Taurus: SCS5 Update #2667
Taurus: SCS5 Update #2667
Conversation
|
|
||
| # Environment ################################################################# | ||
| # | ||
| #export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:$BOOST_LIB | ||
|
|
||
| export PNGWRITER_ROOT=$HOME/lib/pngwriter |
There was a problem hiding this comment.
New spellings:
PNGwriter_DIR
Splash_DIR
There was a problem hiding this comment.
Update export PNGWRITER_ROOT=$HOME/lib/pngwriter to export PNGwriter_DIR=$HOME/lib/pngwriter.
There was a problem hiding this comment.
Correspondingly update export SPLASH_ROOT=$HOME/lib/splash to export Splash_DIR=$HOME/lib/splash.
There was a problem hiding this comment.
Make automatically consistent (only one path needs to be adjusted then):
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:$HOME/lib/pngwriter/lib/ export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:$HOME/lib/splash/lib/
to
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:$PNGwriter_DIR/lib/ export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:$Splash_DIR/lib/
There was a problem hiding this comment.
In general: This proposed solution is messy, but it seems to work.
- The proposed module environment is unfortunately bundled (Boost,HDF5,gnuplot) with the
foss/2018amodule. This introduces a lot of unnecessary modules (51 modules total). The currentmodule list-command results in:
Derzeit geladene Module:
1) modenv/scs5 (S) 14) hwloc/1.11.8-GCCcore-6.4.0 27) gettext/0.19.8.1-GCCcore-6.4.0-libxml2-2.9.7 40) GMP/6.1.2-GCCcore-6.4.0
2) GCCcore/6.4.0 15) OpenMPI/2.1.2-GCC-6.4.0-2.28 28) PCRE/8.41-GCCcore-6.4.0 41) nettle/3.4-GCCcore-6.4.0
3) OpenBLAS/0.2.20-GCC-6.4.0-2.28 16) cURL/7.58.0-GCCcore-6.4.0 29) util-linux/2.31.1-GCCcore-6.4.0 42) libdrm/2.4.91-GCCcore-6.4.0
4) gompi/2018a 17) expat/2.2.5-GCCcore-6.4.0 30) GLib/2.54.3-GCCcore-6.4.0 43) LLVM/5.0.1-GCCcore-6.4.0
5) FFTW/3.3.7-gompi-2018a 18) Perl/5.26.1-GCCcore-6.4.0 31) fontconfig/2.12.6-GCCcore-6.4.0 44) Mesa/17.3.6-foss-2018a
6) ScaLAPACK/2.0.2-gompi-2018a-OpenBLAS-0.2.20 19) git/2.18.0-GCCcore-6.4.0 32) X11/20180131-GCCcore-6.4.0 45) libGLU/9.0.0-foss-2018a
7) foss/2018a 20) bzip2/1.0.6-GCCcore-6.4.0 33) cairo/1.14.12-GCCcore-6.4.0 46) Qt5/5.10.1-foss-2018a
8) binutils/2.28-GCCcore-6.4.0 21) libpng/1.6.34-GCCcore-6.4.0 34) NASM/2.13.03-GCCcore-6.4.0 47) gnuplot/5.2.4-foss-2018a
9) GCC/6.4.0-2.28 22) freetype/2.9-GCCcore-6.4.0 35) libjpeg-turbo/1.5.3-GCCcore-6.4.0 48) Boost/1.66.0-foss-2018a
10) ncurses/6.0-GCCcore-6.4.0 23) pixman/0.34.0-GCCcore-6.4.0 36) libgd/2.2.5-GCCcore-6.4.0 49) Szip/2.1.1-GCCcore-6.4.0
11) CMake/3.10.2-GCCcore-6.4.0 24) libffi/3.2.1-GCCcore-6.4.0 37) HarfBuzz/1.7.5-foss-2018a 50) HDF5/1.10.1-foss-2018a
12) CUDA/9.2.88 25) XZ/5.2.3-GCCcore-6.4.0 38) Pango/1.41.1-foss-2018a 51) zlib/1.2.11-GCCcore-6.4.0
13) numactl/2.0.11-GCCcore-6.4.0 26) libxml2/2.9.7-GCCcore-6.4.0 39) libcerf/1.5-GCCcore-6.4.0
Wo:
S: Das Modul ist angeheftet. Verwenden Sie "--force", um das Modul zu entladen.
To me, this appears to be excessive and will probably invite compatibility issues in the future. Maybe we should just build boost, hdf5 and gnuplot on our own.
- There are many Boost-related warnings during cmake configuration:
CMake Warning at /software/haswell/CMake/3.10.2-GCCcore-6.4.0/share/cmake-3.10/Modules/FindBoost.cmake:801 (message):
New Boost version may have incorrect or missing dependencies and imported
targets
Call Stack (most recent call first):
/software/haswell/CMake/3.10.2-GCCcore-6.4.0/share/cmake-3.10/Modules/FindBoost.cmake:907 (_Boost_COMPONENT_DEPENDENCIES)
/software/haswell/CMake/3.10.2-GCCcore-6.4.0/share/cmake-3.10/Modules/FindBoost.cmake:1542 (_Boost_MISSING_DEPENDENCIES)
/home/s0134766/src/picongpu/include/pmacc/PMaccConfig.cmake:296 (find_package)
CMakeLists.txt:83 (find_package)
...
many more
- There is a long stream of warnings during compiling:
/home/s0134766/src/picongpu/thirdParty/alpaka/include/alpaka/vec/Vec.hpp(925): warning: calling a __host__ function("std::__shared_count<( ::__gnu_cxx::_Lock_policy)2> ::__shared_count") from a __host__ __device__ function("alpaka::dev::DevCpu::DevCpu") is not allowed
/home/s0134766/src/picongpu/thirdParty/alpaka/include/alpaka/vec/Vec.hpp(925): warning: calling a __host__ function("std::__shared_count<( ::__gnu_cxx::_Lock_policy)2> ::__shared_count") from a __host__ __device__ function("alpaka::dev::DevCpu::DevCpu [subobject]") is not allowed
/sw/installed/Boost/1.66.0-foss-2018a/include/boost/mpl/map/aux_/item.hpp(54): warning: "boost::mpl::aux::type_wrapper<T> operator/(const boost::mpl::m_item<Key, T, Base> &, boost::mpl::aux::type_wrapper<Key> *)" declares a non-template function -- add <> to refer to a template instance
/sw/installed/Boost/1.66.0-foss-2018a/include/boost/mpl/map/aux_/item.hpp(55): warning: "boost::mpl::aux::type_wrapper<boost::mpl::pair<Key, T>> operator|(const boost::mpl::m_item<Key, T, Base> &, boost::mpl::next<Base::order>::type *)" declares a non-template function -- add <> to refer to a template instance
/sw/installed/Boost/1.66.0-foss-2018a/include/boost/mpl/map/aux_/item.hpp(56): warning: "char (&operator||(const boost::mpl::m_item<Key, T, Base> &, boost::mpl::aux::type_wrapper<Key> *))[boost::mpl::next<Base::order>::type::value]" declares a non-template function -- add <> to refer to a template instance
/sw/installed/Boost/1.66.0-foss-2018a/include/boost/mpl/map/aux_/item.hpp(71): warning: "boost::mpl::aux::type_wrapper<mpl_::void_> operator/(const boost::mpl::m_mask<Key, Base> &, boost::mpl::aux::type_wrapper<Key> *)" declares a non-template function -- add <> to refer to a template instance
/sw/installed/Boost/1.66.0-foss-2018a/include/boost/mpl/map/aux_/item.hpp(72): warning: "boost::mpl::aux::type_wrapper<mpl_::void_> operator|(const boost::mpl::m_mask<Key, Base> &, boost::mpl::x_order_impl<Base, Key>::type *)" declares a non-template function -- add <> to refer to a template instance
/home/s0134766/src/picongpu/include/pmacc/../pmacc/particles/memory/dataTypes/Particle.hpp(167): warning: __device__ annotation is ignored on a function("operator=") that is explicitly defaulted on its first declaration
/home/s0134766/src/picongpu/include/pmacc/../pmacc/particles/memory/dataTypes/Particle.hpp(167): warning: __host__ annotation is ignored on a function("operator=") that is explicitly defaulted on its first declaration
...
Many more
| # | ||
| module purge |
There was a problem hiding this comment.
we should not purge the environments they set up.
|
|
||
| # Environment ################################################################# | ||
| # | ||
| #export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:$BOOST_LIB | ||
|
|
||
| export PNGWRITER_ROOT=$HOME/lib/pngwriter |
There was a problem hiding this comment.
Update export PNGWRITER_ROOT=$HOME/lib/pngwriter to export PNGwriter_DIR=$HOME/lib/pngwriter.
|
|
||
| # Environment ################################################################# | ||
| # | ||
| #export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:$BOOST_LIB | ||
|
|
||
| export PNGWRITER_ROOT=$HOME/lib/pngwriter |
There was a problem hiding this comment.
Correspondingly update export SPLASH_ROOT=$HOME/lib/splash to export Splash_DIR=$HOME/lib/splash.
|
|
||
| # Environment ################################################################# | ||
| # | ||
| #export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:$BOOST_LIB | ||
|
|
||
| export PNGWRITER_ROOT=$HOME/lib/pngwriter |
There was a problem hiding this comment.
Make automatically consistent (only one path needs to be adjusted then):
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:$HOME/lib/pngwriter/lib/ export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:$HOME/lib/splash/lib/
to
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:$PNGwriter_DIR/lib/ export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:$Splash_DIR/lib/
|
|
||
| # Environment ################################################################# | ||
| # | ||
| #export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:$BOOST_LIB | ||
|
|
||
| export PNGWRITER_ROOT=$HOME/lib/pngwriter |
There was a problem hiding this comment.
In general: This proposed solution is messy, but it seems to work.
- The proposed module environment is unfortunately bundled (Boost,HDF5,gnuplot) with the
foss/2018amodule. This introduces a lot of unnecessary modules (51 modules total). The currentmodule list-command results in:
Derzeit geladene Module:
1) modenv/scs5 (S) 14) hwloc/1.11.8-GCCcore-6.4.0 27) gettext/0.19.8.1-GCCcore-6.4.0-libxml2-2.9.7 40) GMP/6.1.2-GCCcore-6.4.0
2) GCCcore/6.4.0 15) OpenMPI/2.1.2-GCC-6.4.0-2.28 28) PCRE/8.41-GCCcore-6.4.0 41) nettle/3.4-GCCcore-6.4.0
3) OpenBLAS/0.2.20-GCC-6.4.0-2.28 16) cURL/7.58.0-GCCcore-6.4.0 29) util-linux/2.31.1-GCCcore-6.4.0 42) libdrm/2.4.91-GCCcore-6.4.0
4) gompi/2018a 17) expat/2.2.5-GCCcore-6.4.0 30) GLib/2.54.3-GCCcore-6.4.0 43) LLVM/5.0.1-GCCcore-6.4.0
5) FFTW/3.3.7-gompi-2018a 18) Perl/5.26.1-GCCcore-6.4.0 31) fontconfig/2.12.6-GCCcore-6.4.0 44) Mesa/17.3.6-foss-2018a
6) ScaLAPACK/2.0.2-gompi-2018a-OpenBLAS-0.2.20 19) git/2.18.0-GCCcore-6.4.0 32) X11/20180131-GCCcore-6.4.0 45) libGLU/9.0.0-foss-2018a
7) foss/2018a 20) bzip2/1.0.6-GCCcore-6.4.0 33) cairo/1.14.12-GCCcore-6.4.0 46) Qt5/5.10.1-foss-2018a
8) binutils/2.28-GCCcore-6.4.0 21) libpng/1.6.34-GCCcore-6.4.0 34) NASM/2.13.03-GCCcore-6.4.0 47) gnuplot/5.2.4-foss-2018a
9) GCC/6.4.0-2.28 22) freetype/2.9-GCCcore-6.4.0 35) libjpeg-turbo/1.5.3-GCCcore-6.4.0 48) Boost/1.66.0-foss-2018a
10) ncurses/6.0-GCCcore-6.4.0 23) pixman/0.34.0-GCCcore-6.4.0 36) libgd/2.2.5-GCCcore-6.4.0 49) Szip/2.1.1-GCCcore-6.4.0
11) CMake/3.10.2-GCCcore-6.4.0 24) libffi/3.2.1-GCCcore-6.4.0 37) HarfBuzz/1.7.5-foss-2018a 50) HDF5/1.10.1-foss-2018a
12) CUDA/9.2.88 25) XZ/5.2.3-GCCcore-6.4.0 38) Pango/1.41.1-foss-2018a 51) zlib/1.2.11-GCCcore-6.4.0
13) numactl/2.0.11-GCCcore-6.4.0 26) libxml2/2.9.7-GCCcore-6.4.0 39) libcerf/1.5-GCCcore-6.4.0
Wo:
S: Das Modul ist angeheftet. Verwenden Sie "--force", um das Modul zu entladen.
To me, this appears to be excessive and will probably invite compatibility issues in the future. Maybe we should just build boost, hdf5 and gnuplot on our own.
- There are many Boost-related warnings during cmake configuration:
CMake Warning at /software/haswell/CMake/3.10.2-GCCcore-6.4.0/share/cmake-3.10/Modules/FindBoost.cmake:801 (message):
New Boost version may have incorrect or missing dependencies and imported
targets
Call Stack (most recent call first):
/software/haswell/CMake/3.10.2-GCCcore-6.4.0/share/cmake-3.10/Modules/FindBoost.cmake:907 (_Boost_COMPONENT_DEPENDENCIES)
/software/haswell/CMake/3.10.2-GCCcore-6.4.0/share/cmake-3.10/Modules/FindBoost.cmake:1542 (_Boost_MISSING_DEPENDENCIES)
/home/s0134766/src/picongpu/include/pmacc/PMaccConfig.cmake:296 (find_package)
CMakeLists.txt:83 (find_package)
...
many more
- There is a long stream of warnings during compiling:
/home/s0134766/src/picongpu/thirdParty/alpaka/include/alpaka/vec/Vec.hpp(925): warning: calling a __host__ function("std::__shared_count<( ::__gnu_cxx::_Lock_policy)2> ::__shared_count") from a __host__ __device__ function("alpaka::dev::DevCpu::DevCpu") is not allowed
/home/s0134766/src/picongpu/thirdParty/alpaka/include/alpaka/vec/Vec.hpp(925): warning: calling a __host__ function("std::__shared_count<( ::__gnu_cxx::_Lock_policy)2> ::__shared_count") from a __host__ __device__ function("alpaka::dev::DevCpu::DevCpu [subobject]") is not allowed
/sw/installed/Boost/1.66.0-foss-2018a/include/boost/mpl/map/aux_/item.hpp(54): warning: "boost::mpl::aux::type_wrapper<T> operator/(const boost::mpl::m_item<Key, T, Base> &, boost::mpl::aux::type_wrapper<Key> *)" declares a non-template function -- add <> to refer to a template instance
/sw/installed/Boost/1.66.0-foss-2018a/include/boost/mpl/map/aux_/item.hpp(55): warning: "boost::mpl::aux::type_wrapper<boost::mpl::pair<Key, T>> operator|(const boost::mpl::m_item<Key, T, Base> &, boost::mpl::next<Base::order>::type *)" declares a non-template function -- add <> to refer to a template instance
/sw/installed/Boost/1.66.0-foss-2018a/include/boost/mpl/map/aux_/item.hpp(56): warning: "char (&operator||(const boost::mpl::m_item<Key, T, Base> &, boost::mpl::aux::type_wrapper<Key> *))[boost::mpl::next<Base::order>::type::value]" declares a non-template function -- add <> to refer to a template instance
/sw/installed/Boost/1.66.0-foss-2018a/include/boost/mpl/map/aux_/item.hpp(71): warning: "boost::mpl::aux::type_wrapper<mpl_::void_> operator/(const boost::mpl::m_mask<Key, Base> &, boost::mpl::aux::type_wrapper<Key> *)" declares a non-template function -- add <> to refer to a template instance
/sw/installed/Boost/1.66.0-foss-2018a/include/boost/mpl/map/aux_/item.hpp(72): warning: "boost::mpl::aux::type_wrapper<mpl_::void_> operator|(const boost::mpl::m_mask<Key, Base> &, boost::mpl::x_order_impl<Base, Key>::type *)" declares a non-template function -- add <> to refer to a template instance
/home/s0134766/src/picongpu/include/pmacc/../pmacc/particles/memory/dataTypes/Particle.hpp(167): warning: __device__ annotation is ignored on a function("operator=") that is explicitly defaulted on its first declaration
/home/s0134766/src/picongpu/include/pmacc/../pmacc/particles/memory/dataTypes/Particle.hpp(167): warning: __host__ annotation is ignored on a function("operator=") that is explicitly defaulted on its first declaration
...
Many more
|
LWFA test case in
|
|
I will have a look into that issue today. |
|
Without When trying to compile # cmake output
-- The CXX compiler identification is GNU 6.4.0
-- Check for working CXX compiler: /sw/installed/GCCcore/6.4.0/bin/g++
-- Check for working CXX compiler: /sw/installed/GCCcore/6.4.0/bin/g++ -- works
-- Detecting CXX compiler ABI info
-- Detecting CXX compiler ABI info - done
-- Detecting CXX compile features
-- Detecting CXX compile features - done
-- Found ZLIB: /usr/lib64/libz.so (found version "1.2.7")
-- Found PNG: /usr/lib64/libpng.so (found suitable version "1.5.13", minimum required is "1.2.9")
-- Found Freetype: /usr/lib64/libfreetype.so (found version "2.4.11")
PNGwriter build configuration:
PNGwriter Version: 0.7.0
C++ Compiler : GNU 6.4.0
/sw/installed/GCCcore/6.4.0/bin/g++
Installation prefix: /scratch/p_electron/richard/libSCS5/lib/pngwriter
Build Type: Release
Build Options:
FREETYPE: ON
-- Configuring done
-- Generating done
-- Build files have been written to: /scratch/p_electron/richard/libSCS5/build
$ make install
[ 8%] Building CXX object CMakeFiles/PNGwriter.dir/src/pngwriter.cc.o
[ 16%] Linking CXX static library libPNGwriter.a
[ 16%] Built target PNGwriter
[ 25%] Building CXX object CMakeFiles/blackwhite.dir/tests/blackwhite.cc.o
[ 33%] Linking CXX executable blackwhite
libPNGwriter.a(pngwriter.cc.o):pngwriter.cc:function pngwriter::close(): error: undefined reference to 'png_convert_to_rfc1123_buffer'
collect2: Fehler: ld gab 1 als Ende-Status zurück
make[2]: *** [blackwhite] Fehler 1
make[1]: *** [CMakeFiles/blackwhite.dir/all] Fehler 2
make: *** [all] Fehler 2The module |
|
Compiling PIConGPU resulted in a number of boost warnings and some alpaka warnings: /.../picongpu2/thirdParty/alpaka/include/alpaka/vec/Vec.hpp(925): warning: calling a __host__ function("std::__shared_count<( ::__gnu_cxx::_Lock_policy)2> ::__shared_count") from a __host__ __device__ function("alpaka::dev::DevCpu::DevCpu") is not allowed
/.../picongpu2/thirdParty/alpaka/include/alpaka/vec/Vec.hpp(925): warning: calling a __host__ function("std::__shared_count<( ::__gnu_cxx::_Lock_policy)2> ::__shared_count") from a __host__ __device__ function("alpaka::dev::DevCpu::DevCpu [subobject]") is not allowedEdit: pic-build looks fine: final linking error: |
|
With regard to all these warnings - a newer boost, if needed even self compiled, might be a workaround. |
|
Self compiling boost 1.65.1 solved the issue. A run time test showed, that hdf5 data is produced. |
|
PIConGPU ran into a segmentation fault while writing hdf5 data. UPDATE: UPDATE 2: UPDATE 3: |
|
offline discussion with @psychocoderHPC: |
|
Trying to build hdf5 from source caused the following MPI related error during configuration: |
|
could you please test to set the mpi compiler instead compiling direct with the gnu compiler: |
|
with @psychocoderHPC trick, I could compile hdf5. Some test of hdf5 still failed with regard to MPI, bit I could build both splash and picongpu without compile errors. |
|
@ax3l Please rebase your branch against the current dev so that we can update the files easily. |
|
@PrometheusPi updates, feel free to add further commits! :) |
|
PNGwriter issue: you load |
|
@ax3l You were right, the I will open a ticket. Furthermore, when linking /sw/installed/libpng/1.6.34-GCCcore-6.4.0/lib/libpng.so: error: undefined reference to 'inflateValidate', version 'ZLIB_1.2.9'
$ldd /sw/installed/libpng/1.6.34-GCCcore-6.4.0/lib/libpng.so
linux-vdso.so.1 => (0x00007ffef53dc000)
libz.so.1 => /sw/installed/zlib/1.2.11-GCCcore-6.4.0/lib/libz.so.1 (0x00007efda7239000)
libm.so.6 => /usr/lib64/libm.so.6 (0x00007efda6f24000)
libpthread.so.0 => /usr/lib64/libpthread.so.0 (0x00007efda6d08000)
libc.so.6 => /usr/lib64/libc.so.6 (0x00007efda6945000)
/lib64/ld-linux-x86-64.so.2 (0x000055c81a442000)This looks like a mismatch - it links to the correctly loaded zlib module (1.2.11) but internally uses version 1.2.9. I will open a ticket. |
|
Yes, also make sure they expose the right zlib as a module again and that it itself sets |
|
Support ticket to taurus was submitted. ID: |
|
@ComputationalRadiationPhysics/picongpu-developers Do we want to keep the Intel compiler for knl or do we switch to gcc? |
|
In case ICC compiles for us, we should use it for KNL |
|
reply from taurus support (in German): Guten Tag @PrometheusPi, aus meiner Sicht ist da, was unsere Module angeht, nichts falsch gebaut. $ ml libpng/1.6.34-GCCcore-6.4.0 Es wird also ganz normal die 1.2.11er gelinkt, die auch als Abhängigkeit für das libpng Modul gesetzt ist und somit mitgeladen wird. Wenn Sie natürlich in Ihrer Anwendung nun aus irgendwelchen Gründen an anderer Stelle die libz.so aus /usr (1.2.9) gelinkt haben und zeitgleich auch libpng linken, kann so natürlich ein Konflikt entstehen, da ja nicht zwei Shared Libraries gleichen Namens im selben Prozess geladen werden können. D.h. da müssten Sie nochmal darauf schauen, was Ihr Buildsystem genau macht. Was das CMAKE_PREFIX_PATH angeht: ansich setzen die Module alle CPATH, LIBRARY_PATH und PKG_CONFIG_PATH, das sollte dem Buildsystem doch eigentlich ausreichen, um nicht bevorzugt die Bibliotheken aus /usr zu benutzen...? Bei anderer Software funktioniert das schließlich auch. Mit freundlichen Grüßen Therefore we need (for now) to set our |
Nope, we need a CMAKE_PREFIX_PATH, we are not in a pure Unix-systemtools nor autotools world. All modules should expand their CMAKE_PREFIX_PATH to their install prefixes... Please tell them, that all modern software projects based on CMake (C/C++/CUDA even Fortran) need this. Regarding easybuild: it's just a config that ZIH did forget to set. Of course Easybuild supports setting arbitrary vars in modules. Do we have to explain them their own tools? |
|
I wrote them an email explaining things. Nevertheless, did you find any |
|
As mentioned in the pr to your fork: the k20x and k80 run successful with the unchanged |
|
Cool, thanks! |
| @@ -64,6 +67,6 @@ export PYTHONPATH=$PICSRC/lib/python:$PYTHONPATH | |||
|
|
|||
| # "tbg" default options ####################################################### | |||
| # - SLURM (sbatch) | |||
| # - "gpu1" queue | |||
| # - "gpu2" queue | |||
There was a problem hiding this comment.
this comment should be gpu1
There was a problem hiding this comment.
and below it should read k20x.tpl :)
| @@ -13,21 +13,37 @@ export MY_NAME="$(whoami) <$MY_MAIL>" | |||
| # - examples: "nano", "vim", "emacs -nw", "vi" or without terminal: "gedit" | |||
| #export EDITOR="nano" | |||
|
|
|||
| # General modules ############################################################# | |||
| # Modules ############################################################# | |||
| # | |||
| module purge | |||
There was a problem hiding this comment.
purging here still a good idea? Works everywhere?
There was a problem hiding this comment.
It does not hurt and there are no modules loaded before - however it throws a warning that tells you that it cannot unload the following modules - followed by an empty list.
To avoid confusion I vote for removing the "old" module purge 👴
There was a problem hiding this comment.
alright, I will take it out
ax3l
force-pushed
the
topic-taurusSCS5
branch
3 times, most recently
from
f0a2a77 to
9ff9d3b
Compare
|
@PrometheusPi can you check if this PR still contains typos or is everything in good shape? :) |
| export LD_LIBRARY_PATH=$HDF5_ROOT/lib:$LD_LIBRARY_PATH | ||
| export CMAKE_PREFIX_PATH=$HDF5_ROOT:$CMAKE_PREFIX_PATH | ||
|
|
||
| # pngwriter needs to be build by the user: |
There was a problem hiding this comment.
be built, not be build; same on line 52 and in other profiles.
Also the whole construction feels weird to me, but I could not promptly come up with a clearly better alternative.
There was a problem hiding this comment.
| # pngwriter needs to be build by the user: | |
| # pngwriter needs to be built by the user: |
| export LD_LIBRARY_PATH=$HDF5_ROOT/lib:$LD_LIBRARY_PATH | ||
| export CMAKE_PREFIX_PATH=$HDF5_ROOT:$CMAKE_PREFIX_PATH | ||
|
|
||
| # pngwriter needs to be build by the user: |
There was a problem hiding this comment.
| # pngwriter needs to be build by the user: | |
| # pngwriter needs to be built by the user: |
| export CMAKE_PREFIX_PATH=$PNGwriter_DIR:$CMAKE_PREFIX_PATH | ||
| export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:$PNGwriter_DIR/lib/ | ||
|
|
||
| # splash needs to be build by the user: |
There was a problem hiding this comment.
| # splash needs to be build by the user: | |
| # splash needs to be built by the user: |
| export LD_LIBRARY_PATH=$HDF5_ROOT/lib:$LD_LIBRARY_PATH | ||
| export CMAKE_PREFIX_PATH=$HDF5_ROOT:$CMAKE_PREFIX_PATH | ||
|
|
||
| # pngwriter needs to be build by the user: |
There was a problem hiding this comment.
| # pngwriter needs to be build by the user: | |
| # pngwriter needs to be built by the user: |
| export CMAKE_PREFIX_PATH=$PNGwriter_DIR:$CMAKE_PREFIX_PATH | ||
| export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:$PNGwriter_DIR/lib/ | ||
|
|
||
| # splash needs to be build by the user: |
There was a problem hiding this comment.
| # splash needs to be build by the user: | |
| # splash needs to be built by the user: |
| module load HDF5/1.10.1-intel-2018a | ||
| module load libpng/1.6.34-GCCcore-7.3.0 | ||
|
|
||
| # module system does not expand cmake prefix path: |
There was a problem hiding this comment.
Imo this expand is not suitable here, as it does not mean expanding as for environment variable.
ax3l
force-pushed
the
topic-taurusSCS5
branch
from
9ff9d3b to
5d6ef3d
Compare
Co-authored-by: PrometheusPi <r.pausch@hzdr.de>
ax3l
force-pushed
the
topic-taurusSCS5
branch
from
5d6ef3d to
3c499ed
Compare
|
Regarding the knl partition: I |
Prepare an updated environment for the Taurus SCS5 software stack.
In order to use the new software stack, make sure to log into the login nodes
If you are an active Taurus user, please feel free to contribute further to this PR.
Refs