate': '0',
+                'SLURM_MPI_TYPE': 'pmix_v4',
+   }
+elif os.environ.get("CLUSTERNAME") in ["KAROLINA"]:
+   modextravars = {'OMPI_MCA_btl_openib_if_include': 'mlx5_0',
+                'OMPI_MCA_orte_base_help_aggregate': '0',
+                'SLURM_MPI_TYPE': 'pmix_v4',
+   }
+else:
+   modextravars = {'OMPI_MCA_btl_openib_if_include': 'mlx4_0',
+                'OMPI_MCA_oob_tcp_if_include': '10.0.0.0/8',
+                'SLURM_MPI_TYPE': 'pmix_v4',
+   }
+
+moduleclass = 'mpi'

p/pkgconf/pkgconf-2.0.3-NVHPC-23.11.eb

@@ -0,0 +1,32 @@
+# IT4Innovations
+# LK 2024
+
+easyblock = 'ConfigureMake'
+
+name = 'pkgconf'
+version = '2.0.3'
+
+homepage = 'https://github.com/pkgconf/pkgconf'
+
+description = """pkgconf is a program which helps to configure compiler and linker flags for development libraries.
+ It is similar to pkg-config from freedesktop.org."""
+
+toolchain = {'name': 'NVHPC', 'version': '23.11'}
+
+source_urls = ['https://distfiles.ariadne.space/pkgconf/']
+sources = [SOURCELOWER_TAR_GZ]
+checksums = ['a8f25f4b0c1ad48edc9b07ed46101aab4bcb305ba6ae811be49b7499b00eed49']
+
+postinstallcmds = ["cd %(installdir)s/bin && ln -s pkgconf pkg-config"]
+
+sanity_check_paths = {
+    'files': ['bin/pkg-config', 'bin/pkgconf'],
+    'dirs': [],
+}
+
+sanity_check_commands = [
+    "pkg-config --help",
+    "pkgconf --help",
+]
+
+moduleclass = 'devel'

u/UCC-CUDA/UCC-CUDA-1.2.0-GCCcore-13.2.0-CUDA-12.2.0.eb

@@ -0,0 +1,60 @@
+# IT4Innovations
+# LK 2024
+
+easyblock = 'ConfigureMake'
+
+name = 'UCC-CUDA'
+version = '1.2.0'
+versionsuffix = '-CUDA-%(cudaver)s'
+
+homepage = 'https://www.openucx.org/'
+description = """UCC (Unified Collective Communication) is a collective
+communication operations API and library that is flexible, complete, and 
+feature-rich for current and emerging programming models and runtimes.
+
+This module adds the UCC CUDA support.
+"""
+
+toolchain = {'name': 'GCCcore', 'version': '13.2.0'}
+toolchainopts = {'pic': True}
+
+source_urls = ['https://github.com/openucx/ucc/archive/refs/tags']
+sources = ['v%(version)s.tar.gz']
+patches = [
+    '%(name)s-1.0.0_link_against_existing_UCC_libs.patch',
+    '%(name)s-%(version)s_cuda_12_mem_ops.patch',
+]
+checksums = [
+    {'v1.1.0.tar.gz': '74c8ba75037b5bd88cb703e8c8ae55639af3fecfd4428912a433c010c97b4df7'},
+    {'UCC-CUDA-1.0.0_link_against_existing_UCC_libs.patch':
+     '9fa11cf6779174f4e9048df5812096e4261e1769d465cc7f34a6354398876856'},
+    {'UCC-CUDA-1.1.0_cuda_12_mem_ops.patch': 'fc3ea1487d29dc626db2363ef5a79e7f0906f6a7507a363fa6167a812b143eb6'},
+]
+
+builddependencies = [
+    ('binutils', '2.40'),
+    ('Autotools', '20220317'),
+]
+
+dependencies = [
+    ('UCC', '1.2.0'),
+    ('CUDA',  '12.2.0', '', SYSTEM),
+    ('UCX-CUDA', '1.15.0', '-CUDA-%(cudaver)s'),
+    ('NCCL', '2.18.3', '-CUDA-%(cudaver)s', SYSTEM),
+]
+
+preconfigopts = "./autogen.sh && "
+
+buildopts = '-C src/components/mc/cuda V=1 && make -C src/components/tl/nccl V=1'
+installopts = '-C src/components/mc/cuda && make -C src/components/tl/nccl install'
+
+sanity_check_paths = {
+    'files': ['lib/ucc/libucc_mc_cuda.%s' % SHLIB_EXT, 'lib/ucc/libucc_tl_nccl.%s' % SHLIB_EXT],
+    'dirs': ['lib']
+}
+
+sanity_check_commands = ["ucc_info -c"]
+
+modextrapaths = {'EB_UCC_EXTRA_COMPONENT_PATH': 'lib/ucc'}
+
+moduleclass = 'lib'

u/UCX-CUDA/UCX-CUDA-1.15.0-GCCcore-13.2.0-CUDA-12.2.0.eb

@@ -0,0 +1,51 @@
+# IT4Innovations
+# LK 2024
+
+easyblock = 'EB_UCX_Plugins'
+
+name = 'UCX-CUDA'
+version = '1.15.0'
+versionsuffix = '-CUDA-%(cudaver)s'
+
+homepage = 'http://www.openucx.org/'
+description = """Unified Communication X
+An open-source production grade communication framework for data centric
+and high-performance applications
+
+This module adds the UCX CUDA support.
+"""
+
+toolchain = {'name': 'GCCcore', 'version': '13.2.0'}
+toolchainopts = {'pic': True}
+
+source_urls = ['https://github.com/openucx/ucx/releases/download/v%(version)s']
+sources = [{'filename': 'ucx-%(version)s.tar.gz', 'alt_location': 'UCX'}]
+patches = ['%(name)s-1.11.0_link_against_existing_UCX_libs.patch']
+checksums = [
+    ('457187fa020e526609ba91e7750c9941d57bd57d60d6eed317b40ad8824aca93',
+     '4b202087076bc1c98f9249144f0c277a8ea88ad4ca6f404f94baa9cb3aebda6d'),  # ucx-1.15.0.tar.gz
+    {'UCX-CUDA-1.11.0_link_against_existing_UCX_libs.patch':
+     '457187fa020e526609ba91e7750c9941d57bd57d60d6eed317b40ad8824aca93'},
+]
+
+builddependencies = [
+    ('binutils', '2.40'),
+    ('Autotools', '20220317'),
+    ('pkgconf', '2.0.3'),
+]
+
+dependencies = [
+    ('zlib', '1.2.13'),
+    ('UCX', '1.15.0'),
+    ('CUDA', '12.2.0', '', SYSTEM),
+    ('GDRCopy', '2.4'),
+]
+
+configopts = '--enable-optimizations --enable-cma --enable-mt --with-verbs '
+configopts += '--without-java --without-go --disable-doxygen-doc '
+
+import os
+if os.environ.get("CLUSTERNAME") in ["KAROLINA", "BARBORA"]:
+   configopts += '--with-xpmem --with-knem=/opt/knem-1.1.4.90mlnx1'
+
+moduleclass = 'lib'