Difference between revisions of "Welcome to the HPC User Wiki of the University of Oldenburg"

From HPC users
Jump to navigationJump to search
 
(169 intermediate revisions by 8 users not shown)
Line 1: Line 1:
<div style="text-align:justify">
__NOTOC__
__NOEDITSECTION__
<div style="text-align:justify;">
<center>
{| style="text-align:justify;font-size:1.2em;line-height:1.2em;background-color:#eeeeff;" border="1" cellspacing="0"
|-
| [[Image:picture_of_nodes.jpg|155px]]
| [[Image:picture_of_cluster_closed.jpg|70px]]
| ''This is the HPC-Wiki of the University of Oldenburg''<br>
| [[Image:picture_of_gpfs.jpg|82px]]
| [[Image:picture_of_infinyband.jpg|155px]]
|}
</center>


'''Note''': This is a first, '''preliminary''' version (v0.01) of the HPC User Wiki. Its primary purpose is to get you started with our new clusters (FLOW and HERO), enabling you to familiarize with these systems and gather some experience. More elaborate, improved versions will follow, so you may want to check these pages regularly.
= Basic Information =
<center>
{| style="background-color:#eeeeff;" cellpadding="10" border="1" cellspacing="0"
|- style="background-color:#ddddff;"
! HPC Facilities
! Login
! User environment
! Compiling and linking
! Job Management (Queueing) System
! Altix UV 100 system
! Examples
|- valign="top"
|
* [[HPC Facilities of the University of Oldenburg| Overview]]
* [[HPC Facilities of the University of Oldenburg#FLOW| FLOW]]
* [[HPC Facilities of the University of Oldenburg#HERO| HERO]]
* [[HPC Policies| HPC Policies]]
* [[Unix groups| Groups ]]
* [[Acknowledging_the_HPC_facilities| Acknowledging FLOW/HERO]]
* [[User Meetings]]
|
* [[Logging in to the system#From within the University (intranet) | From University]]
* [[Logging in to the system#From outside the University (internet) | From Home]]
|
* [[User environment - The usage of module| Usage of module]]
* [[File system| File System / Quotas]]
* [[Mounting Directories of FLOW and HERO#Windows | Shares under Windows]]
* [[Mounting Directories of FLOW and HERO#Linux | Shares under Linux]]
* [[License servers]]
|
* [[Compiling and linking|Basics]]
* [[GNU Compiler]]
* [[Intel Compiler]]
* [[PGI Compiler]]
* [[Open64 Compiler]]
* [[Using the Altix UV 100 system#Compiling and linking applications| Altix UV 100]]


= Introduction =
|
* [[SGE Job Management (Queueing) System| Overview]]
* [[SGE Job Management (Queueing) System#Submitting jobs| Submitting ]]
* [[SGE Job Management (Queueing) System#Specifying job requirements| Job requirements ]]
* [[SGE Job Management (Queueing) System#Parallel environments (PEs) | Parallel jobs ]]
* [[SGE Job Management (Queueing) System#Interactive jobs | Interactive jobs ]]
* [[SGE Job Management (Queueing) System#Monitoring and managing your jobs | Commands ]]
* [[SGE Job Management (Queueing) System#Array jobs| Job arrays  ]]
* [[SGE Job Management (Queueing) System#Environment variables | Environment variables]]
* [[Brief_Introduction_to_HPC_Computing#Checking_the_status_of_the_job | Checking the job status]] [[Brief_Introduction_to_HPC_Computing#Checking_the_status_of_the_job_2| (par. jobs)]]
* [[Brief_Introduction_to_HPC_Computing#Details_for_finished_jobs| Obtaining details for finished jobs]]
* [[SGE Job Management (Queueing) System#Documentation | Documentation]]
* [[Queues_and_resource_allocation| On Queues and resource allocation]]
|
* [[Using the Altix UV 100 system#Compiling and linking applications| Compiling]]
* [[Using the Altix UV 100 system#Submitting SGE jobs| Submitting]]
* [[Using the Altix UV 100 system#Documentation| Documentation]]
|
* [[Brief Introduction to HPC Computing| Brief Introduction to HPC Computing]]
* [[Matlab Examples using MDCS| Matlab examples using MDCS]]
* [[MDCS Basic Example]] (for R2014b and later)
* [[HPC Tutorial No1| HPC Tutorial 2013]]
* [[HPC Introduction October 6-8, 2014| HPC Tutorial 2014]]
* [[HPC Introduction October 7-9, 2015| HPC Tutorial 2015]]
|-


Presently, the central HPC facilities of the University of Oldenburg comprise three systems:


*FLOW ('''F'''acility for '''L'''arge-Scale C'''O'''mputations in '''W'''ind Energy Research)<br> IBM iDataPlex cluster solution, 2232 CPU cores, 6 TB of (distributed) main memory, QDR InfiniBand interconnect.<br>Theoretical peak performance: '''24 TFlop/s'''.
|}
</center>


*HERO ('''H'''igh-'''E'''nd Computing '''R'''esource '''O'''ldenburg)<br>Hybrid system composed of two components:
= Application Software and Libraries =
**IBM iDataPlex cluster solution, 1800 CPU cores, 4 TB of (distributed) main memory, Gigabit Ethernet interconnect.<br>Theoretical peak performance: '''19.2 TFlop/s'''.
**SGI Altix UltraViolet shared-memory system ("SMP component"), 120 CPU cores, 640 GB of globally addressable memory, NumaLink5 interconnect<br>Theoretical peak performance: '''1.3 TFlop/s'''.


*[http://www.csc.uni-oldenburg.de GOLEM]: older, AMD Opteron-based cluster with 390 cores and 800 GB of (distributed) main memory.<br>Theoretical peak performance: 1.6 TFlop/s.
<center>
{| style="background-color:#eeeeff;" cellpadding="10" border="1" cellspacing="0"
|- style="background-color:#ddddff;"
!Compiler and Development Tools
!Quantum Chemistry
!Computational Fluid Dynamics
!Mathematics/Scripting
!Visualisation
!Libraries
|- valign="top"
|
* [[debugging]]
* [[git]]
* [[GNU Compiler]]
* [[Intel Compiler]]
* [[Open64 Compiler]]
* [[PGI Compiler]]
* [[Profiling_using_gprof| profiling]]
* [[scalasca]]
* [[subversion (svn)]]
* [[valgrind]]


FLOW and HERO use a common, shared storage system (high-performance NAS Cluster) with a net capacity of 130 TB.
|
* [[Gaussian 09]]
* [[MOLCAS]]
* [[MOLPRO]]
* [[NBO]]
* [[ORCA]]
|
* [[Ansys]]
* [[FOAMpro]]
* [[Nektar++]]
* [[Nek 5000]]
* [[OpenFOAM]]
* [[PALM]]
* [[STAR-CCM++]]
* [[THETA]]
* [[WRF/WPS]]


FLOW is used for computationally demanding CFD calculations in wind energy research, conducted by the Research Group [http://twist.physik.uni-oldenburg.de/en/index.html TWiST] (Turbulence, Wind Energy, and Stochastis) and the [http://www.forwind.de/forwind/index.php?article_id=1&clang=1 ForWind] Center for Wind Energy Research. It is, to the best of our knowledge, the largest system in Europe dedicated solely to that purpose.
|
* [[Configuration MDCS]] (2014b and later)  
* [[MATLAB Distributing Computing Server]]
* [[Python]]
* [[R]]
* [[STATA| STATA]]
|
* [[iso99]]
* [[NCL]]
* [[ncview]]
* [[paraview]]
|
* [[BLAS and LAPACK]]
* [[EGSnrc]]
* [[FLUKA]]
* [[GEANT4]]
* [[Gurobi]]
* [[HDF5]]
* [[Intel MPI]]
* [[LEDA]]
* [[NetCDF]]
* [[OpenMPI]]


The main application areas of the HERO cluster are Quantum Chemistry, Theoretical Physics, the Neurosciences, and Audiology. Besides that, the system is used by many other research groups of the [http://www.fk5.uni-oldenburg.de Faculty of Mathematics and Science] and the [http://www.informatik.uni-oldenburg.de Department of Informatics] of the School of Computing Science, Business Administration, Economics, and Law.
|-


= Hardware Overview  =
|}
</center>


== FLOW  ==
= Courses and Tutorials =


*122 "low-memory" compute nodes: IBM dx360 M3, dual socket (Westmere-EP, 6C, 2.66 GHz), 12 cores per server, 24 GB DDR3 RAM, diskless (host names <tt>cfdl001..cfdl122</tt>).
<center>
{| style="background-color:#eeeeff;" cellpadding="10" border="1" cellspacing="0"  
|- style="background-color:#ddddff;"
!Introduction to HPC Courses
!Matlab Tutorials
!New OS
|- valign="top"
|
* [[HPC Introduction October 6-8, 2014]]
* [[HPC Introduction October 7-9, 2015]]
|
* [[Audio Data Processing]]
* [[Using the MEX Compiler]]
|
* [[media:New_OS_On_FLOW.pdf | New OS on FLOW ]]
|-


*64 "high-memory" compute nodes: IBM dx360 M3, dual socket (Westmere-EP, 6C, 2.66 GHz), 12 cores per server, 48 GB DDR3 RAM, diskless (host names <tt>cfdh001..cfdh064</tt>).
|}
</center>


*QDR InfiniBand interconnect (fully non-blocking), 198-port Mellanox IS5200 IB switch (can be extended up to 216 ports).


*Gigabit Ethernet for File-I/O etc.
= Contact =


*10/100 Mb/s Ethernet for management and administrative tasks (IPMI).
<center>
{| style="background-color:#eeeeff;" cellpadding="10" border="1" cellspacing="0"
|- style="background-color:#ddddff;"
!HPC Resource
!EMail
|- valign="top"
|
FLOW and HERO<br>
Both (in case of vacation)<br>
|
Stefan.Harfst@uni-oldenburg.de<br>
hpcuniol@uni-oldenburg.de<br>
|-
|}
</center>


== HERO  ==


*130 "standard" compute nodes: IBM dx360 M3, dual socket (Westmere-EP, 6C, 2.66 GHz), 12 cores per server, 24 GB DDR3 RAM, 1 TB SATAII disk (host names <tt>mpcs001..mpcs130</tt>).
'''''Note:''' This Wiki is under construction and a preliminary version! Contributions are welcome. Please ask Stefan Harfst (Stefan.Harfst(at)uni-oldenburg.de) for further informations.''


*20 "big" compute nodes: IBM dx360 M3, dual socket (Westmere-EP, 6C, 2.66 GHz), 12 cores per server, 48 GB DDR3 RAM, RAID 8 x 300 GB 15k SAS (host names <tt>mpcb001..mpcb020</tt>)
<center>
 
''Only for editors: [[Formatting rules for this Wiki]]''
*Gigabit Ethernet II for communication of parallel jobs (MPI, LINDA, ...).
</center>
 
*Second, independent Gigabit Ethernet for File-I/O etc.
 
*10/100 Mb/s Ethernet for management and administrative tasks (IPMI).
 
*SGI Altix UV 100 shared-memory system, 10 CPUs (Nehalem-EX, "Beckton", 6C, 2.66 GHz), 120 cores in total, 640 GB DDR3 RAM, NumaLink5 interconnect, RAID 20 x 600 GB SAS 15k rpm (host <tt>uv100</tt>).
 
The 1 Gb/s leaf switches have uplinks to a 10 Gb/s backbone (two switches, redundant). The central management interface of both clusters runs on two master nodes (IBM x3550 M3) in an HA setup. Each cluster has two login nodes (IBM x3550 M3).
 
Operating system: '''Scientific Linux 5.5'''
 
Cluster management software: '''Bright Cluster Manager 5.1''' by [http://www.clustervision.com ClusterVision B.V.]
 
= Basic Usage  =
 
== Logging in to the system  ==
 
=== From within the University (intranet)  ===
 
Within the internal net of the University, access to the systems is granted via ssh. Use your favorite ssh client like OpenSSH, PuTTY, etc. For example, on a UNIX/Linux system, users of FLOW may type on the command line (replace "abcd1234" by your own account):
 
ssh abcd1234@flow.hpc.uni-oldenburg.de
 
Similarly, users of HERO login by typing:
 
ssh abcd1234@hero.hpc.uni-oldenburg.de
 
Use "<tt>ssh -X</tt>" for X11 forwarding (i.e., if you need to export the graphical display to your local system).
 
For security reasons, access to the HPC systems is denied from certain subnets. In particular, you cannot login from the WLAN of the University (uniolwlan) or from "public" PCs (located, e.g., in Libraries, PC rooms, or at other places).
 
=== From outside the University (internet)  ===
 
First, you have to establish a VPN tunnel to the University intranet. After that, you can login to HERO or FLOW via ssh as described above. The data of the tunnel are:
 
Gateway      &nbsp;: vpn2.uni-oldenburg.de
Group name  &nbsp;: hpc-vpn
Group password: hqc-vqn
 
Cf. the [http://www.itdienste.uni-oldenburg.de/21240.html instructions] of the IT Services on how to configure the Cisco VPN client. For the HPC systems, a separate VPN tunnel has been installed, which is only accessible for users of FLOW and HERO. Therefore, you have to configure a new VPN connection and enter the data provided above. For security reasons, you cannot login to FLOW or HERO if you are connected to the intranet via the "generic" VPN tunnel of the University.
 
== User Environment  ==
 
We use the module environment, which has a lot of advantages, is very flexible (and user-friendly), and even allows one to use different versions of the same software concurrently on the same system. You can see a list of all available modules by typing
module avail
 
To load a given module:
module load <name of the module>
 
The modules system uses a hierarchical file structure, i.e., sometimes (whenever there are ambiguities) you may have to specify a path, as in:
module load fftw2/gcc/64/double
 
To revert all changes made by a given module (environment variables, paths, etc.):
module unload <name of the module>
 
 
== Compiling and linking  ==
 
This section will be elaborated later and then provide much more detailed information. For the time being, we only give a '''very''' brief overview.
 
The following compilers and MPI libraries are currently available:
 
* GCC, the GNU Compiler Collection: <tt>gcc</tt> Version 4.3.4<pre>module load gcc</pre>This module is loaded per default if you log in to the system.Supported MPI libraries: OpenMPI, MPICH, MPICH2, MVAPICH and MVAPICH2
 
* Intel Cluster Studio 2011, formerly known as Intel Cluster Toolkit Compiler Edition (contains the ''Math Kernel Library'' and other performance libraries, analyzer, and HPC tools):<pre>module load intel/ics</pre>The environment for the Intel MPI library must be loaded separately:<pre>module load intel/impi</pre>The Fortran compuler is invoked by <tt>ifort</tt>, and the C/C++ compiler by <tt>icc</tt>. However, if one wants to build MPI applications, one should generally use the wrapper scripts <tt>mpif77</tt>, <tt>mpif90</tt>, <tt>mpicc</tt>, ...
 
* PGI Cluster Development Kit, Version 11.3: contains a suite of Fortran and C/C++ compiler as well as various other tools (MPI debugger etc.):<pre>module load pgi</pre>. Invoked by <tt>pgf77</tt>, <tt>pgf95</tt>, ... and </tt><tt>pgcc</tt>, <tt>pgcpp</tt>, ... for FORTRAN and C/C++, respectively. Again, wrapper scripts exist for building MPI applications.<br>Supported MPI libraries: MPICH, MPICH2, and MVAPICH.
(At the moment, MPICH and MPICH2 have problems running under the queueing system and thus their use is not recommended, but that problem will be fixed soon.)
 
Is planned to extend the MPI support for various compilers. In particular, OpenMPI will soon be available for the Intel compiler, too.
 
==== Documentation  ====
 
*[http://software.intel.com/sites/products/documentation/hpc/composerxe/en-us/fortran/lin/index.htm Intel Fortran compiler User and Reference Guides]
 
*[http://software.intel.com/sites/products/documentation/hpc/composerxe/en-us/cpp/lin/index.htm Intel C/C++ Compiler]
 
*[http://software.intel.com/sites/products/documentation/hpc/composerxe/en-us/start/lin/cpp/index.htm Intel Getting started tutorial]
 
*[http://software.intel.com/sites/products/documentation/hpc/mkl/userguides/mkl_userguide_lnx/index.htm Intel Math Kernel Library User's Guide]
 
[http://www.pgroup.com/doc/pgiug.pdf PGI User's Guide (PDF)]
 
== Job Management (Queueing) System  ==
 
The queueing system employed to manage user jobs for FLOW and HERO is [http://wikis.sun.com/display/GridEngine/Home Sun Grid Engine] (SGE). For first-time users (especially those acquainted with PBS-based systems), some features of SGE may seem a little unusual and certainly need some getting-accustomed-to. In order to efficiently use the available hardware resources (so that all users may benefit the most from the system), a basic understanding of how SGE works is indispensable. Some of the points to keep in mind are the following:
 
*Unlike other (e.g., PBS-based) queueing systems, SGE does not "know" the concept of "nodes" with a fixed number of CPUs (cores) and users specifying the number of nodes they need, along with the number of CPUs per node, in their job requirements. Instead, SGE logically divides the cluster into '''slots''', where each "slot" may be thought of as a single CPU core. The scheduler assigns free slots to pending jobs. Since in the multi-core area each host offers many slots, this will, in general, lead to jobs of different users running concurrently on the same host (provided that there are sufficient resources like memory, disk space etc. to meet all requirements of all jobs, as specified by the users who submitted them) and usually guarantees efficient resource utilization.
 
*While the scheduling behavior described above may be very efficient in optimally using the available hardware resources, it will have undesirable effects on parallel (MPI, LINDA, ...) jobs. E.g., an MPI job requesting 24 slots could end up running 3 tasks on one host, 12 tasks on another host (fully occupying this host, if it is a server with 2 six-core CPUs, as happens with our clusters), and 9 tasks on a third host. Clearly, such an unbalanced configuration may lead to problems. For certain jobs (multithreaded applications) it is even mandatory that all slots reside on one host (typical examples: OpenMP programs, Gaussian single-node jobs).<br> To deal with the specific demands of parallel jobs, SGE offers so-called '''parallel environments (PEs)''' which are largely configurable. Even if your job does not need several hosts, but runs on only one host using several or all cores of that host, you '''must''' specify a parallel environment. '''It is of crucial importance to choose the "correct" parallel environment''' (meeting the requirements of your application/program) when submitting a parallel job.
 
*Another "peculiarity" of SGE (as compared to its cousins) are the concepts of '''cluster queues''' and '''queue instances'''. Cluster queues are composed of several (typically, many) queue instances, with each instance associated with one particular host. A cluster queue may have a name like, e.g., ''standardqueue.q'', where the .q suffix is a commonly followed convention. Then the queue instances of this queue has names like, e.g. ''standardqueue.q@host001'', ''standardqueue.q@host002'', ... (note the "@" which acts as a delimiter between the queue name and the queue instance).<br> In general, each host will hold several queue instances belonging to different cluster queues. E.g. there may be a special queue for long-running jobs and a queue for shorter jobs, both of which share the same "physical" machines but have different policies. To avoid oversubscription, resource limits can be configure for individual hosts. Since resource limits and other, more complex attributes can also be associated with cluster queues and even queue instances, the system is highly flexible and can be customized for specified needs. On the other hand, the configuration quickly tends to get rather complex, leading to unexpected side effects. E.g., PEs grab slots from all queue instances of all cluster queues they are associated with. Thus, a parallel job may occupy slots on one particular host belonging to different queue instances on that host. While this is usually no problem for the parallel job itself, it blocks resources in both cluster queues which may be unintended. For that reason, it is common practice to associate each PE with one and only one cluster queue and define several (possibly identically configured) PEs in order to avoid that a single PE spans several cluster queues.
 
==== Submitting jobs  ====
 
Sample job submission scripts for both serial and parallel jobs are provided in the subdirectory <tt>Examples</tt> of your homedirectory. You may have to adapt these scripts as needed. Note that a job submission script consists of two principal parts:
 
*SGE directives (lines starting with the "magic" characters <tt>#$</tt>), which fall into three categories:
**general options (which shell to use, name of the job, name of output and error files if differing from default, etc.). The directives are passed to the <tt>qsub</tt> command when the job is submitted.
**Resource requirements (introduced by the <tt>-l</tt> flag), like memory, disk space, runtime (wallclock) limit, etc.
**Options for parallel jobs (parallel environment, number of job slots, etc.)
 
*Commands to be executed by the job (your program, script, etc.), including the necessary set-up of the environment for the application/program to run correctly (loading of modules so that your programs find the required runtime libraries, etc.).
 
The job is submitted by the <tt>qsub</tt> command, e.g. (assuming your submission script is named"<tt>myprog.sge</tt>):
 
qsub myprog.sge
 
==== Specifying job requirements  ====
 
The general philosophy behind SGE is that you should not submit your job to a specific queue or queue instance (although this is possible in principle), but rather define your requirements, and then let SGE decide which queue matches them best (taking into account the current load of the system and other factors). For this "automatic" queue selection to work efficiently and in order to avoid wasting of valuable resources (e.g., requesting much more memory than your job needs, which may prevent the scheduling of jobs of other users) it is important that you give a complete and precise specification of your job requirements in your submission script. The following points are relevant to both serial and parallel jobs.
 
===== Runtime =====
 
Maximum (wallclock) runtime is specified by <tt>-l h_rt=&lt;hh:mm:ss&gt;</tt>. E.g., a maximum runtime of three days is requested by:
<pre>
#$ -l h_rt=72:0:0
</pre>
 
All cluster queues except the "long" queues have a maximum allowed runtime of '''8 days'''. It is highly recommendable that you specify the runtime of your job as accurately as possible (leaving, of course, a margin of error). If the scheduler knows that, e.g., a pending job is a "fast run" which needs only a few hours of walltime, it is likely that it gets executed much earlier than other jobs with higher walltime requirements (so-called '''backfilling''').
 
If your job needs more than 8 days of walltime, your submission script must contain the following line:
<pre>
#$ -l longrun=true
</pre>
It is then automatically directed to one of the "long" queues, which have no runtime limit. The number of long-running jobs per user is limited.
 
===== Memory =====
Maximum memory usage of a job is defined by the <tt>h_vmem</tt> attribute, as in
<pre>
#$ -l h_vmem=4G
</pre>
for a job requesting 4 GB of memory. If your job exceeds the speficied memory limit, it gets killed automatically.
 
'''Note''': The <tt>h_vmem</tt> attribute refers to the memory '''per job slot''' (CPU core), i.e. it gets multiplied by the number of slots for a parallel job.
 
Total memory available on each compute node:
<br>
<ul>
<li>standard compute nodes of HERO (<tt>mpcs001..mpcs130</tt>): 23 GB
<li>big nodes of HERO (<tt>mpcb001..mpcb020</tt>): 46 GB
<li>low-memory nodes of FLOW (<tt>cfdl001..cfdl122</tt>): 22 GB
<li>high-memory nodes of FLOW (<tt>cfdh001..cfdh064</tt>): 46 GB
</ul>
If your job needs one (or several) of the "big nodes" of HERO (<tt>mpcb001..mpcb020</tt>), you must specify your memory requirement '''and''' set the Boolean attribute <tt>bignode</tt> to <tt>true</tt>. Example: A job in the parallel environment "smp" (see below) requests 12 slots and 3 GB per slot (i.e., <tt>h_vmem=3G</tt>). This jobs needs 36 GB of memory on a single node in total, and thus can only run on one of the big nodes. The corresponding section of your submission scrip will then read:
<pre>
#$ -l h_vmem=3G
#$ -l bignode=true
</pre>
 
Similarly, to request one of the high-memory nodes of FLOW (<tt>cfdh001..cfdh064</tt>), you need to set the attribute <tt>highmen</tt> to <tt>true</tt>. Example: for an MPI job with 12 tasks per node and a memory requirement of 3 GB for each task, you would specify:
<pre>
#$ -l h_vmem=3G
#$ -l highmem=true
</pre>
 
===== Local disk space (HERO only) =====
Local scratch space is only available for the HERO, since the compute nodes of FLOW are diskless.
For requesting, e.g., 200 GB of scratch space, the SGE directive reads:
<pre>
#$ -l h_fsize=200G
</pre>
The path to the local scratch directory can be accessed in your job script (or other scripts/programs invoked by your job) via the <tt>$TMPDIR</tt> environment variable. After termination of your job (or if you kill your job manually by <tt>qdel</tt>), the scratch directory is automatically purged.
 
Total amount of scratch space available on each compute node:
<ul>
<li> standard nodes (<tt>mps001..mpcs130</tt>): 800 GB
<li> big nodes (<tt>mpcb001..mpcb020</tt>): 2100 GB
</ul>
 
If your job needs more than 800 GB of scratch space, you must request one of the big nodes. Example:
<pre>
#$ -l h_fsize=1400G
#$ -l bignode=true
</pre>
 
==== Parallel environments (PEs)  ====
 
'''Example''': If you have an MPI program compiled and linked with the Intel Compiler and MPI library,
your job submission script might
#$ -pe intelmpi 96 
#$ -R y
#
load module intel/impi
mpiexec -machinefile $TMPDIR/machines -n $NSLOTS -env I_MPI_FABRICS shm:ofa ./myprog_intelmpi
 
In that case, the MPI job uses the InfiniBand fabric for communication (the I_MPI_FABRICS variable).
Turning on resource reservation (<tt>-R y</tt>) is highly recommended in order to avoid starving of parallel jobs by serial jobs which "block" required slots on specific hosts.The job requests 96 cores. The allocation rule of this PE is "fill-up", i.e. SGE tries to place the MPI tasks on as few hosts as possible (in the "ideal" case, the program would run on exactly 8 hosts (with cores or slots on each host, but there is no guerantee that this is going to happen).
Please have a look at the directory named <tt>Examples</tt> in your homedirectory, which contains other examples how to submit parallel (MPI) jobs.
 
List of all currently available PEs:
*<tt>intelmpi</tt> for using the Intel MPI Library, see above.
 
*<tt>openmpi</tt> for using the OpenMPI Library (so far, only supported with the <tt>gcc</tt> compiler
 
*<tt>mvapich</tt> for MVAPICH library (i.e., InfiniBand interconnects)
 
*<tt>smp</tt>: this PE demands that '''all''' requested slots be on the same host (needed for multithreaded applications, like Gaussian single-node jobs, OpenMP, etc.)
 
*<tt>linda</tt>: special PE for Gaussian Linda jobs, see below.
 
If your job is to run in one of the "long" queues (i.e., requesting more than 8 days of walltime), you must use the corresponding "long" version of the PE: intelmpi_long, openmpi_long, etc.
 
Note that the above list will grow over time. E.g., it is planned to support OpenMPI with the Intel Compiler (not only the <tt>gcc</tt> compiler, as is now the case).
 
... tbc ...
 
<br>
 
==== Interactive jobs  ====
 
Interactive jobs are only allowed for members of certain groups from the Institue of Psychology who have special data pre-processing needs which require manual intervention and cannot be automatized (the prerequesit for writing a batch job script).
 
Users who are entitled to interactive jobs type
qlogin
on the command line (terminal). After that, a graphical Matlab session can be started by issuing the
following two commands:
module load matlab
matlab &
(Sending the Matlab process to the background gives you control over the shell, which may be useful.
 
 
==== Monitoring and managing your jobs  ====
 
A selection of the most frequently used commands for job monitoring and management:
 
*<tt>qstat</tt>: display all (pending, running, ...) jobs of the user (output is empty if user has no jobs in the system).
 
*<tt>qstat -j <jobid></tt>: get a more verbose output, which is particularly useful when analyzing why your job won't run.
 
*<tt>qdel <jobid></tt>: kill job with specified ID (users can, of course, only kill their own jobs).
 
*<tt>qalter</tt>: Modify a pending or running job.
 
*<tt>qhost</tt>: display state of all hosts.
 
Note that there is also a GUI to SGE, invoked by the command <tt>qmon</tt>
 
... tbc ...
 
==== Array jobs ====
 
... are a very efficient way of managing your jobs under certain circumstances (e.g., if you have to run one identical program many times on different data sets, with different initial conditions, etc.). Please see the corresponding [http://wikis.sun.com/display/GridEngine/Submitting+Extended+Jobs+and+Advanced+Jobs#SubmittingExtendedJobsandAdvancedJobs-SubmittingArrayJobs Section] in the official documentation of Sun Grid Engine
 
...tbc...
 
=== Documentation  ===
 
* [http://wikis.sun.com/display/GridEngine/Using+Sun+Grid+Engine Sun Grid Engine User's Guide]
 
 
== Using the Altix UV 100 system  ==
 
The SGI system is used for very specific applications (in need of a large and highly performant shared-memory system) and can presently only be accessed by the Theoretical Chemistry group. Entitled users may login to the system via <tt>ssh</tt> (the same rules as for the login nodes of the main system apply, i.e. access is only possible from within the intranet of the University, otherwise you have to establish a VPN tunnel):
abcd1234@uv100.hpc.uni-oldenburg.de
 
The Altix UV system has a RHEL 6.0 operating system installed.
As for the IBM cluster, the modules environment is used.
 
=== Compiling and linking applications ===
 
It is strongly recommended to use MPT (Message Passing Toolkit), SGI's own implementation of the MPI standard. It is only then that the highly specialized HW architecture of the system can be fully exploited.
The MPT module must be loaded both for compiling and (in general) at runtime in order for you application to find the dynamically linked libraries:
 
module load mpt
 
Note that MPT is not a compiler. SGI does not provide own compilers for x86-64 based systems. One may, e.g., use the Intel compiler:
 
module load intel/ics
 
Basically, you can use the compiler the same way you are accustomed to, and link against the MPT library by setting the flag <tt>-lmpi</tt>. See the documentation provided below, and also for how to run MPI programs.
 
=== Documentation ===
 
* [http://techpubs.sgi.com/library/manuals/3000/007-3773-003/sgi_html/index.html MPT User's Guide]
 
*[http://docs.sgi.com/library/tpl/cgi-bin/getdoc.cgi?coll=linux&db=bks&srch=&fname=/SGI_Developer/LX_86_AppTune/sgi_html/front.html Application Tuning Guide for SGI X86-64 Based Systems]
 
(Both of the above are also available as PDF download.)
 
 
= Application Software and Libraries  =
 
== Computational Chemistry  ==
 
=== Gaussian 09  ===
 
==== Single-node (multi-threaded) jobs ====
 
You have to use the SMP parallel environment (sse above) to ensure that all slots are on the same host,
as in the following example:
#
#$ -pe smp 12
#$ -R y
#
module load gaussian
g09run myinputfile
#
 
Your input file <tt>myinputfile</tt> (link0 section) must then contain the line:
%NProcShared=12
 
Of course, you also have to reserve sufficent memory and disk space for your job.
 
In the above example, you are reserving all 12 slots that a single host can offer.
If you requested less than 12 slots, other users may have jobs running on that host, too.
 
==== Linda jobs ====
 
Use the <tt>linda</tt> parallel environment (cf. above). The number of requested slots '''must''' be an integer multiple of 12 (= the maximum number of slots per host). E.g., for a Linda job requesting four nodes (Linda workers), the relevant section of the submission script would be:
#
#$ -pe linda 48
#$ -R y
#
module load gaussian
g09run myinputfile
#
 
In order to be started as Linda job, it is mandatory that the input file (here, <tt>myinputfile</tt>) contains the line:
%LindaWorkers=
The wrapper script parses the input file and looks for the <tt>LindaWorkers</tt> keyword (anything after the "=" will be ignored). The wrapper replaces this line and fills in the correct node list for the running job. Note that the <tt>%NProcl</tt> directive of older Gaussian versions is deprecated and should no longer be used.
 
=== MOLCAS  ===
 
not yet installed
 
... tbc ...
 
=== MOLPRO  ===
 
not yet installed
 
... tbc ...
 
 
== Matlab  ==
 
Load the module:
 
module load matlab
 
(automatically loads the newest version, if several versions are installed). Matlab can then be invoked as usually by the command (to be used in scripts etc.):
matlab
 
 
== LEDA  ==
 
To set the correct paths and environment variables load the appropriate module:
 
module load leda/6.3
 
for the single-threaded version, or
 
module load leda/6.3-mt
 
in the case of the multi-threaded library.
 
 
= Advanced Usage  =
 
Here will you will find, among others, hints how to analyse and optimize your programs using HPC tools (profiler, debugger, performance libraries), and other useful information.
 
... tbc ...


</div>
</div>
[[HPC User Wiki 2016]]

Latest revision as of 15:08, 6 June 2017


Picture of nodes.jpg Picture of cluster closed.jpg This is the HPC-Wiki of the University of Oldenburg
 Picture of gpfs.jpg Picture of infinyband.jpg

Basic Information

HPC Facilities Login User environment Compiling and linking Job Management (Queueing) System Altix UV 100 system Examples

Application Software and Libraries

Compiler and Development Tools Quantum Chemistry Computational Fluid Dynamics Mathematics/Scripting Visualisation Libraries

Courses and Tutorials

Introduction to HPC Courses Matlab Tutorials New OS


Contact

HPC Resource EMail

FLOW and HERO
Both (in case of vacation)

Stefan.Harfst@uni-oldenburg.de
hpcuniol@uni-oldenburg.de


Note: This Wiki is under construction and a preliminary version! Contributions are welcome. Please ask Stefan Harfst (Stefan.Harfst(at)uni-oldenburg.de) for further informations.

Only for editors: Formatting rules for this Wiki

HPC User Wiki 2016

Retrieved from "https://wiki.hpcuser.uni-oldenburg.de/index.php?title=Welcome_to_the_HPC_User_Wiki_of_the_University_of_Oldenburg&oldid=3748"