Convert to MD

Author: youminhan

.DS_Store

@@ -0,0 +1,138 @@
+---
+highlighter: none
+layout: default
+title: MPI Use Guide
+---
+
+
+Contents
+========
+
+1.  [Overview of installed MPI libraries](#overview)
+2.  [Using the MPI libraries to compile your code](#lib)
+3.  [Running your job](#run)
+
+This page describes how to use the MPI libraries installed on the HPC
+cluster.
+
+<a name="overview"></a>
+
+**1. Overview of Installed MPI Libraries**
+======================================
+
+There are multiple MPI libraries installed on the cluster, many compiled
+in at least two ways (see below). The SLURM scheduler also has
+integrated libraries for most MPI versions, which you can read about
+[here](https://computing.llnl.gov/linux/slurm/mpi_guide.html).
+
+There are ten different MPI libraries installed:
+
+-   OpenMPI version 1.6.4
+-   OpenMPI version 1.10.2
+-   OpenMPI version 2.0.1
+-   OpenMPI version 2.1.0
+-   OpenMPI version 3.1.1
+-   MPICH version 3.0.4
+-   MPICH version 3.1
+-   MPICH2 version 1.5
+-   MVAPICH2 version 1.9
+-   MVAPICH2 version 2.1
+
+**IMPORTANT: If your software can use OpenMPI or MVAPICH2, these are the
+recommended MPI libraries for CHTC\'s HPC Cluster and will perform the
+fastest on the cluster\'s Infiniband networking.** MPICH and MPICH2 do
+not use Infiniband, by default, and will perform slower than OpenMPI or
+MVAPICH2, though we\'ve configured them to work as well as for
+ethernet-only clusters, so they\'ll still work if your software will
+*only* run with MPICH or MPICH2.
+
+Note that many of our MPI libraries have been compiled with different
+base compilers, to accommodate codes that require these base compilers
+to be built successfully. The compiler used to build MPI can be seen in
+the names of our software modules (see [below](#lib)) and include:
+
+-   `gcc` - the base system version of `gcc`
+-   `intel` - compiled with Intel Composer XE and Intel MPI Library
+    Development Kit 4.1 compilers
+-   `intel-2016` -
+-   `2.1.0-GCC-7.3.0-2.30` - version 7.3 of `gcc`
+
+<a name="lib"></a>
+
+**2. Using the MPI Libraries to Compile Your Code**
+===============================================
+
+In order to successfully compile and run your code using these MPI
+libraries you need to set a few environmental variables. To set these
+variables you will be using the Environmental Modules package
+(<http://modules.sourceforge.net>). This package is very easy to use and
+it will automatically set the environmental variables necessary to use
+the flavor and version of MPI that you need.
+
+First, you are going to want to run the following command to see the
+available modules:
+
+``` 
+[alice@service]$ module avail
+```
+{:.term}
+
+When you run the above command you will receive output similar to this:
+
+``` 
+[alice@service]$ module avail
+---------------------------------------- /etc/modulefiles ----------------------------------------
+mpi/gcc/mpich-3.0.4                 mpi/gcc/openmpi-1.6.4               mpi/intel/mvapich2-1.9
+mpi/gcc/mvapich2-1.9                mpi/intel/mpich-3.0.4               mpi/intel/openmpi-1.6.4
+matlab-r2015b                       compile/intel-2016                  mpi/gcc/openmpi/2.1.0-GCC-7.3.0-2.30
+```
+{:.term}
+
+As you can see, the MPI libraries compiled with GCC compilers are listed
+under `mpi/gcc/` and the MPI libraries compiled with Intel compilers are
+listed under `mpi/intel/`.
+
+To load a module (for example, OpenMPI compiled with GCC compilers),
+simply run this command:
+
+``` 
+[alice@service]$ module load mpi/gcc/openmpi-1.6.4
+```
+{:.term}
+
+Now all necessary environmental variables are set correctly and you can
+go ahead and compile your code!
+
+If you loaded the wrong module, let\'s say MPICH compiled with Intel
+compilers, you can unload it by running:
+
+``` 
+[alice@service]$ module unload mpi/intel/mpich-3.0.4
+```
+{:.term}
+
+You can see what modules you already have loaded by running:
+
+``` 
+[alice@service]$ module list
+```
+{:.term}
+
+**NOTE:** Before using any of the MPI libraries under `mpi/intel/` you
+first need to load the `compile/intel` module.
+
+<a name="run"></a>
+
+
+**3. Running Your Job**
+===================
+
+**NOTE:** To run your job you need the same module loaded as when you
+compiled your code. When you log out of your terminal all loaded modules
+are automatically unloaded.
+
+In order to ensure that your job has the appropriate modules loaded when
+it runs, we recommend adding the `module load` command to your submit
+file, with the appropriate modules. See our sample submit file in the
+[HPC Use Guide](/HPCuseguide.shtml#batch-job) to see what this looks
+like.

MPIuseguide.md

@@ -0,0 +1,58 @@
+---
+layout: default
+title: CHTC Projects
+---
+
+<b>As part of it's many services to UW-Madison and beyond, the CHTC is home to or supports the following projects:</b>
+
+<br>
+
+<div id = "tile-wrapper">
+
+	<a href = "http://research.cs.wisc.edu/htcondor/" class = "tile" id = "condor-tile">
+		<h2>HTCondor project</h2>
+
+		<img src = "images/condor.png">
+
+		<p>
+		The HTCondor project provides the globally-used HTCondor software that powers the CHTC's <i>HTC resources</i>. HTCondor enables the CHTC to
+		maximize job throughput for many simultaneous users, each with different needs. It supports
+		complex workflows, the ability to take checkpoints and then resume jobs, and it enables access
+		to additional computing resources across the globe, like the Open Science Grid.
+		</p>
+
+	</a>
+
+	<a href = "http://www.opensciencegrid.org" class = "tile" id = "osg-tile">
+		<h2>Open Science Grid</h2>
+
+		<img src = "images/osg.png">
+
+		<p>
+		The Open Science Grid (OSG) is a high-throughput computing infrastructure that supports science across the country. 
+		It is an expanding alliance of more than 100 universities, national laboratories, 
+		scientific collaborations, and software developers, all combining their computational resources with one 
+		another for maximal throughput of large-scale computational work. 
+		CHTC users have automatica and FREE access to OSG's considerable computing and storage resources.
+		</p>
+
+	</a>
+
+	<a href = "http://www.neos-server.org" class = "tile" id = "neos-tile">
+		<h2>NEOS Online Optimization Service</h2>
+
+		<img src = "images/neos.png">
+
+		<p>
+		NEOS Online Optimization Service is a publicly-available, high-throughput computing infrastructure that provides solvers for a variety of
+		optimization problems. It also provides background information on the field of optimization,
+		in general, as well as case studies that provide an understanding
+		of optimization problems and approaches to solving them.
+		</p>
+
+	</a>
+
+</div>
+
+<p><a href="/projects.shtml">Learn more</a>
+about how the CHTC has helped advance a diverse body of research on campus.</p>

MPIuseguide.shtml archive_html/MPIuseguide.shtml

@@ -0,0 +1,30 @@
+---
+highlighter: none
+layout: default
+title: CHTC Job Policy Description
+---
+
+These policies relate to the general resources which were funded by
+**WARF** to which **UW** researchers have equal access. There are
+additional **OWNED** resources in the center which are made available
+for Opportunistics use.
+
+-   A user may recycle a slot for up to 12 hours without having to
+    renegotiate for the slot. No single job will be interrupted before
+    it has run for 24 hours. If(for any reason) a job is suspended, this
+    does not count against its run time.
+-   A user may request a whole machine for use (ie a whole machine
+    slot). \~45 of the current 200 machines are available in this
+    capcity. When a user claims a machine, their job will be initially
+    suspended while jobs on any single-core slots complete. Like
+    single-core slots, no whole machine job will be interrupted before
+    it has run for 24 hours.
+-   Groups of users are identified by their associated project. (One
+    user working on one project, counts as one group). This is used tp
+    make accounting more readable.
+-   User Priority Factors in this pool are adjusted to attempt to avoid
+    a group getting a larger share of resources only because it has more
+    users (or logical identities) submitting jobs. This policy currently
+    weights all identified groups equal to the weight of one individual
+    user.
+-   User priority is calculated with a 3 day half life.

chtc-projects.shtml archive_html/chtc-projects.shtml

@@ -0,0 +1,55 @@
+---
+highlighter: none
+layout: default
+title: Getting a Submit Node
+---
+
+
+In order to submit jobs to our campus-wide collection of resources, you
+will need access to a submit node. There are several options for getting
+access to a submit node:
+
+1.  **[Use ours](use-submit-node.shtml).** We operate a submit node that
+    is shared by many researchers. This is a great way to get started
+    quickly, and it is sufficient if you do not need to run tens of
+    thousands of jobs with heavy data transfer requirements.
+2.  **Use your department\'s.** Perhaps your department already has its
+    own submit node, in which case you can contact your local
+    administrator for an account. You will still need to provide all the
+    info requested on the [getting started](get-started.shtml) form, so
+    we can set up things on our end. The benefits of using a
+    departmental or group submit node are: access to data on local file
+    systems; limited impact from other, potentially new users; and,
+    greater scalability in the number of simultaneous jobs you can run,
+    as well as the amount of data you can transfer.
+3.  **Set up a new submit node on a server.** If you do not already have
+    one and need access to data on local file systems, or if you believe
+    that you will have a significant job and/or data volume, getting
+    your own submit node is probably the best way to go. Here\'s an
+    example system configuration that we\'ve found works well for a
+    variety of submit work loads. You can expect to spend around
+    \$4,000 - \$5,000 for such a system.
+
+    **Typical submit node configuration**
+
+    -   A 1U rack-mount enclosure, like a Dell PowerEdge 410.
+    -   Two processors with 12 cores total, for example Intel Xeon
+        E5645, 2.4GHz 6-core processors
+    -   24GB of 1.3 GHz RAM
+    -   Two drives for the operating system. 500GB each is enough. You
+        can use mirroring or a RAID configuration like RAID-6 for
+        reliability.
+    -   Two or more 2-3TB drives for data, depending on your needs.
+
+4.  **Use your desktop.** Depending on your department\'s level of
+    system adminstration support, you may be able to have HTCondor
+    installed on your desktop and configured to submit into our campus
+    resources. Another option that is under development is
+    [Bosco](https://twiki.grid.iu.edu/bin/view/CampusGrids/BoSCO), a
+    user-installable software package that lets you submit jobs into
+    resources managed by HTCondor, PBS or SGE.
+
+Still not sure what option is right for you? No worries. This is one of
+the topics we discuss in our initial consultation. To schedule an
+initial consultation, fill out our [getting started](get-started.shtml)
+form.