Accessing Software with Modules
The software environment on the Lovelace cluster is primarily controlled through modules. By loading and switching modules you control the compilers, libraries and software available.
This means that for compiling software on Lovelace you typically set the compiler you wish to use by loading the appropriate compiler module, and load all the required library modules (e.g. numerical libraries such as OpenBLAS, driver related libraries such as CUDA, and IO format libraries).
Additionally, if you are compiling parallel applications using MPI (or SHMEM, etc.) then you will need to load the MPI environment and use the appropriate compiler wrapper scripts.
The list of the modules loaded can be obtained by typing:
[apraveen@uop-hpc ~]$ module list
Currently Loaded Modules:
1) gcc/13.3.1 7) zlib/1.3.1-zkl5wjj 13) hwloc/2.11.1-mqg2qec 19) pigz/2.8-pipe537 25) libxcrypt/4.4.35-2leymva 31) rdma-core/52.0-vibka2e
2) spack-modules/113 8) libxml2/2.13.4-vqoagpe 14) knem/1.1.4-prvvzic 20) zstd/1.5.6-pdk6d4j 26) openssh/9.9p1-agycqww 32) xpmem/2.6.5-36-rxnv6w6
3) glibc/2.28-lmcpkml 9) cuda-toolkit/12.4.1 15) openssl/3.4.0-7ckv34u 21) tar/1.34-dzeehng 27) pmix/5.0.3-2gkjs47 33) ucx/1.17.0-mzix24r
4) gcc-runtime/13.3.1-2kda5rk 10) cuda/12.4.1-b35sufh 16) libevent/2.1.12-tgsoax3 22) gettext/0.22.5-zjglimm 28) check/0.15.2-aflo7jk 34) ucc/1.3.0-h2s7mm6
5) libiconv/1.17-6nlfqwe 11) libpciaccess/0.17-d5wcj52 17) numactl/2.0.18-xy2slzw 23) krb5/1.21.3-i3rz3pq 29) gdrcopy/2.4.1-xyc5lwi 35) openmpi/5.0.5-fp6i5jk
6) xz/5.4.6-rb3mavc 12) ncurses/6.5-pnmzftp 18) bzip2/1.0.8-hwh4cnv 24) libedit/3.1-20240808-fxke3py 30) libnl/3.3.0-haaq6e3
Basic usage of the :bash:`module` command is covered below. For
full documentation please see:
Using the modules environment
Information on the available modules
Finding out which modules (and hence which compilers, libraries and
software) are available on the system is performed using the
module avail command:
[apraveen@uop-hpc ~]$ module avail
------------------------------------------- /users/cicd/.jenkins/workspace/UoP-Spack/main/113/spack/share/spack/lmod/openmpi/5.0.5-fp6i5jk/Core -------------------------------------------
abyss/2.3.10-65lydhd hpl/2.3-f7adg7v py-keras/3.6.0-qm535qo py-torch/2.5.1-h3qeknb xbeach/6118-rfmbfsw
adcirc/56.0.3-4oztlos openfoam/2312-fyzw4ln py-netcdf4/1.7.1.post2-lrnbi6s swan/4145-filbquq
bigdft-core/1.9.5-43un7wm py-bigdft/1.9.5-4xrtwr7 py-tensorflow/2.17.1-mlz3f6e swash/main-5s3s4uc
------------------------------------------------------ /users/cicd/.jenkins/workspace/UoP-Spack/main/113/spack/share/spack/lmod/Core ------------------------------------------------------
bwa/0.7.17-v72ok5n py-geopandas/1.0.1-vultwue py-pip-py3.13/23.1.2-dauydjf py-virtualenv-py3.13/20.26.5-sqb2hgl trinity/2.15.1.FULL-myzlz5u
gatk/4.5.0.0-yvwst4o py-jupyter/1.1.1-32xgg6m py-pip/23.1.2-qwz5sj5 py-virtualenv/20.26.5-fmvfrje ucx/1.17.0-mzix24r (L)
gnuplot/6.0.0-r5k27jt py-matplotlib/3.9.2-hsad7f7 py-scipy/1.14.1-tisu5tg python-py3.12/3.12.5-uni66am
openmpi/4.1.7-q3idazf py-numpy/1.26.4-mde553l py-sphinx-rtd-theme/1.2.2-45lz6nd python-py3.13/3.13.0-irwel6m
openmpi/5.0.5-fp6i5jk (L,D) py-pandas/2.2.3-sgvtc6s py-sphinx/6.2.1-fry74n4 python/3.11.9-bgjodjg
paml/4.10.7-pkdiklo py-pip-py3.12/23.1.2-dor7udl py-virtualenv-py3.12/20.26.5-qp5jrs3 r/4.4.1-flsgxmv
-------------------------------------------------------------------------------- /scratch/software/modules --------------------------------------------------------------------------------
ansys/2025_R1 conda/24.9.2 delft3d-fm/1 hpcx-modules/2.20 julia/1.11.3 spack-modules/113 (L)
aocc/openmpi/4.1.5 cuda-toolkit/12.4.1 (L) fsl/6.0.7.16 hpcx-modules/2.22.1 (D) matlab/R2024b
aocc/4.2.0 delft3d-delft3d4/1 gcc/13.3.1 (L,D) intel-oneapi-toolkit-modules/2024.2 matlab_parallel_server/R2024b
aocc/5.0.0 (D) delft3d-flow2d3d/1 gcc/14.2.1 intel-oneapi-toolkit-modules/2025.0.1 (D) ohpc-modules/2
Where:
D: Default Module
L: Module is loaded
Module defaults are chosen based on Find First Rules due to Name/Version/Version modules found in the module tree.
See https://lmod.readthedocs.io/en/latest/060_locating.html for details.
If the avail list is too long consider trying:
"module --default avail" or "ml -d av" to just list the default modules.
"module overview" or "ml ov" to display the number of modules for each name.
Use "module spider" to find all possible modules and extensions.
Use "module keyword key1 key2 ..." to search for all possible modules matching any of the "keys".
This will list the names and versions of selected modules available on the service. Not all of them may work in your account though due to, for example, licencing restrictions. You will notice that for many modules we have more than one version, each of which is identified by a version number. One of these versions is the default. As the service develops the default version will change.
To list the names of all modules available on the system, run the command below. These are normally hidden as these modules are typically not directly used by users. Instead, these modules are usually automatically loaded as dependencies of other modules. Using this command is not recommended for users due to the excessive number of modules.
module --show_hidden avail
You can list all the modules of a particular type by providing an
argument to the module avail command. For example, to list all
available versions of the Intel libraries, compilers and tools:
[apraveen@uop-hpc ~]$ module avail intel-oneapi-toolkit-modules
-------------------------------------------------------------------------------- /scratch/software/modules --------------------------------------------------------------------------------
intel-oneapi-toolkit-modules/2024.2 intel-oneapi-toolkit-modules/2025.0.1 (D)
...
If you want more info on any of the modules, you can use the module help command:
[apraveen@uop-hpc ~]$ module help python
-------------------------------------------------------------------- Module Specific Help for "python/3.11.9-spkc57n" ---------------------------------------------------------------------
Name : python
Version: 3.11.9
Target : zen4
The Python programming language.
Loading, unloading and swapping modules
To load a module to use module add or module load. For example,
to load the default Intel compilers and their dependencies into the development environment:
module load intel-oneapi-toolkit-modules umf tbb compiler-rt compiler
If you need a specific version of each module, you can add more information:
module load intel-oneapi-toolkit-modules/2025.0.1 umf/0.9.1 tbb/2022.0 compiler-rt/2025.0.4 compiler/2025.0.4
This will load the specified versions of the packages. If you
want to clean up, module remove will remove a loaded module:
module remove intel-oneapi-toolkit-modules/2025.0.1 umf/0.9.1 tbb/2022.0 compiler-rt/2025.0.4 compiler/2025.0.4
(or module rm tbb compiler-rt compiler or
module unload tbb compiler-rt compiler) will unload what ever version of
intel (even if it is not the default) you might have
loaded. There are many situations in which you might want to change the
presently loaded version to a different one, such as trying the latest
version which is not yet the default or using a legacy version to keep
compatibility with old data. This can be achieved most easily by using
module swap oldmodule newmodule.
Available Compiler Suites
Note
As lovelace uses dynamic linking by default you will generally also need to load any modules you used to compile your code in your job submission script when you run your code.
GCC Compiler Suite
The GCC 13.3.1 compiler suite is available by default. This is because the gcc/13.3.1 module is loaded by default.
The compilers are available as:
gfortran- Fortrangcc- Cg++- C++
AMD Optimizing C/C++ and Fortran Compilers
The AOCC compiler suite is also available. This can be loaded with:
module load aocc
The compilers are then available as:
flang- Fortranclang- Cclang++- C++
Intel® oneAPI Compilers
The Intel oneAPI compiler suite is available as in the example above. This can be loaded with:
module load intel-oneapi-toolkit-modules umf tbb compiler-rt compiler
The compilers are then available as:
ifx- Fortranicx- Cicpx- C++
Compiling MPI codes
Lovelace supports multiple MPI libraries including OpenMPI and Intel MPI. Both are available as modules.
Using GCC Compilers and openmpi
OpenMPI 5.0.5 is available by default. This is because the OpenMPI module is loaded by default.
The MPI compilers are then available as
mpifort- Fortran with MPImpicc- C with MPImpicxx- C++ with MPI
Summary
The name of the wrapper script depends on the compiler suite you are using. In summary:
Language |
Intel |
GCC |
|---|---|---|
Fortran |
mpiifort |
mpifort |
C++ |
mpiicpc |
mpicxx |
C |
mpiicc |
mpicc |
Compiler Information and Options
Help is available for the different compiler suites
- GCC
Fortran
gfortran --helporman gfortran, C/C++gcc --helporman gcc- Intel
Fortran
man ifx, C/C++man icx
Useful compiler options
Whilst difference codes will benefit from compiler optimisations in different ways, for reasonable performance on Lovelace, at least initially, we suggest the following compiler options:
- Intel
-O2- GNU
-O2 -funroll-loops -march=native
Spack Modules
Most software modules provided on the Lovelace cluster are generated by the Spack package manager. These modules themselves are provided through the spack-modules metamodule. From time to time, the Spack modules on the Lovelace cluster may be regenerated (e.g. due to the release of a new compiler version). In the event that you would like to build and run your application against a fixed version of these modules, you can note down the version of the spack-modules metamodule that is in use (e.g. spack-modules/113). To use only modules (currently with the exception of the gcc module) provided in this version (and thus, a largely fixed set of binaries), you could use the command below:
module purge
module load spack-modules/113
module load gcc openmpi
You should replace the version number above (113) with the number you noted down.
Software Requests
Please contact the HPC Admin Team if you would like to request or suggest for software to be installed onto the cluster.