Releases: cdslaborg/paramonte
Auxiliary external dependency binaries
This pre-release only contains the ParaMonte library's optional auxiliary software and libraries, most importantly, the relevant Intel MPI runtime libraries. Users can download these libraries for installation on their local systems from this page or directly from the relevant parent websites.
To download the Intel runtime libraries directly (instead of this page), see the original Intel website. Note that the Intel web pages frequently change address. For this reason, important runtime library install binaries are also provided here for users' convenience.
Auxiliary binaries for ParaMonte 2 releases:
-
Remarks on the Intel MPI library for Windows intel-mpi-2021.14.0.785_offline.exe.
- This binary is only relevant to the ParaMonte 2 library releases.
- Once the library is installed on the system, the runtime library variables must be loaded in the Windows Batch command line environment by running the Batch script located at
C:\Program Files (x86)\Intel\oneAPI\setvars.bat
.
To execute this script, open a Windows Batch command line, then type:"C:\Program Files (x86)\Intel\oneAPI\setvars.bat"
-
Remarks on the Intel MPI library for Linux intel-mpi-2021.14.0.791_offline.sh.
- This binary must be executed on a Bash-compatible command line.
You can download it via the command line applications such aswget
:or thewget https://github.com/cdslaborg/paramonte/releases/download/auxil/intel-mpi-2021.14.0.791_offline.sh
curl
application:curl -OL https://github.com/cdslaborg/paramonte/releases/download/auxil/intel-mpi-2021.14.0.791_offline.sh
- Once the binary is downloaded, ensure it is executable by typing the following on the command line:
chmod +x intel-mpi-2021.14.0.791_offline.sh
- Run the executable on the command line to install the Intel MPI library for Linux on your system:
./intel-mpi-2021.14.0.791_offline.sh
- Once the installation is complete, you will have to run a Bash script,
setvars.sh
orvars.sh
, that defines all relevant environment variables.
This script is frequently, though not always, located in the following directory:source ~/intel/oneapi/setvars.sh
- This binary must be executed on a Bash-compatible command line.
-
Remarks on the Intel
ifort
Fortran compiler runtime library for Windows w_ifort_runtime_p_2024.2.1.1084.exe.- This is the last runtime library released by Intel for the
ifort
Fortran compiler, which is now removed from the Intel products and replaced by the new Intel Fortran compilerifx
. - This runtime library is provided here to ensure its longevity and accessibility.
- Despite its availability here, note that all Intel runtime libraries (except MPI) are automatically statically linked when building the ParaMonte library, resolving the need to install this runtime library post-ParaMonte installation.
- This is the last runtime library released by Intel for the
Auxiliary binaries for ParaMonte 1 releases:
-
Remarks on the Intel MPI runtime library for Windows w_mpi-rt_p_2019.4.245.exe.
- This binary is only relevant to the ParaMonte 1 library releases.
- This is an old release of Intel MPI runtime libraries that may not be compatible with newer releases of Intel products or the ParaMonte library.
-
Remarks on the Intel MPI runtime library for Linux l_mpi-rt_2018.2.199.tgz.
- This binary must be executed on a Bash-compatible command line.
You can download it via the command line applications such aswget
:or thewget https://github.com/cdslaborg/paramonte/releases/download/auxil/l_mpi-rt_2018.2.199.tgz
curl
application:curl -OL https://github.com/cdslaborg/paramonte/releases/download/auxil/l_mpi-rt_2018.2.199.tgz
- Once the binary is downloaded, untar it as in the following on the command line:
tar xvzf l_mpi-rt_2018.2.199.tgz
- Run the executable on the command line to install the Intel MPI runtime library for Linux on your system:
cd l_mpi-rt_2018.2.199 && ./install.sh
- This binary must be executed on a Bash-compatible command line.
Developer Warning
This particular pre-release must never be marked as an actual ParaMonte library release, as its contents are frequently updated with the relevant ParaMonte library dependencies. At the same time, it must never contain a specific actual ParaMonte library release.
🚀 ParaMonte 2 C, C++, Fortran prerelease
Version 2024.04.29 -- C/C++/Fortran prerelease
Major
This is a significant revision of the original ParaMonte 1 library toward the full release of the new ParaMonte 2 library.
Compiler Compatibility.
Compiler Suite | Windows (amd64) | Linux (amd64) | macOS (amd64) | macOS (arm64) |
---|---|---|---|---|
GNU Compiler Collection >= 10.3 | ✅ | ✅ | ✅ | ✅ |
Intel OneAPI >= 2021.8.0 | ✅ | ✅ | ✅ | ❌ |
WARNING:
On macOS systems, you must download the prebuilt libraries via thecurl
software as in the following examples:For macOS ARM64 (M1, M2, M3, etc.):
libname=libparamonte_matlab_darwin_arm64 curl -OL https://github.com/cdslaborg/paramonte/releases/download/2024.04.29/$libname.tar.gz tar xvzf $libname.tar.gz && cd $libnameFor macOS AMD64 (Intel chip):
libname=libparamonte_matlab_darwin_amd64 curl -OL https://github.com/cdslaborg/paramonte/releases/download/2024.04.29/$libname.tar.gz tar xvzf $libname.tar.gz && cd $libnameReplace the library names with any other desired download names. Do NOT download the ParaMonte library for macOS through > your web browser.
v1.5.2
Version 1.5.2 -- March 7, 2024
Minor Enhancements
This is a very minor, minimal final release of the original ParaMonte library before the impending new ParaMonte 2 library release.
Compiler Compatibility
Compiler Suite | Windows (64bit) | Linux | macOS |
---|---|---|---|
GNU Compiler Collection > 8.4 | ✅ | ✅ | ✅ |
Intel Parallel Studio > 19.1.1 | ✅ | ✅ | ✅ |
Microsoft C/C++ Compiler > 16.0.0 | ✅ | ❌ | ❌ |
Compiler / MPI library used for this binary release
- Windows: Intel Parallel Studio Version 19.1.1.216 Build 20200306 / Intel(R) MPI Library 2019 Update 7 for Windows
- Linux: Intel Parallel Studio Version 19.1.1.217 20200306 / Intel(R) MPI Library for Linux OS, Version 2019 Update 7 Build 20200312
- Linux: GNU 10.1.0 / Open-MPI 4.0.3
- Linux: GNU 10.1.0 / MPICH 3.2
- macOS: Intel Parallel Studio Version 19.1.0.216 20200306
- macOS: GNU 10.2.0 / Open-MPI 4.0.5
- macOS: GNU 10.2.0 / MPICH 3.3.2
v1.5.1
Version 1.5.1 -- January 1, 2021
Minor Enhancements
-
This is a minor enhancement release, but is a major step toward
further portability of the ParaMonte library. All ParaMonte kernel
library dependencies are now properly handled and recognized at runtime
without such aggressive actions as permanently redefining the environmental
path variables, most importantly,PATH
andLD_LIBRARY_PATH
on Linux/macOS. -
The ParaMonte kernel routines are now capable of handling user-input
file paths that contain white-space (blank) or other exotic characters. -
The Bash build script for the ParaMonte C/C++/Fortran examples can now handle
file paths that contain white-space (blank) or other exotic characters. -
The shared (dynamic) library file naming convention is now changed from
using "dynamic" in the file name to using "shared" in the file name. -
Typo-fixes in the documentation of the library.
Compiler Compatibility
Compiler Suite | Windows (64bit) | Linux | macOS |
---|---|---|---|
GNU Compiler Collection > 8.4 | ✅ | ✅ | ✅ |
Intel Parallel Studio > 19.1.1 | ✅ | ✅ | ✅ |
Microsoft C/C++ Compiler > 16.0.0 | ✅ | ❌ | ❌ |
Compiler / MPI library used for this binary release
- Windows: Intel Parallel Studio Version 19.1.1.216 Build 20200306 / Intel(R) MPI Library 2019 Update 7 for Windows
- Linux: Intel Parallel Studio Version 19.1.1.217 20200306 / Intel(R) MPI Library for Linux OS, Version 2019 Update 7 Build 20200312
- Linux: GNU 10.1.0 / Open-MPI 4.0.3
- Linux: GNU 10.1.0 / MPICH 3.2
- macOS: Intel Parallel Studio Version 19.1.0.216 20200306
- macOS: GNU 10.2.0 / Open-MPI 4.0.5
- macOS: GNU 10.2.0 / MPICH 3.3.2
v1.5.0
Version 1.5 -- Dec 17, 2020
Major Enhancements
- This version introduces numerous performance and accuracy enhancements to the ParaMonte library.
- The entire kernel library is now fully documented and verified with over 866 tests that cover close
to 100% of all lines and functions in the kernel. - New prebuilt libraries with GNU compilers and Open-MPI on Linux and macOS are added.
- New flags are now added to the build scripts of the library that automate the process of code coverage generation.
- The
testing
builds are now removed from the ParaMonte release page as this build is mostly useful for development purposes. - The issue of Windows file locking, that led to the occasional crashes of the
ParaDRAM and ParaDISE simulations inmultiChain
parallelism mode, is now resolved.
Minor Enhancements
- Minor enhancements to the ParaMonte C/C++/Fortran example build scripts
build.sh
andbuild.bat
. - The default build settings are now limited to
heap
memory allocation withdynamic
library builds
for only serial and MPI parallelization for all languages. Thestack
memory allocation results in
a ~10% gain in the efficiency of the code. The benefits of stack-memory builds are marginal and are
often problematic, in particular, for usage with non-compiled languages. However, users can still
build the library withstack
memory allocation by specifying the appropriate build flags with
theinstall.sh
on Unix orinstall.bat
script on Windows systems. For further information,
see the installation guidelines on the ParaMonte documentation website. - All temporary array creations in debug mode are now resolved,
except when Intel compilers are used, in which case, the debug warning messages are silenced.
Compiler Compatibility
Compiler Suite | Windows (64bit) | Linux | macOS |
---|---|---|---|
GNU Compiler Collection > 8.4 | ✅ | ✅ | ✅ |
Intel Parallel Studio > 19.1.1 | ✅ | ✅ | ✅ |
Microsoft C/C++ Compiler > 16.0.0 | ✅ | ❌ | ❌ |
Compiler / MPI library used for this binary release
- Windows: Intel Parallel Studio Version 19.1.1.216 Build 20200306 / Intel(R) MPI Library 2019 Update 7 for Windows
- macOS: Intel Parallel Studio Version 19.1.0.216 20200306
- macOS: GNU 10.2.0 / Open-MPI 4.0.5
- macOS: GNU 10.2.0 / MPICH 3.3.2
- Linux: Intel Parallel Studio Version 19.1.1.217 20200306 / Intel(R) MPI Library for Linux OS, Version 2019 Update 7 Build 20200312
- Linux: GNU 10.2.0 / Open-MPI 4.0.3
- Linux: GNU 10.1.0 / MPICH 3.2
v1.4.1
Version 1.4.1 -- November 15, 2020
Enhancements
- The ParaMonte C/C++/Fortran example build scripts
build.sh
andbuild.bat
now check for the existence of both Intel and GNU compilers in the appropriate
order that is automatically inferred at the compilation time. Also, the dependencies
on the MPI compiler wrappers is now removed as the MPI libraries are not required to
build the ParaMonte examples, even in cases of parallel ParaMonte example builds.
Compiler support
Compiler Suite | Windows (64bit) | Linux | macOS |
---|---|---|---|
GNU Compiler Collection > 7.0 | ✅ | ✅ | ✅ |
Intel Parallel Studio > 18.0.0 | ✅ | ✅ | ✅ |
Microsoft C/C++ Compiler > 16.0.0 | ✅ | ❌ | ❌ |
Compiler / MPI library used for this binary release
- Windows: Intel Parallel Studio Version 19.0.4.245 Build 20190417 / Intel(R) MPI Library 2019 Update 4 for Windows
- Linux: Intel Parallel Studio Version 18.0.2 20180210 / Intel(R) MPI Library for Linux OS, Version 2018 Update 2 Build 20180125
- Linux: GNU 10.1.0 / MPICH 3.2
- macOS: Intel Parallel Studio Version 19.1.0.216 20200306
- macOS: GNU 10.2.0 / Open-MPI 4.0.5
v1.4.0
Version 1.4.0 -- October 29, 2020
Enhancements
-
The IO debugging info of all ParaMonte samplers have been enhanced.
In cases of wrong syntax or syntax-breaking input values in the simulation
output files, the error messages are now more informative and point directly
to the exact location of of error in the input file. -
The Integrated Autocorrelation (IAC) for sample refinement in ParaDRAM
sampler of ParaMonte is now set to the average of all variables' IAC values
instead of the maximum IAC value. This will lead to less aggressive decorrelation
of the final sample, which means significantly larger final sample sizes, without
compromising the i.i.d. property of the final refined sample. This behavior can
be reversed back to the original by specifying "max" or "maximum" along with
the requested refinement method,SampleRefinementMethod = "batchmeans max"
orSampleRefinementMethod = "BatchMeans-max"
(case-insensitive).
Compiler support
Compiler Suite | Linux | macOS | Windows (64bit) |
---|---|---|---|
GNU Compiler Collection > 7.1 | ✓ | ✓ | ✗ |
Intel Parallel Studio > 18.0.0 | ✓ | ✓ | ✓ |
Compiler / MPI library used for this binary release
- Windows: Intel Parallel Studio Version 19.0.4.245 Build 20190417 / Intel(R) MPI Library 2019 Update 4 for Windows
- Linux: Intel Parallel Studio Version 18.0.2 20180210 / Intel(R) MPI Library for Linux OS, Version 2018 Update 2 Build 20180125
- Linux: GNU 9.1 / MPICH 3.2
- macOS: Intel Parallel Studio Version 19.1.0.216 20200306
- macOS: GNU 10.2.0 / Open-MPI 4.0.5
v1.3.0
Version 1.3.0 -- October 10, 2020
Enhancements
- A new simulation specification
overwriteRequested
has
been added to all ParaMonte samplers. If TRUE and the
ParaMonte sampler detects an existing set of old simulation
output files in the output path of the current simulation with
the same names as the output file names of the current simulation,
then, the ParaMonte sampler will overwrite the existing simulation files.
Compiler support
- Intel Parallel Studio (>2018.0.0)
- GNU Compiler Collection (>7.0.0)
Compiler / MPI library used for this binary release
- Windows: Intel Parallel Studio Version 19.0.4.245 Build 20190417 / Intel(R) MPI Library 2019 Update 4 for Windows
- Linux: Intel Parallel Studio Version 18.0.2 20180210 / Intel(R) MPI Library for Linux OS, Version 2018 Update 2 Build 20180125
- Linux: GNU 9.1 / MPICH 3.2
- macOS: Intel Parallel Studio Version 19.1.0.216 20200306
- macOS: GNU 10.2.0 / Open-MPI 4.0.5
Version 1.2.0
Version 1.2.0 -- September 22, 2020
Enhancements
-
The post-processing report in the output report file
of ParaDRAM simulation has been significantly improved:- The parallel simulation summary now also provides the
predicted strong-scaling speedup behavior of the parallel
ParaDRAM simulations in "single chain" parallelism mode.
This can help make wiser decisions regarding the the number
of processors for similar parallel simulations in the future.
- The parallel simulation summary now also provides the
-
The ParaDRAM restart output file in ASCII mode now contains all
proposal updates, including the first user-specified proposal specs. -
All parallel simulations now avoid the unnecessary creation of
temporary files by all processors for System and OS operations.
This is particularly important for large-scale parallel simulations.
As a side effect, this will also potentially improve the runtime
performances of the simulation. -
Major enhancements has been made to the parallel simulation
performance analysis reported in the post-processing section
of the ParaDRAM simulation output_report.txt
files. -
The ParaMonte C/C++ build processes are now separate from each other.
Bug fixes
- minor typo fixes
Compiler support
- Intel Parallel Studio (>2018.0.0)
- GNU Compiler Collection (>7.0.0)
Compiler / MPI library used for this binary release
- Windows: Intel Parallel Studio Version 19.0.4.245 Build 20190417 / Intel(R) MPI Library 2019 Update 4 for Windows
- Linux: Intel Parallel Studio Version 18.0.2 20180210 / Intel(R) MPI Library for Linux OS, Version 2018 Update 2 Build 20180125
- Linux: GNU 9.1 / MPICH 3.2
- macOS: Intel Parallel Studio Version 19.1.0.216 20200306
- macOS: GNU 10.2.0 / Open-MPI 4.0.5
Version 1.1.0
Version 1.1.0 -- May 27, 2020
New features
- ParaMonte Interface to MATLAB in addition to the currently-available C/C++/Fortran/Python Programming languages.
Enhancements
- The simulation summary report in the output report file has been improved.
- The error handling in the build Batch-scripts on Windows is now greatly improved.
Bug fixes
- The build Batch-script for the ParaMonte examples on Windows now properly builds and runs coarray applications in parallel.
- The fully-deterministic restart functionality is now functional also when chainFileFormat="verbose" in ParaDRAM simulations.
Compiler support
- Intel Parallel Studio (>2018.0.0)
- GNU Compiler Collection (>7.0.0)
Compiler / MPI library used for this binary release
- Windows: Intel Parallel Studio Version 19.0.4.245 Build 20190417 / Intel(R) MPI Library 2019 Update 4 for Windows
- Linux: Intel Parallel Studio Version 18.0.2 20180210 / Intel(R) MPI Library for Linux OS, Version 2018 Update 2 Build 20180125
- Linux: GNU 8.3 / MPICH 3.2
- macOS: Intel Parallel Studio Version 19.1.0.166 20191121
- macOS: GNU 9.3 / Open-MPI 4.0
For more information on the usage, visit cdslab.org/paramonte/notes/installation/readme/