Differences between revisions 27 and 28
Revision 27 as of 2014-09-28 15:16:46
Size: 9818
Editor: MarcinDulak
Comment: gcc-4.8.3
Revision 28 as of 2018-04-06 14:16:40
Size: 84
Comment:
Deletions are marked like this. Additions are marked like this.
Line 5: Line 5:
.. Contents::

In order to compile your code for the desired compute node architecture
login to the corresponding login node
`login-to-niflheim <Getting_Started#login-to-niflheim>`_.

gnu
===

Gnu compilers are available by default and
the corresponding openmpi packages are available as modules,
see `Parallelization <Parallelization>`_ for details.

Newer gnu compiler versions are available through modules::

   module load libmpc/0.8-3.el6
   module load gcc/4.8.3-`echo $FYS_PLATFORM | cut -d "-" -f 1`-1

intel
=====

List available intel compilers with::

  module avail intel-compilers

and load the desired version, e.g. the recommended one with::

  module load intel-compilers

See `Installed_software <Installed_software>`_ for more about modules.

Each compiler set has the corresponding openmpi package installed, see `Parallelization <Parallelization>`_ for details.

Intel documentation
-------------------

Intel offers documentation
http://software.intel.com/en-us/articles/intel-software-technical-documentation/.
The compiler documentation is listed here:

* `C++ Compiler Professional Edition for Linux <http://software.intel.com/en-us/articles/intel-c-compiler-professional-edition-for-linux-documentation/>`_.

* `Fortran Compiler Professional Edition for Linux <http://software.intel.com/en-us/articles/intel-fortran-compiler-professional-edition-for-linux-documentation/>`_.

Intel Math Kernel Library (MKL)
-------------------------------

The `Intel Math Kernel Library <http://software.intel.com/en-us/intel-mkl/>`_ contains numerous mathematical operations.
Please see the `MKL documentation <http://software.intel.com/en-us/articles/intel-math-kernel-library-documentation/>`_.

Linking your code with MKL can be quite complex.
Therefore Intel offers a `Math Kernel Library Link Line Advisor <http://software.intel.com/en-us/articles/intel-mkl-link-line-advisor/>`_
to assist with the selection of proper linking commands.

nvcc
====

`NVIDIA CUDA Compiler Driver nvcc <http://docs.nvidia.com/cuda/cuda-compiler-driver-nvcc/>`_
is available on the Niflheim login nodes `Getting_Started <Getting_Started#login-to-niflheim>`_, after loading the modules::

  module load NVIDIA-Linux-x86_64
  module load cuda

An example of nvcc usage would be::

  nvcc -arch sm_35 -c example.cu -Xcompiler -fPIC

See `Installed_software <Installed_software>`_ for more about modules.


Installation and configuration of NVIDIA/cuda
---------------------------------------------

The information is this section is relevant only for system administration.

There are several documents describing installation and configuration of NVIDIA/cuda,
see for example https://access.redhat.com/site/solutions/64300
http://docs.nvidia.com/cuda/cuda-getting-started-guide-for-linux/index.html#runfile-installation
https://ask.fedoraproject.org/question/23982/how-to-disable-nouveau-in-fedora-18/

In order to install NVIDIA/cuda on an el6 x86_64 compute nodes the following steps were performed:

1. configure the node to boot in init 3 mode::

     sed -i 's/id:.:initdefault:/id:3:initdefault:/' /etc/inittab

2. install packages (libraries are required for some of the `NVIDIA_CUDA-5.5_Samples` to compile)::

     yum -y install kernel-devel wget make gcc-c++ freeglut-devel libXi-devel libXmu-devel mesa-libGLU-devel

3. rebuild `initramfs` in order to get rid of the `nouveau` driver::

     mv /boot/initramfs-$(uname -r).img /boot/initramfs-$(uname -r)-nouveau.img
     dracut --omit-drivers nouveau /boot/initramfs-$(uname -r).img $(uname -r)
     reboot

   Verify that nouveau is not present::

     lsmod | grep nouveau

   Note that after removal of `nouveau` from `initramfs` no blacklisting of `nouveau`
   (`echo "blacklist nouveau" > /etc/modprobe.d/blacklist-nouveau.conf`) is necessary.
   Note however that blacklisting of `nouveau` alone is not sufficient - there are
   other modules that load `nouveau` trigerred by /boot/initramfs.
   See https://bbs.archlinux.org/viewtopic.php?id=123421

4. install NVIDIA drivers::

     wget http://us.download.nvidia.com/XFree86/Linux-x86_64/331.20/NVIDIA-Linux-x86_64-331.20.run
     sh NVIDIA-Linux-x86_64-331.20.run --silent

   Note that in order to make NVIDIA libraries packages in NVIDIA-Linux-x86_64-331.20.run (e.g. `libcuda.so`)
   available on the login nodes, which itself do not contain GPU devices, one can unpack the installer::

     sh NVIDIA-Linux-x86_64-331.20.run --extract-only

   as non-*root*, copy the shared objects and includes into a desired *location* and
   create links to shared objects manually::

     for f in /home/opt/common/NVIDIA-Linux-x86_64-331.20/lib/*.so.*;
     do
        bn=`echo $f | sed 's/.so/:/' | rev | cut -d":" -f 2 | rev`.so # *.so
        ln -sv `basename $f` $prefix/lib/`basename $bn`.1 # *.so.* -> *.so.1
        ln -sv `basename $bn`.1 $prefix/lib/`basename $bn` # *.so.1 -> *.so
     done

   For the applications which use the NVIDIA libraries set::

     export LD_LIBRARY_PATH=/home/opt/common/NVIDIA-Linux-x86_64-331.20/lib:${LD_LIBRARY_PATH}

5. due to the fact that X is not running (the node is init 3 mode) `/dev/nvidia*` devices
   need to be created at every boot. This is achieved by `/etc/init.d/nvidia` script
   available for example at https://access.redhat.com/site/solutions/64300::

     chmod a+x /etc/init.d/nvidia
     chkconfig nvidia on
     service nvidia start

   Verify that `nvidia` module is loaded::

     lsmod | grep nvidia

   you will see something like::

     nvidia 10613805 0
     i2c_core 31276 1 nvidia

   Note that step 3. may be unnecessary. Without the removal of `nouveau`
   from `initramfs`, and after installation of NVIDIA drivers (step 4.) you will see::

     i2c_core 31276 4 nvidia,nouveau,drm_kms_helper,drm

   however this does not seem to influence of the GPU devices.

6. install cuda (again this can be performed as non-*root*)::

     wget http://developer.download.nvidia.com/compute/cuda/5_5/rel/installers/cuda_5.5.22_linux_64.run
     prefix=/home/opt/el6/common/cuda-5.5.22
     sh cuda_5.5.22_linux_64.run -silent -toolkit -toolkitpath=$prefix -samples -samplespath=$prefix

   Set the *PATH*, *LD_LIBRARY_PATH* and *INCLUDE* variables accordingly.

7. test the installation on a system containing a GPU device.
   Note that the `make` step requires the NVIDIA libraries to be installed under `/usr/lib*`.
   As recommended above one can install the NVIDIA libraries in the default location on the compute nodes,
   and in a custom location on the login nodes. Test the installation on the compute node with::

     cd $prefix/NVIDIA_CUDA-5.5_Samples
     cp -rp ../NVIDIA_CUDA-5.5_Samples ../NVIDIA_CUDA-5.5_Samples.orig
     make
     ./bin/x86_64/linux/release/deviceQuery

open64
======

List available open64 compilers with::

  module avail open64

and load the desired version, e.g. the recommended one with::

  module load open64

Each compiler set has the corresponding openmpi package installed, see `Parallelization <Parallelization>`_ for details.

Additional information about Pathscale compilers
------------------------------------------------

After the bankruptcy of `Pathscale <http://www.pathscale.com>`_, `Aertia <http://www.aertia.com/en/productos.asp?pid=24>`_
you are encouraged to use `open64 <http://www.open64.net/>`_ compiler instead.

If you prefer to use pathscale still, the login node *fjorm.fysik.dtu.dk* has a single-user license for the
`Pathscale EKOPATH compiler for AMD64 and EM64T <http://www.pathscale.com/>`_.
This was probably the best compiler on the market for AMD Opteron CPUs.

The Pathscale Fortran-90 compiler command is ``pathf90``, and the other
languages are ``pathf95`` (Fortran 95), ``pathcc`` (C), ``pathCC`` (C++).

Pathscale license manager
-------------------------

Since we have only a single license the Pathscale compiler, the license manager daemon
will allocate the use of the compiler to only **one user at a time**,
with a time-out period of 15 minutes of non-usage.

If you are having trouble with PathScale subscription management while using
the compilers, use the ``-subverbose`` option as you compile. The information from
this option is useful for diagnosing problems. You will need to do an actual
compile with the ``-subverbose`` option to see the output.
For example, you might type::

  pathcc -subverbose hello.c

ACML libraries
--------------

We have also installed the `AMD ACML Math library <http://developer.amd.com/acml.aspx>`_.
Please see the AMD `ACML User Guide <http://developer.amd.com/wordpress/media/2012/10/acml_userguide.pdf.pdf>`_ (pdf).
You have to link to different ACML libraries according to which compiler your use, for example:

 * **GNU gcc EL5**::

      gcc -m64 -I/opt/acml/4.0.1/gfortran64/include/ *.o -L/opt/acml/4.0.1/gfortran64/lib/ -lacml -lgfortran

 * **GNU gcc 4.3 EL5**::

      gcc43 -m64 -I/opt/acml/4.3.0/gfortran4364/include/ *.o -L/opt/acml/4.3.0/gfortran4364/lib/ -lacml -lgfortran

 * **GNU gcc 4.4 EL6**::

      gcc -m64 -I/home/opt/common/acml-gfortran-64bit-4.4.0/include *.o -L/home/opt/common/acml-gfortran-64bit-4.4.0/lib/ -lacml -lgfortran

 * **open64 EL6**::

      opencc -I/home/opt/common/acml-open64-64bit-4.4.0/include *.o -L/home/opt/common/acml-open64-64bit-4.4.0/ -lacml

**Note**: problems with `dgemm <http://developer.amd.com/cpu/Libraries/acml/downloads/assets/ReleaseNotes.txt>`_ acml 4.1.0 and 4.2.0 have been reported, use these versions at your own risk!
Please see the Niflheim7_Getting_started_ page.

Compilers

Please see the Niflheim7_Getting_started page.

Niflheim: Compilers (last edited 2018-04-06 14:16:40 by OleHolmNielsen)