HIGH-END COMPUTING CAPABILITY PORTFOLIO - William Thigpen NASA Advanced Supercomputing Division - NASA ...
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
HIGH-END COMPUTING
CAPABILITY PORTFOLIO
William Thigpen
NASA Advanced Supercomputing Division
July 10, 2021
National Aeronautics and Space Administration July 10, 2021 High-End Computing Capability Portfolio
Resource Management System Version 1.4.8 Released
• HECC, in collaboration with NCCS, released version 1.4.8 of the IMPACT: Developing in-house software to manage
Request Management System (RMS) on June 21, 2021. This tool supercomputer resource allocation requests allows
replaced REI eBooks in March and is being updated to add NASA’s High End Computing Program more ownership of
features. the data and simplifies the process for reviewing
allocations and targets.
• Version 1.4.8 concentrates on multi-year requests and allocations,
as well as preparing for the mission directorates’ specific needs.
• The new RMS tool enables new features such as:
– Easy navigation for requests from the home screen: Request more
resources, Request more time, Cancel request, or Print PDF.
– Allowing bulk allocation through Excel rather than allocating each
individual allocation by hand.
– Greater flexibility in customization and significant cost savings.
• On June 22, the new operational period (NOP) opened to all
mission directorates on RMS.
• On June 28, an RMS user information session was held and was
well-attended, with more than 55 participants.
• Completion of this milestone concluded the final phase of the multi- Banner announcement of the Resource Management System
phase version 1.x project. Work on RMS 2.0, which will enhance version 1.4.8 release. NASA
administrative abilities, started immediately after the release.
National Aeronautics and Space Administration July 10, 2021 High-End Computing Capability Portfolio 2
Applications Team Updates Key Performance Benchmark
• HECC’s Application Performance and Productivity (APP) team IMPACT: Benchmarks help maximize system productivity
updated their version of the mesoscale weather benchmark: by identifying performance issues on existing systems and
predicting the performance of architectures under
Weather Research and Forecasting (WRF) to version 3.9.1.1.
consideration for acquisition.
– WRF is an important component of HECC’s benchmark suite, which is
heavily used to measure system health and performance.
– It is also the best single representative of the entire SBU suite and is
therefore used extensively for system architecture evaluations,
including procurements.
• The previous WRF benchmark (v3.1) is more than 10 years old.
The new version provides bug fixes, compatibility with more recent
datasets, and backward compatibility with existing datasets.
– Both WRF v3.9.1.1 and WRF v4.2.2 were evaluated against the
Typhoon Morakot dataset in use since 2011.
– WRF v3.9.1.1 was found to be more compatible with the dataset.
Timings were found to fall within 5% of established wall times across
architectures.
– The APP team also improved validation testing by using the fldcor
function from Climate Data Operators (CDO) to measure the correlation HECC’s Application Performance and Productivity team uses an
between sample variables. established set of benchmarks to determine the health of existing
systems, such as the Pleiades supercomputer (pictured above), and
• APP will continue evaluating WRFv4, which will likely involve the to predict the performance of potential new systems by comparing
performance against reference statistics. NASA/Ames
creation of a new test with a different dataset.
National Aeronautics and Space Administration July 10, 2021 High-End Computing Capability Portfolio 3
HECC Contributes to Successful GPU Bootcamps
• HECC recently helped ensure the success of two virtual 6-hour GPU IMPACT: Easing the process of porting legacy codes for
bootcamps organized by Eric Nielsen and George Switzer from Langley use on accelerators will help ensure NASA’s effective use
Research Center and staff from NVIDIA. The bootcamps are part of of future high performance computing platforms.
preparations for the NASA GPU Hackathon 2021, which is scheduled for
September.
• More than 50 people attended the first event, which provided training on
writing CUDA-enabled HPC applications to individuals with no prior GPU
experience. More than 30 people attended the second event, which was
tailored to artificial intelligence and machine learning applications. Many of
the participants were new users who required NAS User Services assistance
for account setup on short notice.
• HECC reserved 21 V100 GPU nodes for use by participants and provided
technical support both before and during the events, including: devised and
provided step-by-step instructions detailing the use of the Singularity
container, Jupyter Lab, and SSH tunneling from users’ local machines to the
GPU nodes; tested NVIDIA’s Singularity containers and prepared them for
use by participants; and helped more than 20 participants by answering their
questions and/or setting up their environments. Despite being given A chart included in one of the sessions of the first GPU bootcamp,
highlighting the technique of splitting compute intensive functions
instructions to test logins and ssh tunneling in advance, many users waited and sequential code (top) and architectural differences (bottom)
until the time of the event to start setting up their accounts. between graphics processing units and central processing units.
Image courtesy of NVIDIA.
• Eric Nielsen was highly appreciative of the team’s help. He wrote: “HUGE
thanks to NAS for all of their support!”
National Aeronautics and Space Administration July 10, 2021 High-End Computing Capability Portfolio 4
HECC Completes COVID-19 Consortium Research Projects
• As part of the COVID-19 High-Performance Computing IMPACT: HECC resources and services enabled COVID-
Consortium, NASA provided access to HECC resources and 19 scientists to conduct rapid research in the fight to stop
the SARS-CoV-2 virus and find solutions that can help
expertise to support researchers in the fight to understand the
bring an end to the pandemic.
pandemic and to develop treatments and vaccines.
• Three research projects completed their research utilizing their
awarded supercomputing time on the HECC systems:
– MIT: “Drug-repurposing for COVID-19 with 3D-aware Machine
Learning.” The models produced can accurately screen unknown
compounds for SARS-CoV-2 inhibition. The work has led to 2 preprints,
a publicly available dataset, and a publicly available code repository.
– NASA Ames: “COVID-19 – RNA-seq Analysis to Identify Potential
Biomarkers Indicative of Disease Severity.” The researchers
successfully analyzed 845 COVID-19 patient samples and
environmental swab data to improve the understanding of SARS-CoV-2
interaction with the host. Results contributed to the collective scientific
knowledge of COVID-19 through several publications, manuscripts-in-
preparation, and presentations at scientific conferences including the
2020 supercomputing conference, SC20.
– Virginia Commonwealth University: “Modeling the Dynamic Behavior Simulation of the latanoprost molecule. A quantum chemistry
of Surface Spike Glycoprotein of COVID19 Coronavirus and Designing calculation was performed at each step, yielding the energy, charge
Biomimetic Therapeutic Compounds.” This work helped uncover density, and atomic forces. Warm colors signify charge concentration
dynamic and structural changes in spike protein variants that may lead and cool colors signify charge depletion. Chris Henze, NASA/Ames
to increased transmission rates. The project resulted in 2 publications.
National Aeronautics and Space Administration July 10, 2021 High-End Computing Capability Portfolio 5
HECC Support Teams Hold First Virtual User Meeting
• HECC completed the first of three virtual user meetings aimed at IMPACT: User outreach is critical to ensuring that
highlighting and promoting the project’s extensive resources and scientists and engineers make the most of their compute
resources and have an opportunity to give HECC staff a
services. Presentations covered the following support areas:
better understanding of their needs.
– Overview of the HECC Project – Accounts and Allocations
– User Services and Control Room – Supercomputers/Filesystems
– Cloud Offerings – Application Optimization
– Networks – Data Analysis/Visualization
– Data Analytics/Publication/Discovery – Publications & Media
• These virtual meetings take the place of regular in-person
outreach. Each meeting is targeted for a specific time zone (PDT,
CDT, and EDT), to ensure the maximum number of participants.
– The first meeting was very well attended, reaching 76 participants over
the course of the session.
– The remaining meetings will take place on July 7 and August 11.
• While not a perfect substitute for face-to-face meetings, this virtual
component allows HECC to reach a much wider user base and will Slide from the HECC Overview presentation. All slides will be made
likely be used to supplement future outreach activities when available on the HECC website. Bill Thigpen
regular travel resumes.
National Aeronautics and Space Administration July 10, 2021 High-End Computing Capability Portfolio 6
Modeling the Impact of Martian Dust on Entry Vehicles*
• The presence of suspended dust in the Martian atmosphere is a unique IMPACT: Results of this work can be used to determine
aspect of Mars atmospheric entry that current and future NASA missions the types of test data needed from future experiments and
to the Red Planet must address. Researchers from Stanford University guide the development of a simple, reliable physical model
used HECC resources to develop a computational methodology to for use in the design of future missions to Mars.
efficiently and accurately simulate supersonic and hypersonic particle-
laden flows over blunt bodies in order to better predict how entry vehicles
will interact with the Martian atmosphere.
• The team used discontinuous Galerkin (DG) methods to perform
simulations of hypersonic dusty flows and conducted a parametric study
to address the lack of experimental data needed for validation.
• They also developed a Lagrangian particle method in the DG framework
with algorithms to track particles on arbitrary elements, as well as to
handle particle-fluid coupling and particle-particle interactions in an
efficient and accurate manner.
• With this new methodology, the Stanford team found that their predictions
of dust-induced surface heat flux augmentation agreed well with
experimental tests, demonstrating the applicability of the computational Visualization of a hypersonic dusty flow over a capsule forebody.
models to accurately simulate the impact of suspended dust particles on The flow field is colored by the temperature of the gas (blue
thermal protection systems during atmospheric entry. indicates low and red indicates high). The circles, which represent
particles, are scaled in size by particle diameter and colored by
streamwise velocity. Eric Ching, Matthias Ihme, Stanford University
* HECC provided supercomputing resources and services in support of this work.
National Aeronautics and Space Administration July 10, 2021 High-End Computing Capability Portfolio 7
Exploring the “Dark Genome” Response to SARS-CoV-2*
• A team of genetic researchers from NASA Ames, universities, and IMPACT: This work was part of the COVID-19 HPC
laboratories used Pleiades to perform transcriptomic (RNA) Consortium, which provided high-performance computing
access to researchers around the world studying the
analysis of next-generation sequencing data from patients tested
global transmission and impact of the COVID-19 virus.
for SARS-CoV-2 to gain a better understanding of how the virus
hijacks cells’ metabolic pathways to facilitate its own replication.
– Human viruses may accomplish this by selectively modifying the host’s
genomic output via interactions with genomic regulators such as non-
coding RNA molecules (ncRNAs, also known as the “dark genome”).
• Results showed that SARS-CoV-2 infection induces a
transcriptional shift—a shift in the genomic output—of the host
cells, mainly characterized by an up-regulation of RNA transcripts
that includes a specific group of long ncRNAs (lncRNAs).
– The team postulated that lncRNAs appearing in higher amounts in
patients infected with SARS-CoV-2—versus patients without
SARS-CoV-2—will, in response to the infection, establish interaction
networks with other cellular molecules such as RNA-binding proteins.
• Further characterization of these interactions may open a new
Host transcriptomic analysis in SARS-CoV-2 infected samples. The
window for the development of therapeutic strategies that could chart shows Principal Component Analysis (PCA) of differentially
target lncRNA function in SARS-CoV-2 and other viral infections. expressed long non-coding RNAs (lncRNAs) in the infected
samples. Francisco J. Enguita, University of Lisbon, Portugal
* HECC provided supercomputing resources and services in support of this work
National Aeronautics and Space Administration July 10, 2021 High-End Computing Capability Portfolio 8Papers
• “Core Formation in High-z Massive Haloes: Heating by Post Compaction Satellites and Response to AGN Outflows,”
A. Dekel, et al., arXiv:2106.01378 [astro-ph.GA], June 2, 2021. *
https://arxiv.org/abs/2106.01378
• “Molecular Dynamics Simulations of Ultrafast Radiation Induced Melting at Metal-Semiconductor Interfaces,”
A. Ravichandran, et al., Journal of Applied Physics, vol .129, issue 21, published online June 2, 2021. *
https://aip.scitation.org/doi/abs/10.1063/5.0045766
• “One Year in the Life of Young Suns: Data Constrained Corona-Wind Model of kappa1 Ceti,” V. Airapetian, et al.,
arXiv:2106.01284 [astro-ph.SR], June 2, 2021. *
https://arxiv.org/abs/2106.01284
• “An Automated Bolide Detection Pipeline for GOES GLM,” J. Smith, R. Morris, C. Rumpf, R. Logenbaugh, N. McCurdy, C. Henze,
J. Dotson, Icarus, vol. 368, June 4, 2021. *
https://www.sciencedirect.com/science/article/pii/S0019103521002451
• “The Interaction Between Rising Bubbles and Cold Fronts in Cool Core Clusters,” A. C. Fabian, J. Zuhone, S. A. Walker,
arXiv:2106.03663 [astro-ph.GA], June 7, 2021. *
https://arxiv.org/abs/2106.03662
• “Planet Hunters TESS III: Two Transiting Planets around the Bright G Dwarf HD 1528443,” N. Eisner, et al., arXiv:2106.04603
[astro-ph.EP], June 8, 2021. *
https://arxiv.org/abs/2106.04603
* HECC provided supercomputing resources and services in support of this work
National Aeronautics and Space Administration July 10, 2021 High-End Computing Capability Portfolio 9Papers (cont.)
• “Dynamical Effects of Cosmic Rays on the Medium Surrounding Their Sources,” B. Schroer, et al., The Astrophysical Journal
Letters, vol. 914, no. 1, June 9, 2021. *
https://iopscience.iop.org/article/10.3847/2041-8213/ac02cd/meta
• “From Electrons to Janskys: Full Stokes Polarized Radiative Transfer in 3D Relativistic Particle-in-Cell Jet Simulations,”
N. MacDonald, K.-I. Nishikawa, arXiv:2106.04915 [astro-ph.HE], June 9, 2021. *
https://arxiv.org/abs/2106.04915
• “A Perspective on Collective Properties of Atoms on 2D Materials,” A. Del Maestro, et al., arXiv:210.07685 [cond-mat.mes-hall],
June 14, 2021. *
https://arxiv.org/abs/2106.07685
• “Lepton-Driven Non-Resonant Streaming Instability,” S. Gupta, et al., arXiv:2106.07672 [astro-ph.HE], June 14, 2021. *
https://arxiv.org/abs/2106.07672
• “TESS Data for Astroseismology: Photometry,” R. Handberg, et al., arXiv:2106.08341 [astro-ph.IM], June 15, 2021. *
https://arxiv.org/abs/2106.08341
• “Atomic-Scale Evidence for Open-System Thermodynamics in the Early Solar Nebula,” T. Zega, et al., The Planetary Science
Journal, vol. 2, no. 3, June 17, 2021. *
https://iopscience.iop.org/article/10.3847/PSJ/abf5e5/meta
* HECC provided supercomputing resources and services in support of this work
National Aeronautics and Space Administration July 10, 2021 High-End Computing Capability Portfolio 10Papers (cont.)
• “Efficiency of Mixed-Element USM3D for Benchmark Three-Dimensional Flows,” M. Pandya, D. Jespersen, B. Diskin,
J. Thomas, N. Frink, AIAA Journal, published online June 21, 2021. *
https://arc.aiaa.org/doi/full/10.2514/1.J059720
* HECC provided supercomputing resources and services in support of this work
National Aeronautics and Space Administration July 10, 2021 High-End Computing Capability Portfolio 11News and Events
• Which Way Does the Solar Wind Blow? Texas Advanced Computing Center, June 3, 2021—Using supercomputers at
NAS, as well as the Texas Advanced Computing Center and the San Diego Supercomputing Center, researchers are
improving models and methods to predict space weather, including the arrival of coronal mass ejections.
https://www.tacc.utexas.edu/-/which-way-does-the-solar-wind-blow-
• Which Way Does the Solar Wind Blow? Supercomputers Improve Space Weather Prediction, SciTech Daily, June 6, 2021.
https://scitechdaily.com/which-way-does-the-solar-wind-blow-supercomputers-improve-space-weather-prediction/
• Space Weather Prediction Gets a Supercomputing Boost, HPCwire, June 9, 2021.
https://www.hpcwire.com/2021/06/09/space-weather-prediction-gets-a-supercomputing-boost/
• UAH-Led Space Weather Prediction Research Could be Critical to U.S. Space Command, University of Alabama in Huntsville,
June 24, 2021.
https://www.uah.edu/news/items/uah-led-space-weather-prediction-research-could-be-critical-to-space-force-command
• NASA Partners with Khaled bin Sultan Living Oceans Foundation to Expand Efforts to Map Corals, NASA Ames,
June 8, 2021—NASA’s Ames Research Center is partnering with the Khaled bin Sultan Living Oceans Foundation to use
their extensive high-resolution data about reefs and expand NASA’s coral mapping capabilities even further. The
partnership will use the massive Global Reef Expedition dataset in conjunction with the neural network and the Pleiades
supercomputer at the NAS facility.
https://www.nasa.gov/image-feature/ames/nasa-expands-efforts-to-map-corals
• Mapping the World’s Coral Reefs, World Ocean Forum, June 8, 2021.
https://medium.com/world-ocean-forum/mapping-of-the-worlds-coral-reefs-e29ba653a049
National Aeronautics and Space Administration July 10, 2021 High-End Computing Capability Portfolio 12News and Events: Social Media
• Coverage of NAS Stories
• Aitken compute hours milestone:
• NAS: Twitter 2 retweets, 6 favorites.
• NASA Supercomputing: Facebook 639 users reached, 101 engagements, 7 likes, 4 shares.
• NAS Software Available to Public (part of NASA’s updated software catalog campaign):
• NAS: Twitter 5 retweets, 31 favorites.
• NASA Supercomputing: Twitter 3 retweets, 7 favorites; Facebook 236 users reached, 8 engagements, 4 likes.
• Asteroid Threat Assessment Project (Asteroid Day):
• NASA Ames: Twitter 28 retweets, 102 favorites; Facebook 40 likes, 17 comments, 15 shares.
• NASA Supercomputing: Facebook 62 users reached, 11 engagements, 5 likes.
National Aeronautics and Space Administration July 10, 2021 High-End Computing Capability Portfolio 13HECC Utilization
100%
Furlough
Share Limit
90%
Job Drain
80% Dedtime Drain
Unused Devel Queue
70%
Insufficient CPUs
Held
60%
Queue Not Schedulable
50% Not Schedulable
No Jobs
40% Dedicated
Down
30%
Degraded
Boot
20%
Free/Testing
10% Used
0%
Aitken Pleaides Electra Endeavour Production
June 2021
National Aeronautics and Space Administration July 10, 2021 High-End Computing Capability Portfolio 14HECC Utilization Normalized to 30-Day Month
13,000,000
12,000,000
11,000,000
10,000,000 MISC
9,000,000
Standard Billing Units
NAS
8,000,000
NLCS
7,000,000
NESC
6,000,000
SMD
5,000,000
HEOMD
4,000,000
3,000,000 ARMD
2,000,000 Alloc. to Orgs
1,000,000
0
Au 9
Au 0
Se 9
O 9
De 9
Ja 9
Se 0
O 0
De 0
Ja 0
Ju 0
Ju 1
Fe 0
Ju 0
Fe 1
21
M 0
Ap 0
M 1
Ap 1
No 9
M 20
No 0
M 21
2
-2
2
-2
-1
-2
l-1
l-2
1
1
1
1
2
2
2
2
-2
-2
2
2
2
r-
r-
g-
p-
v-
c-
g-
p-
v-
c-
n-
n-
n-
n-
b-
b-
ct
ct
ar
ar
ay
ay
Ju
National Aeronautics and Space Administration July 10, 2021 High-End Computing Capability Portfolio 15HECC Utilization Normalized to 30-Day Month
7,000,000 6,000,000
SMD 3 5 ARMD
6,000,000 5,000,000
Standard Billing Units 4
Standard Billing Units
5,000,000
2 4,000,000
1
4,000,000
3,000,000
345
3,000,000 1 2
2,000,000
2,000,000
1,000,000 1,000,000
0
ov 9 0
ay 0
ov 0
ay 1
Fe -20
Ju 20
Fe -21
21
M 20
Ap -2 0
M 21
Ap -2 1
Au l-19
Au l-20
ov 9
ay 0
ov 0
ay 1
Se -19
O -19
D -19
Ja -19
Se -20
O -20
D -20
Ja -20
Ju -20
Ju -21
Fe -20
Ju 20
Fe -21
21
M 20
Ap -2 0
M 21
Ap -2 1
Ju -20
Ju -21
Au l-19
Au l-20
Se -19
O -19
D -19
Ja -19
Se -20
O -20
D -20
Ja -20
N -1
M r-2
N -2
M r-2
N -1
M r-2
N -2
M r-2
n-
n-
b-
b-
n-
n-
b-
b-
g
p
g
p
ec
ar
ec
ar
ct
ct
n
n
g
p
g
p
ec
ar
ec
ar
ct
ct
n
n
Ju
Ju
SMD SMD Allocation With Agency Reserve ARMD ARMD Allocation With Agency Reserve
6,000,000
HEOMD, NESC
5,000,000
1 Skylake Tesla GPU V100 Nodes installed
Standard Billing Units
4,000,000
34
3,000,000 5 2 4 Cascade Lake E cells introduced into Aitken
1 2
2,000,000
3 8 Rome Apollo racks introduced into Aitken
1,000,000
0 4 Endeavour replaced with new hardware
N -1 9
ay 0
N -2 0
ay 1
Fe -20
Ju 20
Fe -21
21
M 20
Ap -2 0
M 21
Ap -2 1
Au l-19
Au l-20
Se -19
O -19
D -19
Ja -19
Se -20
O -20
D -20
Ja -20
Ju -20
Ju -21
M r-2
M r-2
n-
n-
b-
b-
g
p
g
p
ov
ec
ar
ov
ec
ar
ct
ct
n
n
5 Merope Retired
Ju
HEOMD NESC HEOMD+NES C Allocatio n
National Aeronautics and Space Administration July 10, 2021 High-End Computing Capability Portfolio 16Tape Archive Status
1200
1100
Capacity
1000
Used
900
HECC
800
Non Mission
700 Specific
Peta Bytes
NAS
600
NLCS
500
NESC
400
SMD
300
HEOMD
200
ARMD
100
0
Unique File Data Unique Tape Data Total Tape Data Tape Capacity Tape Library Capacity
June 2021
National Aeronautics and Space Administration July 10, 2021 High-End Computing Capability Portfolio 17Tape Archive Status
1200
1100
1000
900
800
Tape Library Capacity
700
Peta Bytes
Tape Capacity
600
Total Tape Data
500 Unique Tape Data
400
300
200
100
0
Au 9
Au 0
Se 9
O 9
De 9
Ja 9
Se 0
O 0
De 0
Ja 0
Ju 0
Ju 1
Fe 0
20
Fe 1
21
M 0
Ap 0
M 1
Ap 1
No 9
M 20
No 0
M 21
2
-2
2
-2
-1
-2
l-1
l-2
1
1
1
1
2
2
2
2
-2
-2
2
2
r-
r-
g-
p-
v-
c-
g-
p-
v-
c-
n-
n-
n-
n-
b-
b-
ct
ct
ar
ar
ay
ay
Ju
Ju
National Aeronautics and Space Administration July 10, 2021 High-End Computing Capability Portfolio 18Pleiades: SBUs Reported, Normalized to 30-Day Month
6,000,000
MISC
5,000,000
NAS
NLCS
Standard Billing Units
4,000,000
NESC
3,000,000
SMD
2,000,000
HEOMD
ARMD
1,000,000
Alloc. to Orgs
0
0
20
20
0
0
1
21
21
21
1
0
1
-2
-2
r -2
l-2
-2
-2
-2
g-
p-
n-
n-
b-
ct
v
c
ar
ay
Ju
Ap
Au
Se
No
De
Ja
Ju
Fe
O
M
M
National Aeronautics and Space Administration July 10, 2021 High-End Computing Capability Portfolio 19Pleiades: Devel Queue Utilization
800,000
MISC
700,000
NAS
600,000
Standard Billing Units
NLCS
500,000
NESC
400,000
SMD
300,000
HEOMD
200,000
100,000 ARMD
0 Devel Queue
Alloc.
0
20
20
0
0
1
21
21
21
1
0
1
-2
-2
r -2
l-2
-2
-2
-2
g-
p-
n-
n-
b-
ct
v
c
ar
ay
Ju
Ap
Au
Se
No
De
Ja
Ju
Fe
O
M
M
National Aeronautics and Space Administration July 10, 2021 High-End Computing Capability Portfolio 20Pleiades: Monthly Utilization by Job Length
800,000
700,000
600,000
Standard Billing Units
500,000
400,000
300,000
200,000
100,000
0
0-1 >1 -4 >4 -8 > 8 - 24 > 24 - 48 > 48 - 72 > 72 - 96 > 96 - 120 > 120
Job Run Time (hours)
June 2021
National Aeronautics and Space Administration July 10, 2021 High-End Computing Capability Portfolio 21Pleiades: Monthly Utilization by Job Size
1,500,000
NAS
1,250,000
NLCS
Standard Billing Units
1,000,000
NESC
750,000 SMD
HEOMD
500,000
ARMD
250,000
0
1 - 32 33 - 64 65 - 129 - 257 - 513 - 1025 - 2049 - 4097 - 8193 - 16385 - 32769 - 65537 -
128 256 512 1024 2048 4096 8192 16384 32768 65536 262144
Job Size (cores) June 2021
National Aeronautics and Space Administration July 10, 2021 High-End Computing Capability Portfolio 22Pleiades: Monthly Utilization by Size and Length
1,500,000
> 120 hours
1,250,000
> 96 - 120 hours
> 72 - 96 hours
Standard Billing Units
1,000,000
> 48 - 72 hours
750,000 > 24 - 48 hours
> 8 - 24 hours
500,000
> 4 - 8 hours
250,000 > 1 - 4 hours
0 - 1 hours
0
1 - 32 33 - 64 65 - 129 - 257 - 513 - 1025 - 2049 - 4097 - 8193 - 16385 - 32769 - 65537-
128 256 512 1024 2048 4096 8192 16384 32768 65536 262144
Job Size (cores)
June 2021
National Aeronautics and Space Administration July 10, 2021 High-End Computing Capability Portfolio 23Pleiades: Average Time to Clear All Jobs
1,080
1351
960
840
720
ARMD
Hours
600
HEOMD
SMD
480
360
240
120
0
Jul-20 Aug-20 Sep-20 Oct-20 Nov-20 Dec-20 Jan-21 Feb-21 Mar-21 Apr-21 May-21 Jun-21
National Aeronautics and Space Administration July 10, 2021 High-End Computing Capability Portfolio 24Pleiades: Average Expansion Factor
10.00
9.00
8.00
7.00
Expansion Factor
6.00
ARMD
HEOMD
5.00
SMD
4.00
3.00
2.00
1.00
Jul-20 Aug-20 Sep-20 Oct-20 Nov-20 Dec-20 Jan-21 Feb-21 Mar-21 Apr-21 May-21 Jun-21
National Aeronautics and Space Administration July 10, 2021 High-End Computing Capability Portfolio 25Aitken: SBUs Reported, Normalized to 30-Day Month
5,000,000
MISC
4,500,000
4,000,000 NAS
3,500,000
NLCS
Standard Billing Units
3,000,000
NESC
2,500,000
SMD
2,000,000
1,500,000 HEOMD
1,000,000 ARMD
500,000
Alloc. to Orgs
0
0
20
20
0
0
1
21
21
21
1
0
1
-2
-2
r -2
l-2
-2
-2
-2
g-
p-
n-
n-
b-
ct
v
c
ar
ay
Ju
Ap
Au
Se
No
De
Ja
Ju
Fe
O
M
M
National Aeronautics and Space Administration July 10, 2021 High-End Computing Capability Portfolio 26Aitken: Devel Queue Utilization
175,000
MISC
150,000
NAS
125,000
NLCS
Standard Billing Units
100,000 NESC
SMD
75,000
HEOMD
50,000
ARMD
25,000
Devel Queue
Allocation
0
0
20
20
0
0
1
21
21
21
1
0
1
-2
-2
r -2
l-2
-2
-2
-2
g-
p-
n-
n-
b-
ct
v
c
ar
ay
Ju
Ap
Au
Se
No
De
Ja
Ju
Fe
O
M
M
National Aeronautics and Space Administration July 10, 2021 High-End Computing Capability Portfolio 27Aitken: Monthly Utilization by Job Length
1,200,000
1,000,000
Standard Billing Units
800,000
600,000
400,000
200,000
0
0-1 >1 -4 >4 -8 > 8 - 24 > 24 - 48 > 48 - 72 > 72 - 96 > 96 - 120 > 120
Job Run Time (hours)
June 2021
National Aeronautics and Space Administration July 10, 2021 High-End Computing Capability Portfolio 28Aitken: Monthly Utilization by Job Size
1,500,000
NAS
1,250,000
NLCS
Standard Billing Units
1,000,000
NESC
750,000 SMD
HEOMD
500,000
ARMD
250,000
0
1 - 32 33 - 64 65 - 129 - 257 - 513 - 1025 - 2049 - 4097 - 8193 - 16385 - 32769 - 65537 -
128 256 512 1024 2048 4096 8192 16384 32768 65536 262144
Job Size (cores) June 2021
National Aeronautics and Space Administration July 10, 2021 High-End Computing Capability Portfolio 29Aitken: Monthly Utilization by Size and Length
1,500,000
> 120 hours
1,250,000
> 96 - 120 hours
> 72 - 96 hours
Standard Billing Units
1,000,000
> 48 - 72 hours
750,000 > 24 - 48 hours
> 8 - 24 hours
500,000
> 4 - 8 hours
250,000 > 1 - 4 hours
0 - 1 hours
0
1 - 32 33 - 64 65 - 129 - 257 - 513 - 1025 - 2049 - 4097 - 8193 - 16385 - 32769 - 65537-
128 256 512 1024 2048 4096 8192 16384 32768 65536 262144
Job Size (cores)
June 2021
National Aeronautics and Space Administration July 10, 2021 High-End Computing Capability Portfolio 30Aitken: Average Time to Clear All Jobs
1,200
1,080
960
840
720
Hours
ARMD
600 HEOMD
SMD
480
360
240
120
0
Jul-20 Aug-20 Sep-20 Oct-20 Nov-20 Dec-20 Jan-21 Feb-21 Mar-21 Apr-21 May-21 Jun-21
National Aeronautics and Space Administration July 10, 2021 High-End Computing Capability Portfolio 31Aitken: Average Expansion Factor
8.00
7.00
6.00
Expansion Factor
5.00 ARMD
HEOMD
4.00 SMD
3.00
2.00
1.00
Jul-20 Aug-20 Sep-20 Oct-20 Nov-20 Dec-20 Jan-21 Feb-21 Mar-21 Apr-21 May-21 Jun-21
National Aeronautics and Space Administration July 10, 2021 High-End Computing Capability Portfolio 32Electra: SBUs Reported, Normalized to 30-Day Month
4,500,000
MISC
4,000,000
NAS
3,500,000
NLCS
Standard Billing Units
3,000,000
2,500,000 NESC
2,000,000 SMD
1,500,000
HEOMD
1,000,000
ARMD
500,000
Alloc. to Orgs
0
0
20
20
0
0
1
21
21
21
1
0
1
-2
-2
r -2
l-2
-2
-2
-2
g-
p-
n-
n-
b-
ct
v
c
ar
ay
Ju
Ap
Au
Se
No
De
Ja
Ju
Fe
O
M
M
National Aeronautics and Space Administration July 10, 2021 High-End Computing Capability Portfolio 33Electra: Devel Queue Utilization
500,000
450,000 NAS
400,000
NLCS
350,000
Standard Billing Units
NESC
300,000
250,000
SMD
200,000
HEOMD
150,000
100,000 ARMD
50,000
Devel Queue
Allocation
0
0
20
20
0
0
1
21
21
21
1
0
1
-2
-2
r -2
l-2
-2
-2
-2
g-
p-
n-
n-
b-
ct
v
c
ar
ay
Ju
Ap
Au
Se
No
De
Ja
Ju
Fe
O
M
M
National Aeronautics and Space Administration July 10, 2021 High-End Computing Capability Portfolio 34Electra: Monthly Utilization by Job Length
1,000,000
750,000
Standard Billing Units
500,000
250,000
0
0-1 >1 -4 >4 -8 > 8 - 24 > 24 - 48 > 48 - 72 > 72 - 96 > 96 - 120 > 120
Job Run Time (hours)
June 2021
National Aeronautics and Space Administration July 10, 2021 High-End Computing Capability Portfolio 35Electra: Monthly Utilization by Job Size
1,000,000
NAS
750,000 NLCS
Standard Billing Units
NESC
500,000
SMD
250,000 HEOMD
ARMD
0
1 - 32 33 - 64 65 - 129 - 257 - 513 - 1025 - 2049 - 4097 - 8193 - 16385 - 32769 - 65537 -
128 256 512 1024 2048 4096 8192 16384 32768 65536 131072
Job Size (cores) June 2021
National Aeronautics and Space Administration July 10, 2021 High-End Computing Capability Portfolio 36Electra: Monthly Utilization by Size and Length
1,000,000
> 120 hours
> 96 - 120 hours
750,000
> 72 - 96 hours
Standard Billing Units
> 48 - 72 hours
500,000 > 24 - 48 hours
> 8 - 24 hours
> 4 - 8 hours
250,000
> 1 - 4 hours
0 - 1 hours
0
1 - 32 33 - 64 65 - 129 - 257 - 513 - 1025 - 2049 - 4097 - 8193 - 16385 - 32769 - 65537 -
128 256 512 1024 2048 4096 8192 16384 32768 65536 131072
Job Size (cores) June 2021
National Aeronautics and Space Administration July 10, 2021 High-End Computing Capability Portfolio 37Electra: Average Time to Clear All Jobs
1,080
1985 1113
960
840
720
Hours
600 ARMD
HEOMD
480 SMD
360
240
120
0
Jul-20 Aug-20 Sep-20 Oct-20 Nov-20 Dec-20 Jan-21 Feb-21 Mar-21 Apr-21 May-21 Jun-21
National Aeronautics and Space Administration July 10, 2021 High-End Computing Capability Portfolio 38Electra: Average Expansion Factor
7.00
6.00
5.00
Expansion Factor
ARMD
4.00
HEOMD
SMD
3.00
2.00
1.00
Jul-20 Aug-20 Sep-20 Oct-20 Nov-20 Dec-20 Jan-21 Feb-21 Mar-21 Apr-21 May-21 Jun-21
National Aeronautics and Space Administration July 10, 2021 High-End Computing Capability Portfolio 39Endeavour: SBUs Reported, Normalized to 30-Day Month
60,000
55,000 NAS
50,000
NLCS
45,000
Standard Billing Units
40,000
NESC
35,000
30,000 SMD
25,000
HEOMD
20,000
15,000
ARMD
10,000
5,000 Alloc. to Orgs
0
0
20
20
0
0
1
21
21
21
1
0
1
-2
-2
r -2
l-2
-2
-2
-2
g-
p-
n-
n-
b-
ct
v
c
ar
ay
Ju
Ap
Au
Se
No
De
Ja
Ju
Fe
O
M
M
National Aeronautics and Space Administration July 10, 2021 High-End Computing Capability Portfolio 40Endeavour: Monthly Utilization by Job Length
10,000
9,000
8,000
7,000
Standard Billing Units
6,000
5,000
4,000
3,000
2,000
1,000
0
0-1 >1 -4 >4 -8 > 8 - 24 > 24 - 48 > 48 - 72 > 72 - 96 > 96 - 120 > 120
Job Run Time (hours) June 2021
National Aeronautics and Space Administration July 10, 2021 High-End Computing Capability Portfolio 41Endeavour: Monthly Utilization by Job Size
20,000
18,000 NAS
16,000
NESC
Standard Billing Units
14,000
12,000
SMD
10,000
8,000 HEOMD
6,000
ARMD
4,000
2,000
0
1 - 32 33 - 64 65 - 128 129 - 256 257 - 512 513 - 1024
Job Size (cores)
June 2021
National Aeronautics and Space Administration July 10, 2021 High-End Computing Capability Portfolio 42Endeavour: Monthly Utilization by Size and Length
20,000
> 120 hours
18,000
16,000 > 96 - 120 hours
14,000 > 72 - 96 hours
Standard Billing Units
12,000 > 48 - 72 hours
10,000 > 24 - 48 hours
8,000 > 8 - 24 hours
6,000
> 4 - 8 hours
4,000
> 1 - 4 hours
2,000
0 - 1 hours
0
1 - 32 33 - 64 65 - 128 129 - 256 257 - 512 513 - 1024
Job Size (cores)
June 2021
National Aeronautics and Space Administration July 10, 2021 High-End Computing Capability Portfolio 43Endeavour: Average Time to Clear All Jobs
144
120
96
Hours
ARMD
72 HEOMD
SMD
48
24
0
Jul-20 Aug-20 Sep-20 Oct-20 Nov-20 Dec-20 Jan-21 Feb-21 Mar-21 Apr-21 May-21 Jun-21
National Aeronautics and Space Administration July 10, 2021 High-End Computing Capability Portfolio 44Endeavour: Average Expansion Factor
13.00
20.86
11.00
9.00
Expansion Factor
ARMD
7.00
HEOMD
SMD
5.00
3.00
1.00
Jul-20 Aug-20 Sep-20 Oct-20 Nov-20 Dec-20 Jan-21 Feb-21 Mar-21 Apr-21 May-21 Jun-21
National Aeronautics and Space Administration July 10, 2021 High-End Computing Capability Portfolio 45You can also read
