Releases: tikk3r/flocs
LOFAR-SKSP 3.3.5
Fixes
- Add LofarStMan to
LD_LIBRARY_PATH
Changes
- Disable AVX512 by default as AMD doesn't support it
LOFAR-SKSP 3.3.4
Fixes
- Fix LofarStMan being removed after installation...
- Fix OpenBLAS and OMP threading variables not getting set to the number of cores.
- Fix ddf_public recipe setting OpenBLAS and OMP threading variables to 1.
- Fix AVX512 flag to actually work.
- Fix some CMake-related paths
Updates
- Update DP3 to 5.2.0 (
8e662f6f
) - Update EveryBeam to
083357c1
- Update IDG to
f178b254
- Update RMextract to
v0.4.4
- Update WSClean to 3.1 (
c22ab355
)
Changes
- Base and non-base recipes merged into a single Singularity recipe.
LOFAR-SKSP 3.3.3
Fixes
HAS_CUDA
andHAS_MKL
are now evaluated properly during build and set properly in the containers
Updates
- Added legacystamps Python package.
Changes
$LOFARROOT/bin/NDPPP
is now a symlink to $INSTALLDIR/DP3/bin/DP3.- Add BLAS libraries explicitely to buid parameters to ensure linking to the correct ones.
LOFAR-SKSP 3.3.1
Updates
- Bump DP3 to e5eff2e
- Bump EveryBeam to 14f35c85 (newer than 0.2.2).
- Bump WSClean to c0a654ca
- Fix DDFacet CPU patch.
Changes
- OMP_MAX_THREADS, OMP_MAX_THREADS and OPENBLAS_MAX_THREADS are now set automatically to the available number of cores.
LOFAR-SKSP 3.3
Updates
- DP3 updated to v5.1
- EveryBeam set to 0c26c65d
- WSClean set to 3.0 (41c651d1 specifically)
- Add PyLRU Python package
- Add GDB
- Manually build HDF5 with thread safety (useful for WSClean facet functionality).
- Fixed patch for DDFacet's AsyncProcessPool.py to make it more compatible with Slurm
Changes
- Python environment set up slightly differently.
- Architecture related settings march and mtune are now specified more flexibly through environment variables.
- MKL and CUDA can now be set through environment variables to reduce the clutter of recipes.
- Move some DNF installs to the base container where they belong.
LOFAR-SKSP 3.2
This update changes the container base to Fedora 31 for a newer GCC version that better supports and optimizes for recent architectures (specifically AVX512). Multiple recipes are now provided as well for the base container to be built with MKL and/or CUDA (relevant for IDG).
Updates
- Container is now Fedora 31 instead of 27
- Updated CUDA to 11.3
- Updated DS9 to Fedora 30 8.2.1
- Updated HDF5 to 1.10.5
- Updated WSClean to 2.10.1
- Added astLib, astro-kittens, astro-pyxis, astropy-healpix, astroquery, bcrypt, beautifulsoup4, cffi, crytography, deap, emcee, entrypoints, enum34, ephem, future, html5lib, ipython, keyring, lxml, meqtrees-cattery, montage_wrapper, mysqlclient, owlcat, paramiko, pathlib2, pexpect, pickleshare, Pillow, Polygon2, prettytable, prompt-toolkit, psutil, ptyprocess, purr, py-cpuinfo, pybind11, pycparser, pyephem, pyFFTW, pyfits, Pygments, PyMySQL, PyNaCl, reproject, scandir, SecretStorage, SharedArray, simplegeneric, soupsieve, sshtunnel, traitlets, wcwidth and webencodings Python 2 packages.
- Added ddf recipe that uses the public branches
- Added msoverview
- Added back ParmDB functionality
- Added back support for non-dysco measurement sets
Changes
- Removed MKL_DEBUG_CPU_TYPE workaround for AMD as it no longer works
- Make building for specific architectures simpler
LOFAR-SKSP 3.1
Update recipes to include new dependencies and other miscellaneous changes in e.g. branches or general structure.
Updates
- Container is now based on Fedora 27 instead of CentOS 7
- Python packages moved to requirements.txt and requirements3.txt for Python 2 and 3 respectively
- Both Python 2 and 3 now included through virtualenvs
- Added Montage
- Added EveryBeam library
- Added small test script to assess basic functionality of the container
- Updated HDF5 to 1.8.18
- Updated Intel MKL to 2020.0-088
- Updated WSClean repository to Gitlab
LOFAR-SKSP 3.0
Container: https://lofar-webdav.grid.sara.nl/software/shub_mirror/tikk3r/lofar-grid-hpccloud/lofar_sksp-3.0.sif
Version bump to 3.0.0 to synchronize major version with prefactor v3.0.
This container provides an installation of the LOFAR software in order to run the genericpipeline, along with additional software that is required to run prefactor and potentially other pipelines. The pipelines themselves are purposely not included. The software versions are intended to be compatible with the prefactor 3.0 releas.
Updates
- Added difmap
- Added nano
- Added vim
- All AVX512 instructions disabled
- Freeze most software versions for prefactor 3.0
Bug fixes
- Fix PyTables installation for Python 2.7 causing other packages to fail