* petsc: add a +mkl-pardiso variant
mkl_pardiso solver is distributed with intel-mkl
* petsc: depend on mkl instead of intel-mkl
Co-authored-by: Adam J. Stewart <ajstewart426@gmail.com>
Co-authored-by: Adam J. Stewart <ajstewart426@gmail.com>
Add version 4.12.6, 5.0.3
I think, the preferred was there to keep version 4.
But that's why we have spack, because people can install
whatever version they want.
And root has a properly versioned dependency.
* mumps: Fix for problematic src/makefile patch (#20590)
Minor change in src/Makefile between 5.2.0 and 5.3.3 causing patch to
break. Split into 2 patchfiles
* mumps: Additional patch for fixing #20590
This is to fix issue wherein build fails on Ubuntu due to undefined
symbols, despite symbols being included in other libraries referenced
on the compilation line. I believe the issue is that the inclusion
of libsmumps.so was (due to my original patch) causing
libmumps_common.so to be automatically loaded, but since libpords.so
was not also required, the error was occuring. I have added libpords.so
along with libmumps_common.so to be explicit dependencies of
libsmumps.so, etc., which seems to resolve the issue.
* ArrayFire: Add version 3.7.2.
* ArrayFire: Allow using MKL as the FFTW provider.
* ArrayFire: Ensure the libraries are properly found.
The required backend(s) can be specified in the library query.
* openssl: remove preprocessor flags incompatible with NVIDIA HPC SDK
* Update var/spack/repos/builtin/packages/openssl/package.py
Co-authored-by: Adam J. Stewart <ajstewart426@gmail.com>
Co-authored-by: Scott McMillan <smcmillan@nvidia.com>
Co-authored-by: Adam J. Stewart <ajstewart426@gmail.com>
* [py-statsmodels] added version 0.12.1 and updated dependencies accordingly
* [py-statsmodels] added python requirements for new version and fixed formatting for readability
* added m4 dep to PVM recipe
* added libtirpc dep to PVM recipe
* decode str or bytestr string to unicode
* Resolved comments from @adamjstewart on setup_build_environment
* When the SCR spec specifies a resource_manager=SLURM or LSF flag, propagate the spec through to
the libyogrt scheduler=slurm or lsf
* Use libyogrt default scheduler option when the SCR spec does not specify LSF or SLURM
* updated relion for new versions
* Switched to checksum versions
* Enabled spack tracking for MKL and TBB when CPU optimizations are enabled
* Added variants to control MKL FFT and Ppatent feature
* Replaced tags with sha256 for older versions an and switched to virtual packages
* py-funcy: new recipe
* Update var/spack/repos/builtin/packages/py-funcy/package.py
add build and run python dependencies
Co-authored-by: Adam J. Stewart <ajstewart426@gmail.com>
Co-authored-by: Adam J. Stewart <ajstewart426@gmail.com>
* sbang pushed back to callers;
star moved to util.lang
* updated unit test
* sbang test moved; local tests pass
Co-authored-by: Nathan Hanford <hanford1@llnl.gov>
* fixing outdated metis link
* updated url to the official website since the previous url was a GitHub repo that is an unofficial mirror that only contains the latest version
* py-dictdiffer: new recipie
* Update var/spack/repos/builtin/packages/py-dictdiffer/package.py
add correct setuptools dependency
Co-authored-by: Adam J. Stewart <ajstewart426@gmail.com>
Co-authored-by: Adam J. Stewart <ajstewart426@gmail.com>
* Update NEURON simulator package
- update recipe to support autoconf as well as cmake
- new versions >=7.8 support cmake
- remove old variants
- added patch for latest bug fix release 7.8.2
Co-authored-by: Kumbhar Pramod Shivaji <kumbhar@bbpv1.epfl.ch>
Co-authored-by: Kumbhar Pramod Shivaji <kumbhar@bb-c02vf1h0hv2r.epfl.ch>
* NAMD: FIX build +cuda
Hi,
If I try to compile NAMD with CUDA support, it fails because cannot file the file "{self.arch}.cuda" because it is undet the "arch" folder.
* NAMD: FIX mpi ~smp
Fix `spack install namd ^charmpp backend=mpi ~smp`
* ssht: New version 1.3.4
ssht changed configuration mechanism from "home-grown" to "cmake. The previously current version 1.2b1 (a beta release) is thus unfortunately not available any more.
* ssht: Don't set build type
Co-authored-by: Massimiliano Culpo <massimiliano.culpo@gmail.com>
Co-authored-by: Massimiliano Culpo <massimiliano.culpo@gmail.com>
* Don't use CUDA for hipblas
* old versions use TRY_CUDA
* Update var/spack/repos/builtin/packages/hipblas/package.py
Co-authored-by: Adam J. Stewart <ajstewart426@gmail.com>
Co-authored-by: Adam J. Stewart <ajstewart426@gmail.com>
* add version 2.2-2 to r-gwmodel
* Update var/spack/repos/builtin/packages/r-gwmodel/package.py
Fix comma, space issue.
Co-authored-by: Adam J. Stewart <ajstewart426@gmail.com>
Co-authored-by: Adam J. Stewart <ajstewart426@gmail.com>
* add version 0.3.17 to r-inline
* Drop R version constraint
A really old version of R was specified in the 0.3.14 and 0.3.15
versions of r-inline. This constraint was dropped in the 0.3.17 version.
Drop it from the spack recipe as well.
* vasp: fix build with gfortran 10
Avoid Error: Type mismatch between actual argument at (1) and actual argument at (2)
* vasp: add version 6.1.1
* vasp 6: allow building without CUDA
* Adding PiP recipe
* pip@1 recipe (it seems working)
* change install dir hierarchy
* installing PiP man pages
* add pip-glibc & pip-gdb
* fix configure option designations, fix dependency types
* fix dependency type of pip
* use AutotoolsPackage in pip recipe
* add patch for pip-glibc & pip-gdb to enable 'disable-werror'
* change glibc install directory
* add linux distro check to pip-gdb
* create process-in-process package
* use flag_handler and join_path
* add gcc version constraint, change install-test to check-installed
* fix gcc version designations on conflicts()
* add constraint of target cpu, fix flake8 warnings
* add version constraint to resource()
* Some fixes to adapt the current version
not to execute 'piplnlibs'
change documentation install command
* Update
new branch name of PiP-gdb
adapting PiP-Testsuite
* update pip-gdb github urls
* The very first commit of Process-in-Process (PiP)
details can be found at https://github.com/RIKEN-SysSoft/PiP
* Fix comment style issues
* New Package: Process-in-Process (PiP) -- 2nd trial
* fix style issue
* change inline comments style (required to have two spaces)
Co-authored-by: Daiki Matsunaga <daikim@axe.bz>
Imagemagick-7.0.8 needs to link against libltdl. Otherwise, the build will fail with:
```
2 errors found in build log:
503 checking for libltdl...
504 checking ltdl.h usability... no
505 checking ltdl.h presence... no
506 checking for ltdl.h... no
507 checking for lt_dlinit in -lltdl... no
508 checking if libltdl package is complete... no
>> 509 configure: error: in `/tmp/gpjohnsn/spack-stage/spack-stage-imagemagick-7.0.8-7-4y44gaklhhciiwjzhfpxjfwdj5q
ltjp3/spack-src':
>> 510 configure: error: libltdl is required for modules and OpenCL builds
511 See `config.log' for more details
```
* add version 3.8.2 to r-gtools
* Improve formatting of description
In case the list gets formatted as a non-list:
- added semicolons to end of list items
- replaced dashes with [#]
* add version 1.30 to r-knitr
* Fix version constraints
- r-digest
- r-formatr
The version constraints on those packages should actually be in the `when`
clause.
'date' is a C++ header library offering extensive date and time
functionality for the C++11, C++14 and C++17 standards written by Howard
Hinnant and released under the MIT license. A slightly modified version
has been accepted (along with 'tz.h') as part of C++20. This package
regroups all header files from the upstream repository by Howard Hinnant
so that other R packages can use them in their C++ code. At present, few
of the types have explicit 'Rcpp' wrapper though these may be added as
needed.
Designed to ease the application and comparison of multiple hypothesis
testing procedures for FWER, gFWER, FDR and FDX. Methods are
standardized and usable by the accompanying 'mutossGUI'.
Utility functions that enhance the 'parallel' package and support the
built-in parallel backends of the 'future' package. For example,
availableCores() gives the number of CPU cores available to your R
process as given by the operating system, 'cgroups' and Linux
containers, R options, and environment variables, including those set by
job schedulers on high-performance compute clusters. If none is set, it
will fall back to parallel::detectCores(). Another example is
makeClusterPSOCK(), which is backward compatible with
parallel::makePSOCKcluster() while doing a better job in setting up
remote cluster workers without the need for configuring the firewall to
do port-forwarding to your local computer.
