OpenHPC: Project Overview and Updates

OpenHPC: Project Overview and
Updates
Karl W. Schulz, Ph.D.
Software and Services Group, Intel
Technical Project Lead, OpenHPC
5th Annual MVAPICH User Group (MUG) Meeting
August 16, 2017 w Columbus, Ohio
http://openhpc.community

Outline
• Brief project overview
• New items/updates from last year
• Highlights from latest release
2

OpenHPC: Mission and Vision
• Mission: to provide a reference collection of open-source HPC
software components and best practices, lowering barriers to
deployment, advancement, and use of modern HPC methods
and tools.
• Vision: OpenHPC components and best practices will enable
and accelerate innovation and discoveries by broadening
access to state-of-the-art, open-source HPC methods and tools
in a consistent environment, supported by a collaborative,
worldwide community of HPC users, developers, researchers,
administrators, and vendors.
3

OpenHPC: a brief History…
4
ISC’15
BoF on the
merits/interest in a
community effort
SC’15
seed initial 1.0 release,
gather interested parties
to work with Linux
Foundation
ISC’16
v1.1.1 release
Linux Foundation
announces technical
leadership, founding
members, and formal
governance structure
June 2015
Nov 2015
June 2016
SC’16
v1.2 release
BoF
Nov 2016
June 2017
ISC’17
v1.3.1 release
BoF
MUG’16
OpenHPC Intro

OpenHPC Project Members
5
Project member participation interest? Please contact
Jeff ErnstFriedman jernstfriedman@linuxfoundation.org
Argonne
National
Laboratory
Indiana
University
University of
Cambridge
Mixture of Academics, Labs,
OEMs, and ISVs/OSVs

6
Governance: Technical Steering Committee (TSC)
Role Overview
OpenHPC
Technical Steering Committee (TSC)
Maintainers
Integration
Testing
Coordinator(s)
Upstream Component
Development
Representative(s)
Project
Leader
End-User / Site
Representative(s)
Note: We just completed election of TSC
members for the 2017-2018 term.
• terms are for 1-year

7
OpenHPC TSC – Individual Members
• Reese Baird, Intel (Maintainer)
• David Brayford, LRZ (Maintainer)
• Eric Coulter, Indiana University (End-User/Site Representative)
• Leonordo Fialho, ATOS (Maintainer)
• Todd Gamblin, LLNL (Component Development Representative)
• Craig Gardner, SUSE (Maintainer)
• Renato Golin, Linaro (Testing Coordinator)
• Jennifer Green, Los Alamos National Laboratory (Maintainer)
• Douglas Jacobsen, NERSC (End-User/Site Representative)
• Chulho Kim, Lenovo (Maintainer)
• Janet Lebens, Cray (Maintainer)
• Thomas Moschny, ParTec (Maintainer)
• Nam Pho, New York Langone Medical Center (Maintainer)
• Cyrus Proctor, Texas Advanced Computing Center (Maintainer)
• Adrian Reber, Red Hat (Maintainer)
• Joseph Stanfield, Dell (Maintainer)
• Karl W. Schulz, Intel (Project Lead, Testing Coordinator)
• Jeff Schutkoske, Cray (Component Development Representative)
• Derek Simmel, Pittsburgh Supercomputing Center (End-User/Site Representative)
• Scott Suchyta, Altair (Maintainer)
• Nirmala Sundararajan, Dell (Maintainer)
https://github.com/openhpc/ohpc/wiki/Governance-Overview
New for 2017-2018

8
Target system architecture
• Basic cluster architecture:
head node (SMS) + computes
• Ethernet fabric for mgmt.
network
• Shared or separate out-of-band
(BMC) network
• High speed fabric (InfiniBand,
Omni-Path)
to compute eth interface
to compute BMC interface
compute
nodes
high speed network
Parallel File System
tcp networking
eth0 eth1

[ Key takeaway ]
• OpenHPC provides a collection of pre-built ingredients common in
HPC environments; fundamentally it is a package repository
• The repository is published for use with Linux distro package
managers:
• yum (CentOS/RHEL)
• zypper (SLES)
• You can pick relevant bits of interest for your site:
• if you prefer a resource manager that is not included, you can build
that locally and still leverage the scientific libraries and development
environment
• similarly, you might prefer to utilize a different provisioning system
OpenHPC: a building block repository
9

10
Newish Items/Updates
changes and new items since we were last together at
MUG’16

11
• For those who prefer to mirror a repo
locally, we have historically provided
an ISO that contained all the
packages/repodata
• Beginning with v1.2 release,
switched to tarball based distribution
• Distribution tarballs
available at:
http://build.openhpc.community/dist
• A “make_repo.sh” script is provided
that will setup a locally hosted
OpenHPC repository using the
contents from downloaded tarball
Switched from ISOs -> distribution tarballs
Index of /dist/1.2.1
Name Last modified Size Description
Parent Directory -
OpenHPC-1.2.1.CentOS_7.2_aarch64.tar 2017-01-24 12:43 1.3G
OpenHPC-1.2.1.CentOS_7.2_src.tar 2017-01-24 12:45 6.8G
OpenHPC-1.2.1.CentOS_7.2_x86_64.tar 2017-01-24 12:43 2.2G
OpenHPC-1.2.1.SLE_12_SP1_aarch64.tar 2017-01-24 12:40 1.1G
OpenHPC-1.2.1.SLE_12_SP1_src.tar 2017-01-24 12:42 6.2G
OpenHPC-1.2.1.SLE_12_SP1_x86_64.tar 2017-01-24 12:41 1.9G
OpenHPC-1.2.1.md5s 2017-01-24 12:50 416
# tar xf OpenHPC-1.2.1.CentOS_7.2_x86_64.tar
# ./make_repo.sh
--> Creating OpenHPC.local.repo file in /etc/yum.repos.d
--> Local repodata stored in /root/repo
# yum repolist | grep OpenHPC
OpenHPC-local OpenHPC-1.2 - Base
OpenHPC-local-updates OpenHPC-1.2.1 - Updates

More Generic Repo Paths
• Starting with the v1.3 release, we adopted more generic paths
for underlying distros
12
[OpenHPC]
name=OpenHPC-1.3 - Base
baseurl=http://build.openhpc.community/OpenHPC:/1.3/CentOS_7
gpgcheck=1
gpgkey=file:///etc/pki/rpm-gpg/RPM-GPG-KEY-OpenHPC-1
[OpenHPC-updates]
name=OpenHPC-1.3 - Updates
baseurl=http://build.openhpc.community/OpenHPC:/1.3/updates/CentOS_7
gpgcheck=1
gpgkey=file:///etc/pki/rpm-gpg/RPM-GPG-KEY-OpenHPC-1
Similar approach for SLES12 repo config -> SLE_12

Release Roadmap Published
13
Clone this wiki locally
Roadmap
The table below highlights expected future releases that are currently on the near term
OpenHPC roadmap. Note that these are dynamic in nature and will be updated based on
directions from the OpenHPC TSC. Approximate release date(s) are provided to aid in
planning, but please be aware that these dates and releases are subject to change, YMMV,
etc.
Release Target Release Date Expectations
1.3.2 August 2017 New component additions and version upgrades.
1.3.3 November 2017 New component additions and version upgrades.
Previous Releases
A history of previous OpenHPC releases is highlighted below. Clicking on the version string will
take you to the !Release!Notes!for more detailed information on the changes in a particular
release.
Release Date
1.3.1 June 16, 2017
1.3 March 31, 2017
1.2.1 January 24, 2017
1.2 November 12, 2016
1.1.1 June 21, 2016
1.1 April 18, 2016
1.0.1 February 05, 2016
1.0 November 12, 2015
Add a custom footer
Pages 12
Home
ARM Tech Preview
ARM Tech Preview (1.3.1)
Component List
Component Suggestions
Governance Overview
Open Build Service
Papers and Presentations
Release History and
Roadmap
Technical Steering
Committee Meetings
TSC Membership
Participation and Selection
Process
TSC Weekly Meeting Call
Info:
Add a custom sidebar
https://github.com/openhpc/ohpc.wiki.git
Clone in Desktop
• Have had some requests for a roadmap for future
releases
• High-level roadmap now maintained on GitHub wiki:
https://github.com/openhpc/ohpc/wiki/Release-History-
and-Roadmap

Component Submission Site
• A common question posed to the
project has been how to request new
software components?
• We now have a simple submission site
for new requests:
- https://github.com/openhpc/submissions
- requests reviewed on rolling basis at
roughly a quarterly cadence
• Example software/recipes that have
been added via this process:
- xCAT
- BeeGFS
- PLASMA, SLEPc
- clang/LLVM
koomie commented 15 days ago
Software Name
Public URL
Technical Overview
Latest stable version number
Open-source license type
Relationship to component?
contributing developer
user
other
If other, please describe:
Build system
autotools-based
CMake
other
If other, please describe:
Does the current build system support staged path installati
For example: !make!install!DESTIR=/tmp/foo!(or equivalent)
Subset of information requested
during submission process
14
- MPICH
- PBS Professional
- Singularity
- Scalasca

Opt-in System Registry Now Available
• Interested users can
now register their
usage on a public
system registry
• Helpful for us to have
an idea as to who is
potentially benefitting
from this community
effort
• Accessible from top-
level GitHub page
15
OpenHPC System Registry
This opt-in form can be used to register your system to let us (and the community) know that you
are using elements of OpenHPC.
* Required
CentOS/RHEL
SLES
Other:
Name of Site/Organization *
Your answer
What OS distribution are you using? *
Site or System URL
Your answer
System Registry

Multiple Architecture Builds
• Starting with v1.2 release, we also include builds for aarch64
- both SUSE and RHEL/CentOS now have aarch64 variants available for
latest versions (SLES 12 SP2, CentOS 7.3)
• Recipes/packages being made available as a Tech Preview
- some additional work required for provisioning
- significant majority of development packages testing OK, but there are a
few known caveats
- please see https://github.com/openhpc/ohpc/wiki/ARM-Tech-Preview for
latest info
v1.3.1 RPM counts
16
misc better tar excludes
tests ﬁx test path (#472)
README.md Updating available package counts to reﬂect fact that some of packag…
README.md
OpenHPC: Community building blocks for HPC systems. (v1.3.1)
components availablecomponents available 6767 new additionsnew additions 33 updatesupdates 44%44%
Introduction
This stack provides a variety of common, pre-built ingredients required to deploy and manage an HPC Linux
including provisioning tools, resource management, I/O clients, runtimes, development tools, and a variety o
libraries.
The compatible OS version(s) for this release and the total number of pre-packaged binary RPMs available p
architecture type are summarized as follows:
Base OS x86_64 aarch64 noarch
CentOS 7.3 424 265 32
SLES 12 SP2 429 274 32
A detailed list of all available components is available in the "Package Manifest" appendix located in each of
companion install guide documents, and a list of updated packages can be found in the release notes. The r
notes also contain important information for those upgrading from previous versions of OpenHPC.

Dev Environment Consistency
17
x86_64
aarch64
OpenHPC providing
consistent
development
environment to the
end user across
multiple architectures

18
End-user software addition tools
OpenHPC repositories include two additional tools that can be used to
further extend a user’s development environment
- EasyBuild and Spack
- leverages other community efforts for build reproducibility and best
practices for configuration
- modules available for both after install
# module load gcc7
# module load spack
# spack compiler find
==> Added 1 new compiler to /root/.spack/linux/compilers.yaml gcc@7.1.0
# spack install darshan-runtime
...
# . /opt/ohpc/admin/spack/0.10.0/share/spack/setup-env.sh
# module avail
----- /opt/ohpc/admin/spack/0.10.0/share/spack/modules/linux-sles12-x86_64 -----
darshan-runtime-3.1.0-gcc-7.1.0-vhd5hhg m4-1.4.17-gcc-7.1.0-7jd575i
hwloc-1.11.4-gcc-7.1.0-u3k6dok ncurses-6.0-gcc-7.1.0-l3mdumo
libelf-0.8.13-gcc-7.1.0-tsgwr7j openmpi-2.0.1-gcc-7.1.0-5imqlfb
libpciaccess-0.13.4-gcc-7.1.0-33gbduz pkg-config-0.29.1-gcc-7.1.0-dhbpa2i
libsigsegv-2.10-gcc-7.1.0-lj5rntg util-macros-1.19.0-gcc-7.1.0-vkdpa3t
libtool-2.4.6-gcc-7.1.0-ulicbkz zlib-1.2.10-gcc-7.1.0-gy4dtna

Variety of recipes now available
Choose your own adventure…
Initially, we started
off with a single
recipe with the
intent to expandOpenHPC (v1.0.1)
Cluster Building Recipes
CentOS7.1 Base OS
Base Linux* Edition
Document Last Update: 2016-02-05
Document Revision: 7e93115
Latest v1.3.1 release continues to expand with multiple
resource managers, OSes, provisioners, and architectures:
• Install_guide-CentOS7-Warewulf-PBSPro-1.3.1-x86_64.pdf
• Install_guide-CentOS7-Warewulf-SLURM-1.3.1-aarch64.pd
• fInstall_guide-CentOS7-Warewulf-SLURM-1.3.1-x86_64.pdf
• Install_guide-CentOS7-xCAT-SLURM-1.3.1-x86_64.pdf
• Install_guide-SLE_12-Warewulf-PBSPro-1.3.1-x86_64.pdf
• Install_guide-SLE_12-Warewulf-SLURM-1.3.1-aarch64.pdf
• fInstall_guide-SLE_12-Warewulf-SLURM-1.3.1-x86_64.pdf
• Additional resource
manager (PBS
Professional)
with v1.2
• Additional provisioner
(xCAT) with v1.3.1
19

20
Template scripts
Template recipe scripts are proved that encapsulate commands presented in
the guides:
# yum/zypper install docs-ohpc
# ls /opt/ohpc/pub/doc/recipes/*/*/*/*/recipe.sh
/opt/ohpc/pub/doc/recipes/centos7/aarch64/warewulf/slurm/recipe.sh
/opt/ohpc/pub/doc/recipes/centos7/x86_64/warewulf/pbspro/recipe.sh
/opt/ohpc/pub/doc/recipes/centos7/x86_64/warewulf/slurm/recipe.sh
/opt/ohpc/pub/doc/recipes/centos7/x86_64/xcat/slurm/recipe.sh
/opt/ohpc/pub/doc/recipes/sles12/aarch64/warewulf/slurm/recipe.sh
/opt/ohpc/pub/doc/recipes/sles12/x86_64/warewulf/pbspro/recipe.sh
/opt/ohpc/pub/doc/recipes/sles12/x86_64/warewulf/slurm/recipe.sh
# ls /opt/ohpc/pub/doc/recipes/*/input.local
/opt/ohpc/pub/doc/recipes/centos7/input.local
/opt/ohpc/pub/doc/recipes/sles12/input.local
input.local + recipe.sh == installed system
# compute hostnames
c_name[0]=c1
c_name[1]=c2
…
# compute node MAC addresses
c_mac[0]=00:1a:2b:3c:4f:56
c_mac[1]=00:1a:2b:3c:4f:56
…

Test Suite
• Initiated from discussion/requests at SC’16 BoF, the OpenHPC test
suite is now available as an installable RPM (introduced with v1.3
release)
• # yum/zypper install test-suite-ohpc
- creates/relies on “ohpc-test” user to perform user testing (with
accessibility to run jobs through resource manager)
- related discussion added to recipes in Appendix C
but is instead, devised to confirm component builds are functional and interoperable within the modular
OpenHPC environment. The test suite is generally organized by components and the OpenHPC CI workflow
relies on running the full suite using Jenkins to test multiple OS configurations and installation recipes. To
facilitate customization and running of the test suite locally, we provide these tests in a standalone RPM.
[sms]# yum -y install test-suite-ohpc
The RPM installation creates a user named ohpc-test to house the test suite and provide an isolated
environment for execution. Configuration of the test suite is done using standard GNU autotools semantics
and the BATS shell-testing framework is used to execute and log a number of individual unit tests. Some
tests require privileged execution, so a di↵erent combination of tests will be enabled depending on which user
executes the top-level configure script. Non-privileged tests requiring execution on one or more compute
nodes are submitted as jobs through the SLURM resource manager. The tests are further divided into
“short” and “long” run categories. The short run configuration is a subset of approximately 180 tests to
demonstrate basic functionality of key components (e.g. MPI stacks) and should complete in 10-20 minutes.
The long run (around 1000 tests) is comprehensive and can take an hour or more to complete.
Most components can be tested individually, but a default configuration is setup to enable collective
testing. To test an isolated component, use the configure option to disable all tests, then re-enable the
desired test to run. The --help option to configure will display all possible tests. Example output is
shown below (some output is omitted for the sake of brevity).
[sms]# su - ohpc-test
[test@sms ~]$ cd tests
[test@sms ~]$ ./configure --disable-all --enable-fftw
checking for a BSD-compatible install... /bin/install -c
checking whether build environment is sane... yes
...
---------------------------------------------- SUMMARY ---------------------------------------------
Package version............... : test-suite-1.3.0
Build user.................... : ohpc-test
Build host.................... : sms001
Configure date................ : 2017-03-24 15:41
Build architecture............ : x86 64
Compiler Families............. : gnu
MPI Families.................. : mpich mvapich2 openmpi
Resource manager ............. : SLURM
Test suite configuration...... : short
...
Libraries:
21

Project CI infrastructure
• TACC is kindly hosting some CI
infrastructure for the project
(Austin, TX)
• Using for build servers and
continuous integration (CI)
testbed.
http://test.openhpc.community:8080
Many thanks to TACC and
vendors for hardware
donations!!: Intel, Cavium, Dell
x86_64
aarch64
22

23
Community Test System for CI in use
http://test.openhpc.community:8080
All recipes exercised in CI system (start w/ bare-metal installs + integration test suite)
nHPC CI Infrastructure
nks to the Texas Advanced Computing Center (TACC) for hosting support and to Intel, Cavium, and Dell for hardware donations.
add description
W Name ↓ Last Success Last Failure Last Duration
(1.2) - (centos7.2,x86_64) - (warewulf+pbspro) - long cycle 1 day 8 hr - #39 N/A 58 min
(1.2) - (centos7.2,x86_64) - (warewulf+pbspro) - short cycle 1 day 4 hr - #155 4 days 10 hr - #109 14 min
(1.2) - (centos7.2,x86_64) - (warewulf+slurm) - long cycle 2 days 4 hr - #244 4 days 4 hr - #219 1 hr 0 min
(1.2) - (centos7.2,x86_64) - (warewulf+slurm) - short cycle 2 hr 46 min - #554 8 days 15 hr - #349 14 min
(1.2) - (centos7.2,x86_64) - (warewulf+slurm+PXSE) - long cycle 1 day 6 hr - #39 4 days 10 hr - #20 2 hr 29 min
(1.2) - (sles12sp1,x86_64) - (warewulf+pbspro) - short cycle 1 day 3 hr - #166 4 days 10 hr - #86 12 min
(1.2) - (sles12sp1,x86_64) - (warewulf+slurm) - short cycle 1 day 2 hr - #259 8 days 20 hr - #72 14 min
(1.2) - (sles12sp1,x86_64) - (warewulf,slurm) - long test cycle 1 day 5 hr - #97 6 days 19 hr - #41 54 min
(1.2) - aarch64 - (centos7.2) - (warewulf+slurm) 2 days 21 hr - #3 N/A 0.41 sec
(1.2) - aarch64 - (sles12sp1) - (warewulf+slurm) 1 day 8 hr - #45 2 days 21 hr - #41 2 hr 13 min
n: S M L Legend RSS for all RSS for failures RSS for just latest builds
KARL W. SCHULZ | LOG OUTsearch
1.1 All Interactive admin +1.2
ENABLE AUTO REFRESH
New Item
People
Build History
Edit View
Project Relationship
Check File Fingerprint
Manage Jenkins
My Views
Lockable Resources
Credentials
No builds in the queue.
master
1 Idle
2 Idle
3 Idle
OpenHPC CI Infrastructure
Thanks to the Texas Advanced Computing Center (TACC) for hosting support and to Intel, Cavium, and
S W Name ↓ Last Success
(1.2) - (centos7.2,x86_64) - (warewulf+pbspro) - long cycle 1 day 8 hr - #39
(1.2) - (centos7.2,x86_64) - (warewulf+pbspro) - short cycle 1 day 4 hr - #155
(1.2) - (centos7.2,x86_64) - (warewulf+slurm) - long cycle 2 days 4 hr - #244
(1.2) - (centos7.2,x86_64) - (warewulf+slurm) - short cycle 2 hr 46 min - #554
(1.2) - (centos7.2,x86_64) - (warewulf+slurm+PXSE) - long cycle 1 day 6 hr - #39
(1.2) - (sles12sp1,x86_64) - (warewulf+pbspro) - short cycle 1 day 3 hr - #166
(1.2) - (sles12sp1,x86_64) - (warewulf+slurm) - short cycle 1 day 2 hr - #259
(1.2) - (sles12sp1,x86_64) - (warewulf,slurm) - long test cycle 1 day 5 hr - #97
(1.2) - aarch64 - (centos7.2) - (warewulf+slurm) 2 days 21 hr - #3
(1.2) - aarch64 - (sles12sp1) - (warewulf+slurm) 1 day 8 hr - #45
Icon: S M L Legend RSS for all
search
Build Queue
Build Executor Status
1.1.1 All Interactive admin +1.2
Jenkins
1.3.x
OpenHPC CI Infrastructure
Thanks to the Texas Advanced Computing Center (TACC) for hosting support and to Intel, Cavium, and Dell for hardware
donations.
add description
S Name ↓ Last Success Last Duration
(1.3 to 1.3.1) - (centos7.3,x86_64) - (warewulf+slurm) 2 days 12 hr - #41 1 hr 12 min
(1.3.1) - (centos7.3,x86_64) - (warewulf+pbspro) - UEFI 2 days 7 hr - #390 56 min
(1.3.1) - (centos7.3,x86_64) - (warewulf+slurm) - long cycle 2 days 3 hr - #892 1 hr 2 min
(1.3.1) - (centos7.3,x86_64) - (warewulf+slurm) - tarball REPO 2 days 5 hr - #48 1 hr 2 min
(1.3.1) - (centos7.3,x86_64) - (warewulf+slurm+PSXE) 2 days 4 hr - #726 2 hr 14 min
(1.3.1) - (centos7.3,x86_64) - (warewulf+slurm+PSXE+OPA) 2 days 7 hr - #80 1 hr 51 min
(1.3.1) - (centos7.3,x86_64) - (xcat+slurm) 2 days 13 hr - #271 1 hr 1 min
(1.3.1) - (sles12sp2,x86_64) - (warewulf+pbspro) - tarball REPO 2 days 8 hr - #45 44 min
(1.3.1) - (sles12sp2,x86_64) - (warewulf+pbspro) - UEFI 2 days 4 hr - #70 45 min
(1.3.1) - (sles12sp2,x86_64) - (warewulf+slurm) 2 days 8 hr - #780 53 min
(1.3.1) - (sles12sp2,x86_64) - (warewulf+slurm+PSXE) - long cycle 2 days 3 hr - #114 1 hr 53 min
(1.3.1) - (sles12sp2,x86_64) - (warewulf+slurm+PSXE+OPA) 2 days 8 hr - #36 1 hr 33 min
KARL W. SCHULZ | LOG OUTsearch1
1.3 All +1.3.1
ENABLE AUTO REFRESH

24
OpenHPC v1.3.1 Release
June 16, 2017

25
OpenHPC v1.3.1 - Current S/W components
Functional Areas Components
Base OS CentOS 7.3, SLES12 SP2
Architecture x86_64, aarch64 (Tech Preview)
Administrative Tools
Conman, Ganglia, Lmod, LosF, Nagios, pdsh, pdsh-mod-slurm, prun, EasyBuild,
ClusterShell, mrsh, Genders, Shine, Spack, test-suite
Provisioning Warewulf, xCAT
Resource Mgmt. SLURM, Munge, PBS Professional
Runtimes OpenMP, OCR, Singularity
I/O Services Lustre client (community version), BeeGFS client
Numerical/Scientific
Libraries
Boost, GSL, FFTW, Metis, PETSc, Trilinos, Hypre, SuperLU, SuperLU_Dist,
Mumps, OpenBLAS, Scalapack
I/O Libraries HDF5 (pHDF5), NetCDF (including C++ and Fortran interfaces), Adios
Compiler Families GNU (gcc, g++, gfortran),
MPI Families MVAPICH2, OpenMPI, MPICH
Development Tools Autotools (autoconf, automake, libtool), Valgrind,R, SciPy/NumPy, hwloc
Performance Tools PAPI, IMB, mpiP, pdtoolkit TAU, Scalasca, ScoreP, SIONLib
Notes:
• Additional
dependencies that are
not provided by the
BaseOS or community
repos (e.g. EPEL) are
also included
• 3rd Party libraries are
built for each
compiler/MPI family
• Resulting repositories
currently comprised of
~450 RPMs
new with v1.3.1
Future additions
approved for inclusion
in v1.3.2 release:
• PLASMA
• SLEPc
• pNetCDF
• Scotch
• Clang/LLVM

OpenHPC Component Updates
• Part of motivation for community effort like OpenHPC is the rapidity of
S/W updates in our space
• Rolling history of updates/additions:
v1.1$Release$
Updated$$ New$Addi2on$ No$change$
v1.2%Release%
Updated%% New%Addi3on% No%change%
v1.3%Release%
38.6%
63.3%
29.1%
v1.3.1%Release%
44%
Apr 2016 Nov 2016 Mar 2017 June 2017

27
Other new items for v1.3.1 Release
Meta RPM packages
introduced and adopted in
recipes:
• these replace previous use
of groups/patterns
• general convention
remains
- names that begin with
“ohpc-*” are typically
metapackges
- intended to group
related collections of
RPMs by functionality
• some names have been
updated for consistency
during the switch over
• updated list available in
Appendix E
E Package Manifest
This appendix provides a summary of available meta-package groupings and all of the individual RPM
packages that are available as part of this OpenHPC release. The meta-packages provide a mechanism to
group related collections of RPMs by functionality and provide a convenience mechanism for installation. A
list of the available meta-packages and a brief description is presented in Table 2.
Table 2: Available OpenHPC Meta-packages
Group Name Description
ohpc-autotools Collection of GNU autotools packages.
ohpc-base Collection of base packages.
ohpc-base-compute Collection of compute node base packages.
ohpc-ganglia Collection of Ganglia monitoring and metrics packages.
ohpc-gnu7-io-libs Collection of IO library builds for use with GNU compiler toolchain.
ohpc-gnu7-mpich-parallel-libs Collection of parallel library builds for use with GNU compiler toolchain and
the MPICH runtime.
ohpc-gnu7-mvapich2-parallel-libs Collection of parallel library builds for use with GNU compiler toolchain and
the MVAPICH2 runtime.
ohpc-gnu7-openmpi-parallel-libs Collection of parallel library builds for use with GNU compiler toolchain and
the OpenMPI runtime.
ohpc-gnu7-parallel-libs Collection of parallel library builds for use with GNU compiler toolchain.
ohpc-gnu7-perf-tools Collection of performance tool builds for use with GNU compiler toolchain.
ohpc-gnu7-python-libs Collection of python related library builds for use with GNU compiler
toolchain.
ohpc-gnu7-runtimes Collection of runtimes for use with GNU compiler toolchain.
ohpc-gnu7-serial-libs Collection of serial library builds for use with GNU compiler toolchain.
ohpc-intel-impi-parallel-libs Collection of parallel library builds for use with Intel(R) Parallel Studio XE
toolchain and the Intel(R) MPI Library.
ohpc-intel-io-libs Collection of IO library builds for use with Intel(R) Parallel Studio XE software
suite.
ohpc-intel-mpich-parallel-libs Collection of parallel library builds for use with Intel(R) Parallel Studio XE
toolchain and the MPICH runtime.
ohpc-intel-mvapich2-parallel-libs Collection of parallel library builds for use with Intel(R) Parallel Studio XE
toolchain and the MVAPICH2 runtime.
ohpc-intel-openmpi-parallel-libs Collection of parallel library builds for use with Intel(R) Parallel Studio XE
toolchain and the OpenMPI runtime.
ohpc-intel-perf-tools Collection of performance tool builds for use with Intel(R) Parallel Studio XE
toolchain.
ohpc-intel-python-libs Collection of python related library builds for use with Intel(R) Parallel Studio
XE toolchain.
ohpc-intel-runtimes Collection of runtimes for use with Intel(R) Parallel Studio XE toolchain.
ohpc-intel-serial-libs Collection of serial library builds for use with Intel(R) Parallel Studio XE
toolchain.
ohpc-nagios Collection of Nagios monitoring and metrics packages.
ohpc-slurm-client Collection of client packages for SLURM.
ohpc-slurm-server Collection of server packages for SLURM.
ohpc-warewulf Collection of base packages for Warewulf provisioning.
33 Rev: 208896e1
ohpc-gnu7-mvapich2-parallel-libs

28
Other new items for v1.3.1 Release
• A new compiler variant (gnu7) was introduced
- in the case of a fresh install, recipes default to installing the new variant along with
matching runtimes and libraries
- if upgrading a previously installed system, administrators can opt-in to enable
the gnu7 variant
• The meta-packages for “gnu7” provide a convenient mechanism to add on:
- upgrade discussion in recipes (Appendix B) amended to highlight this workflow
Install Guide (v1.3.1): CentOS7.3/x86 64 + xCAT + SLURM
# Update default environment
[sms]# yum -y remove lmod-defaults-gnu-mvapich2-ohpc
[sms]# yum -y install lmod-defaults-gnu7-mvapich2-ohpc
# Install GCC 7.x-compiled meta-packages with dependencies
[sms]# yum -y install ohpc-gnu7-perf-tools
ohpc-gnu7-serial-libs
ohpc-gnu7-io-libs
ohpc-gnu7-python-libs
ohpc-gnu7-runtimes
ohpc-gnu7-mpich-parallel-libs
ohpc-gnu7-openmpi-parallel-libs
ohpc-gnu7-mvapich2-parallel-libs
parallel libs for
gnu7/mpich
note: could skip this to leave
previous gnu toolchain as
default

29
Coexistence of multiple variants
$ module list
Currently Loaded Modules:
1) autotools 2) prun/1.1 3) gnu7/7.1.0 4) mvapich2/2.2 5) ohpc
$ module avail
--------------------------- /opt/ohpc/pub/moduledeps/gnu7-mvapich2 -----------------------------
adios/1.11.0 imb/4.1 netcdf-cxx/4.3.0 scalapack/2.0.2 sionlib/1.7.1
boost/1.63.0 mpiP/3.4.1 netcdf-fortran/4.4.4 scalasca/2.3.1 superlu_dist/4.2
fftw/3.3.6 mumps/5.1.1 petsc/3.7.6 scipy/0.19.0 tau/2.26.1
hypre/2.11.1 netcdf/4.4.1.1 phdf5/1.10.0 scorep/3.0 trilinos/12.10.1
------------------------------- /opt/ohpc/pub/moduledeps/gnu7 ----------------------------------
R_base/3.3.3 metis/5.1.0 numpy/1.12.1 openmpi/1.10.7
gsl/2.3 mpich/3.2 ocr/1.0.1 pdtoolkit/3.23
hdf5/1.10.0 mvapich2/2.2 (L) openblas/0.2.19 superlu/5.2.1
--------------------------------- /opt/ohpc/pub/modulefiles ------------------------------------
EasyBuild/3.2.1 gnu/5.4.0 ohpc (L) singularity/2.3
autotools (L) gnu7/7.1.0 (L) papi/5.5.1 valgrind/3.12.0
clustershell/1.7.3 hwloc/1.11.6 prun/1.1 (L)
previously
installed from
1.3 release
everything
else from
1.3.1
updates
(add-on)
Consider an example of system originally installed from 1.3 base
release and then added gnu7 variant using commands from last slide

Summary
• Technical Steering Committee just completed it’s first year of
operation; new membership selection for 2017-2018 in place
• Provided a highlight of changes/evolutions that have occurred since
MUG’16
- 4 releases since last MUG
- architecture addition (ARM)
- multiple provisioner/resource manager recipes
• “Getting Started” Tutorial held at PEARC’17 last month
- more in-depth overview: https://goo.gl/NyiDmr
30
http://openhpc.community (general info)
https://github.com/openhpc/ohpc (main GitHub site)
https://github.com/openhpc/submissions (new submissions)
https://build.openhpc.community (build system/repos)
http://www.openhpc.community/support/mail-lists/ (mailing lists)
Community Resources

OpenHPC: Project Overview and Updates

Recommended

More Related Content

What's hot (20)

Similar to OpenHPC: Project Overview and Updates (20)

More from inside-BigData.com (20)

Recently uploaded (20)

OpenHPC: Project Overview and Updates