Difference between revisions of "Main Page"

From SciNet Users Documentation
Jump to: navigation, search
(Hyperthreading: Logical CPUs vs. cores)
(476 intermediate revisions by 10 users not shown)
Line 1: Line 1:
= Niagara =
{| style="border-spacing:10px; width: 95%"
| style="padding:1em; padding-top:.1em; border:2px solid #0645ad; background-color:#f6f6f6; border-radius:7px"|
==System architecture==
==System Status==
<!-- Use "Up" or "Down"; these are templates. -->
<li>Total of 60,000 Intel x86-64 cores.</li>
<li>1,500 Lenovo SD530 nodes</li>
<li>2x Intel Skylake 6148 CPUs (40 cores @2.4GHz per node) (with hyperthreading to 80 threads &amp; AVX512).</li>
<li>3.02 PFlops delivered / 4.6 PFlops theoretical (would've been #42 on the TOP500 in Nov 2017).</li>
<li>188 GiB / 202 GB RAM per node (at least 4 GiB/core for user jobs).</li>
<li>Operating system: Linux (CentOS 7).
<li>Interconnect: EDR InfiniBand, Dragonfly+ topology with Adaptive Routing</li>
<p>1:1 up to 432 nodes, effectively 2:1 beyond that.</p></li>
<li><p>No GPUs, no local disk.</p></li>
<li><p>Replaces the General Purpose Cluster (GPC) and Tightly Coupled System (TCS).</p></li></ul>
== Using Niagara: Logging in ==
As with all SciNet and CC (Compute Canada) compute systems, access to Niagara is via ssh (secure shell) only.
To access SciNet systems, first open a terminal window (e.g. MobaXTerm on Windows).
Then ssh into the Niagara login nodes with your CC credentials:
<source lang="bash">
$ ssh -Y MYCCUSERNAME@niagara.scinet.utoronto.ca</source>
<source lang="bash">$ ssh -Y MYCCUSERNAME@niagara.computecanada.ca</source>
* The Niagara login nodes are where you develop, edit, compile, prepare and submit jobs.
* These login nodes are not part of the Niagara compute cluster, but have the same architecture, operating system, and software stack.
* The optional <code>-Y</code> is needed to open windows from the Niagara command-line onto your local X server.
* To run on Niagara's compute nodes, you must submit a batch job.
== Migration to Niagara ==
=== Migration for Existing Users of the GPC ===
Niagara is replacing the General Purpose Cluster (GPC) and the Tightly Coupled Cluster (TCS) at SciNet. The TCS was decommissioned last fall, and GPC will be decommissioned very soon: the compute nodes of the GPC will be decommissioned on April 21, 2018, while the storage attached to the GPC will be decommissioned on May 9, 2018.
Active GPC Users got access to the new system, Niagara, on April 9, 2018.
Users' home and project folder were last copied over from the GPC to Niagara on April 5th, 2018, except for files whose name start with a period that were in their home directories (these files were never synced).
It is the user's responsibility to copy over data generated on the GPC after April 5th, 2018.
Data stored in scratch has also not been transfered automatically. Users are to clean up their scratch space on the GPC as much as possible (remember it's temporary data!). Then they can transfer what they need using datamover nodes.
To enable this transfer, there will be a short period during which you can have access to Niagara as well as to the GPC storage resources. This period will end on May 9, 2018.
To copy substantial amounts of data (i.e.,more than 10 GB), please use the datamovers of both the GPC (called gpc-logindm01 and gpc-logindm02) and the Niagara datamovers (called nia-dm1 and nia-dm2). For instance, to copy a directory abc from your GPC scratch to your Niagara scratch directory, you can do the following:
<source lang="bash">
$ ssh CCUSERNAME@niagara.computecanada.ca
$ ssh nia-dm1
$ scp -r SCINETUSERNAME@gpc-logindm01:\$SCRATCH/abc $SCRATCH/abc
For many of you, CCUSERNAME amd SCINETUSERNAME will be the same. Make sure you use the slash (\) before the first $SCRATCH; it cause the value of scratch on the remote node (i.e., here, gpc-logindm01) to be used. Note that the gpc-logindm01 will ask for your SciNet password.
You can also go the other way:
<source lang="bash">
$ ssh SCINETUSERNAME@login.scinet.utoronto.ca
$ ssh gpc-logindm01
$ scp -r $SCRATCH/abc CCUSERNAME@nia-dm1:\$SCRATCH/abc
Again, pay attention to the slash in front of the last occurrence of $SCRATCH.
If you are using rsync, we advice to refrain from using the -a flags, and if using cp, refrain from using the -a and -p flags.
=== For Non-GPC Users ===
Those of you new to SciNet, but with 2018 RAC allocations on Niagara, will have your accounts created and ready for you to login.
New, non-RAC users: we are still working out the procedure to get access. If you can't wait, for now, you can follow the old route of requesting a SciNet Consortium Account on the CCDB site.
== Storage Systems and Locations ==
=== Home and scratch ===
You have a home and scratch directory on the system, whose locations will be given by
For example:
<source lang="bash">
nia-login07:~$ pwd
nia-login07:~$ cd $SCRATCH
nia-login07:rzon$ pwd
=== Project location ===
Users from groups with a RAC allocation will also have a project directory.
'''''IMPORTANT: Future-proof your scripts'''''
Use the environment variables (HOME, SCRATCH, PROJECT) instead of the actual paths!  The paths may change in the future.
=== Storage Limits on Niagara ===
{| class="wikitable"
! location
! quota
!align="right"| block size
! expiration time
! backed up
! on login
! on compute
|{{Up|Jupyter Hub|Jupyter_Hub}}
| 100 GB
|align="right"| 1 MB
|{{Up|File system|Niagara_Quickstart#Storage_and_quotas}}
|{{Up|Burst Buffer|Burst_Buffer}}
| yes
| yes
| read-only
|{{Up|Login Nodes|Niagara_Quickstart#Logging_in}}
| 25 TB
|{{Up|External Network|Niagara_Quickstart#Logging_in}}
|align="right"| 16 MB
| 2 months
| no
| yes
| yes
| by group allocation
|align="right"| 16 MB
| yes
| yes
| yes
| by group allocation
| dual-copy
| no
| no
| ?
|align="right"| 1 MB
| very short
| no
| ?
| ?
<li>Compute nodes do not have local storage.</li>
<li>Archive space is on [https://wiki.scinet.utoronto.ca/wiki/index.php/HPSS HPSS].</li>
<li>Backup means a recent snapshot, not an achive of all data that ever was.</li>
<li><p><code>$BBUFFER</code> stands for the Burst Buffer, a functionality that is still being set up.  This will be a faster parallel storage tier for temporary data.</p></li></ul>
=== Moving data ===
Jan 15, 2021: <b>Globus access update on Mon, Jan/18/2021 and Tue, Jan/19/2021:</b>
<p>Please be advised we start preparations on Monday to perform update to Globus access on Tuesday. We'll be adopting oauth instead of myproxy from that point on. During this period expect sporadic disruptions of service. On Monday we'll already block access to nia-dm2, so please refrain from starting new login sessions or ssh tunnels via nia-dm2 from this weekend already.
'''''Move amounts less than 10GB through the login nodes.'''''
<!-- Current Messages: -->
<!--  When removing system status entries, please archive them to: https://docs.scinet.utoronto.ca/index.php/Previous_messages -->
{|style="border-spacing: 10px;width: 100%"
|valign="top" style="margin: 1em; padding:1em; padding-top:.1em; border:2px solid #000; background-color:#fff; border-radius:7px; width: 49.5%" |
* Only Niagara login nodes visible from outside SciNet.
* Use scp or rsync to niagara.scinet.utoronto.ca or niagara.computecanada.ca (no difference).
* This will time out for amounts larger than about 10GB.
'''''Move amounts larger than 10GB through the datamover nodes.'''''
== QuickStart Guides ==
* [[Niagara Quickstart]]
* [[HPSS | HPSS archival storage]]
* [[SOSCIP_GPU | SOSCIP GPU cluster]]
* [[Mist| Mist Power 9 GPU cluster]]
* [[Teach|Teach cluster]]
* [[FAQ | FAQ (frequently asked questions)]]
* [[Acknowledging SciNet]]
| valign="top" style="margin: 1em; padding:1em; padding-top:.1em; border:2px solid #000; background-color:#fff; border-radius:7px; width: 49.5%" |
* From a Niagara login node, ssh to <code>nia-datamover1</code> or  <code>nia-datamover2</code>.
== Tutorials, Manuals, etc. ==
* Transfers must originate from this datamover.
* [https://support.scinet.utoronto.ca/education/browse.php SciNet education material]
* The other side (e.g. your machine) must be reachable from the outside.
* [https://www.youtube.com/c/SciNetHPCattheUniversityofToronto SciNet's YouTube channel]
* If you do this often, consider using Globus, a web-based tool for data transfer.
* [[Modules specific to Niagara|Software Modules specific to Niagara]]
* [[Commercial software]]
'''''Moving data to HPSS/Archive/Nearline using the scheduler.'''''
* [[Burst Buffer]]
* [[SSH Tunneling]]
* [https://wiki.scinet.utoronto.ca/wiki/index.php/HPSS HPSS] is a tape-based storage solution, and is SciNet's nearline a.k.a. archive facility.
* [[SSH#Two-Factor_authentication|Two-Factor Authentication]]
* Storage space on HPSS is allocated through the annual [https://www.computecanada.ca/research-portal/accessing-resources/resource-allocation-competitions Compute Canada RAC allocation].
* [[Visualization]]
* [[Running Serial Jobs on Niagara]]
== Software and Libraries ==
* [[Jupyter Hub]]
=== Modules ===
Other than essentials, all installed software is made available using module commands. These modules set environment variables (PATH, etc.) This allows multiple, conflicting versions of a given package to be available. module spider shows the available software.
For example:
<source lang="bash">
nia-login07:~$ module spider
The following is a list of the modules currently av
  CCEnv: CCEnv
  NiaEnv: NiaEnv/2018a
  anaconda2: anaconda2/5.1.0
  anaconda3: anaconda3/5.1.0
  autotools: autotools/2017
    autoconf, automake, and libtool
  boost: boost/1.66.0
  cfitsio: cfitsio/3.430
  cmake: cmake/3.10.2 cmake/3.10.3
Common module subcommands are:
<li><p><code>module load &lt;module-name&gt;</code></p>
<p>use particular software</p></li>
<li><p><code>module purge</code></p>
<p>remove currently loaded modules</p></li>
<li><p><code>module spider</code></p>
<p>(or <code>module spider &lt;module-name&gt;</code>)</p>
<p>list available software packages</p></li>
<li><p><code>module avail</code></p>
<p>list loadable software packages</p></li>
<li><p><code>module list</code></p>
<p>list loaded modules</p></li></ul>
On Niagara, there are really two software stacks:
<ol style="list-style-type: decimal;">
<li><p>A Niagara software stack tuned and compiled for this machine. This stack is available by default, but if not, can be reloaded with</p>
<source lang="bash">module load NiaEnv</source></li>
<li><p>The same software stack available on Compute Canada's General Purpose clusters [https://docs.computecanada.ca/wiki/Graham Graham] and [https://docs.computecanada.ca/wiki/Cedar Cedar], compiled (for now) for a previous generation of CPUs:</p>
<source lang="bash">module load CCEnv</source>
<p>If you want the same default modules loaded as on Cedar and Graham, then afterwards also <code>module load StdEnv</code>.</p></li></ol>
Note: the <code>*Env</code> modules are '''''sticky'''''; remove them by <code>--force</code>.
=== Tips for loading software ===
<li><p>We advise '''''against''''' loading modules in your .bashrc.</p>
<p>This could lead to very confusing behaviour under certain circumstances.</p></li>
<li><p>Instead, load modules by hand when needed, or by sourcing a separate script.</p></li>
<li><p>Load run-specific modules inside your job submission script.</p></li>
<li><p>Short names give default versions; e.g. <code>intel</code> → <code>intel/2018.2</code>.</p>
<p>It is usually better to be explicit about the versions, for future reproducibility.</p></li>
<li><p>Handy abbreviations:</p></li></ul>
<pre class="sh">        ml → module list
        ml NAME → module load NAME  # if NAME is an existing module
        ml X → module X</pre>
* Modules sometimes require other modules to be loaded first.<br />
Solve these dependencies by using <code>module spider</code>.
=== Module spider ===
Oddly named, the module subcommand spider is the search-and-advice facility for modules.
Suppose one wanted to load the openmpi module. Upon trying to load the module, one may get the following message:
<source lang="bash">
nia-login07:~$ module load openmpi
Lmod has detected the error:  These module(s) exist but cannot be loaded as requested: "openmpi"
  Try: "module spider openmpi" to see how to load the module(s).
So while that fails, following the advice that the command outputs, the next command would be:
<source lang="bash">
nia-login07:~$ module spider openmpi
  For detailed information about a specific "openmpi" module (including how to load the modules) use
  the module s full name.
  For example:
    $ module spider openmpi/3.1.0rc3
So this gives just more detailed suggestions on using the spider command. Following the advice again, one would type:
<source lang="bash">
nia-login07:~$ module spider openmpi/3.1.0rc3
  openmpi: openmpi/3.1.0rc3
    You will need to load all module(s) on any one of the lines below before the "openmpi/3.1.0rc3"
    module is available to load.
      NiaEnv/2018a  gcc/7.3.0
      NiaEnv/2018a  intel/2018.2
These are concrete instructions on how to load this particular openmpi module. Following these leads to a successful loading of the module.
<source lang="bash">
nia-login07:~$ module load NiaEnv/2018a  intel/2018.2  # note: NiaEnv is usually already loaded
nia-login07:~$ module load openmpi/3.1.0rc3
<source lang="bash">
nia-login07:~$ module list
Currently Loaded Modules:
  1) NiaEnv/2018a (S)  2) intel/2018.2  3) openmpi/3.1.0.rc3
  S:  Module is Sticky, requires --force to unload or purge</source>
== Can I Run Commercial Software? ==
* Possibly, but you have to bring your own license for it.
* SciNet and Compute Canada have an extremely large and broad user base of thousands of users, so we cannot provide licenses for everyone's favorite software.
* Thus, the only commercial software installed on Niagara is software that can benefit everyone: Compilers, math libraries and debuggers.
* That means no Matlab, Gaussian, IDL,
* Open source alternatives like Octave, Python, R are available.
* We are happy to help you to install commercial software for which you have a license.
* In some cases, if you have a license, you can use software in the Compute Canada stack.
== Compiling on Niagara: Example ==
Suppose one want to compile an application from two c source files, appl.c and module.c, which use the Gnu Scientific Library (GSL). This is an example of how this would be done:
<source lang="bash">
nia-login07:~$ module list
Currently Loaded Modules:
  1) NiaEnv/2018a (S)
  S:  Module is Sticky, requires --force to unload or purge
nia-login07:~$ module load intel/2018.2 gsl/2.4
nia-login07:~$ ls
appl.c module.c
nia-login07:~$ icc -c -O3 -xHost -o appl.o appl.c
nia-login07:~$ icc -c -O3 -xHost -o module.o module.c
nia-login07:~$ icc  -o appl module.o appl.o -lgsl -mkl
nia-login07:~$ ./appl
* The optimization flags -O3 -xHost allow the Intel compiler to use instructions specific to the architecture CPU that is present (instead of for more generic x86_64 CPUs).
* The GSL requires a cblas implementation, for is contained in the Intel Math Kernel Library (MKL). Linking with this library is easy when using the intel compiler, it just requires the -mkl flags.
* If compiling with gcc, the optimization flags would be -O3 -march=native. For the way to link with the MKL, it is suggested to use the [https://software.intel.com/en-us/articles/intel-mkl-link-line-advisor MKL link line advisor].
== Testing ==
You really should test your code before you submit it to the cluster to know if your code is correct and what kind of resources you need.
<li><p>Small test jobs can be run on the login nodes.</p>
<p>Rule of thumb: couple of minutes, taking at most about 1-2GB of memory, couple of cores.</p></li>
<li><p>You can run the the ddt debugger on the login nodes after <code>module load ddt</code>.</p></li>
<li><p>Short tests that do not fit on a login node, or for which you need a dedicated node, request an<br />
interactive debug job with the salloc command</p>
<source lang="bash">nia-login07:~$ salloc -pdebug --nodes N --time=1:00:00</source>
<p>where N  is the number of nodes. The duration of your interactive debug session can be at most one hour, can use at most 4 nodes, and each user can only have one such session at a time.</p>
<p>Alternatively, on Niagara, you can use the command</p>
<source lang="bash">nia-login07:~$ debugjob N</source>
<p>where N is the number of nodes, If N=1, this gives an interactive session one 1 hour, when N=4 (the maximum), it give you 30 minutes.</p>
== Submitting jobs ==
Niagara uses SLURM as its job scheduler.
You submit jobs from a login node by passing a script to the sbatch command:
<source lang="bash">nia-login07:~$ sbatch jobscript.sh</source>
This puts the job in the queue. It will run on the compute nodes in due course.
Jobs will run under their group's RRG allocation, or, if the group has none, under a RAS allocation (previously called `default' allocation).
Keep in mind:
<li><p>Scheduling is by node, so in multiples of 40-cores.</p></li>
<li><p>Maximum walltime is 24 hours.</p></li>
<li><p>Jobs must write to your scratch or project directory (home is read-only on compute nodes).</p></li>
<li><p>Compute nodes have no internet access.</p>
<p>Download data you need beforehand on a login node.</p></li></ul>
== Scheduling by Node ==
<li><p>All job resource requests on Niagara are scheduled as a multiple of '''nodes'''.</p></li>
<li>The nodes that your jobs run on are exclusively yours.
<li>No other users are running anything on them.</li>
<li>You can ssh into them to see how things are going.</li></ul>
<li><p>Whatever your requests to the scheduler, it will always be translated into a multiple of nodes.</p></li>
<li><p>Memory requests to the scheduler are of no use.</p>
<p>Your job gets N x 202GB of RAM if N is the number of nodes.</p></li>
<li><p>You should '''use all 40 cores on each of the nodes''' that your job uses.</p>
<p>You will be contacted if you don't, and we will help you get more science done.</p></li></ul>
== Hyperthreading: Logical CPUs vs. cores ==
* Hyperthreading, a technology that leverages more of the physical hardware by pretending there are twice as many logical cores than real ones, is enabled on Niagara.
* So the OS and scheduler see 80 logical cores.
* 80 logical cores vs. 40 real cores typically gives about a 5-10% speedup (YMMV).
'''Because Niagara is scheduled by node, hyperthreading is actually fairly easy to use:'''
<li><p>Ask for a certain number of nodes N for your jobs.</p></li>
<li><p>You know that you get 40xN cores, so you will use (at least) a total of 40xN mpi processes or threads.</p>
<p>(mpirun, srun, and the OS will automaticallly spread these over the real cores)</p></li>
<li><p>But you should also test if running 80xN mpi processes or threads gives you any speedup.</p></li>
<li><p>Regardless, your usage will be counted as 40xNx(walltime in years).</p></li></ul>
=== Example submission script (OpenMP) ===
<source lang="bash">#!/bin/bash
#SBATCH --nodes=1
#SBATCH --cpus-per-task=40
#SBATCH --time=1:00:00
#SBATCH --job-name openmp_job
#SBATCH --output=openmp_output_%j.txt
module load intel/2018.2
# or "srun ./openmp_example".
Submit this script with the command:
<source lang="bash">nia-login07:~$ sbatch openmp_job.sh</source>
* First line indicates that this is a bash script.
* Lines starting with <code>#SBATCH</code> go to SLURM.
* sbatch reads these lines as a job request (which it gives the name <code>openmp_job</code>) .
* In this case, SLURM looks for one node with 40 cores to be run inside one task, for 1 hour.
* Once it found such a node, it runs the script:
** Change to the submission directory;
** Loads modules;
** Sets an environment variable;
** Runs the <code>openmp_example</code> application.
* To use hyperthreading, just change <code>--cpus-per-task=40</code> to <code>--cpus-per-task=80</code>.
=== Example submission script (MPI) ===
<source lang="bash">#!/bin/bash
#SBATCH --nodes=8
#SBATCH --ntasks=320
#SBATCH --time=1:00:00
#SBATCH --job-name mpi_job
#SBATCH --output=mpi_output_%j.txt
module load intel/2018.2
module load openmpi/3.1.0rc3
mpirun ./mpi_example
# or "srun ./mpi_example"
Submit this script with the command:
<source lang="bash">nia-login07:~$ sbatch mpi_job.sh</source>
<li><p>First line indicates that this is a bash script.</p></li>
<li><p>Lines starting with <code>#SBATCH</code> go to SLURM.</p></li>
<li><p>sbatch reads these lines as a job request (which it gives the name <code>mpi_job</code>)</p></li>
<li><p>In this case, SLURM looks for 8 nodes with 40 cores on which to run 320 tasks, for 1 hour.</p></li>
<li><p>Once it found such a node, it runs the script:</p>
<li>Change to the submission directory;</li>
<li>Loads modules;</li>
<li>Runs the <code>mpi_example</code> application.</li>
<li>To use hyperthreading, just change --ntasks=320 to --ntasks=640, and add --bind-to none to the mpirun command (the latter is necessary for OpenMPI only, not when using IntelMPI).</li>
== Monitoring queued jobs ==
Once the job is incorporated into the queue, there are some command you can use to monitor its progress.
<li><p><code>squeue</code> or <code>qsum</code> to show the job queue (<code>squeue -u $USER</code> for just your jobs);</p></li>
<li><p><code>squeue -j JOBID</code> to get information on a specific job</p>
<p>(alternatively, <code>scontrol show job JOBID</code>, which is more verbose).</p></li>
<li><p><code>squeue --start -j JOBID</code> to get an estimate for when a job will run; these tend not to be very accurate predictions.</p></li>
<li><p><code>scancel -i JOBID</code> to cancel the job.</p></li>
<li><p><code>sinfo -pcompute</code> to look at available nodes.</p></li>
<li><p><code>jobperf JOBID</code> to get an instantaneous view of the cpu and memory usage of the nodes of the job while it is running.</p></li>
<li><p><code>sacct</code> to get information on your recent jobs.</p></li>
<li><p>More utilities like those that were available on the GPC are under development.</p></li></ul>
== Data Management and I/O Tips ==
* $HOME, $SCRATCH, and $PROJECT all use the parallel file system called GPFS.
* Your files can be seen on all Niagara login and compute nodes.
* GPFS is a high-performance file system which provides rapid reads and writes to large data sets in parallel from many nodes.
* But accessing data sets which consist of many, small files leads to poor performance.
* Avoid reading and writing lots of small amounts of data to disk.<br />
* Many small files on the system would waste space and would be slower to access, read and write.
* Write data out in binary. Faster and takes less space.
* Burst buffer (to come) is better for i/o heavy jobs and to speed up checkpoints.
== Further information ==
=== Useful sites ===
* SciNet: https://www.scinet.utoronto.ca
* Niagara: https://docs.computecanada.ca/wiki/niagara
* System Status: https://wiki.scinet.utoronto.ca/wiki/index.php/System_Alerts
* Training: https://support.scinet.utoronto.ca/education
=== Support ===
* support@scinet.utoronto.ca
* niagara@computecanada.ca

Revision as of 20:46, 15 January 2021

System Status

Niagara HPSS Mist Teach
Jupyter Hub Scheduler File system Burst Buffer
Login Nodes External Network Globus

Jan 15, 2021: Globus access update on Mon, Jan/18/2021 and Tue, Jan/19/2021:

Please be advised we start preparations on Monday to perform update to Globus access on Tuesday. We'll be adopting oauth instead of myproxy from that point on. During this period expect sporadic disruptions of service. On Monday we'll already block access to nia-dm2, so please refrain from starting new login sessions or ssh tunnels via nia-dm2 from this weekend already.

QuickStart Guides

Tutorials, Manuals, etc.