1. The document describes an HPC experiment that aims to make HPC resources available on demand for small and medium enterprises (SMEs) and their engineering applications.
2. It details how the experiment works by matching SMEs with software/hardware providers and experts to test running jobs on remote HPC clouds.
3. One example team tested running Abaqus simulations on the SGI Cyclone supercomputer and remotely visualizing results using NICE DCV software to assess the feasibility of using cloud HPC resources.
2. The UberCloud HPC Experiment
An open, voluntary, collaborative community
Objective:
Making HPC as a Service available, for everybody, on demand
How?
For SMEs and their engineering applications
we provide the opportunity to explore
the end-to-end process of using remote computing resources,
as a service, on demand, at your finger tip,
and learning how to resolve the roadblocks.
3. What is the HPC Experiment?
Started in mid-2012 as a voluntary effort
Demonstrate the potential of HPC in the Cloud
Uncover and overcome the obstacles
Now with almost 500 participants Round 3 has concluded
Approaching 95 teams as of today
Please submit your project ideas when you register at
http://www.hpcexperiment.com
3
4. WhyThis Experiment?
Foster use of HPC in Digital Manufacturing
Focus on: remote resources in HPC Centers & in HPC Clouds
Support initiatives from Intel, NCMS, and many others to
uncover and support the ‘missing middle’ (SMEs)
Observation: business clouds are becoming widely accepted,
but acceptance of simulation clouds in industry is still in early
adopter stage (CAE, Bio, Finance,Oil & Gas, …)
Some Barriers: Complexity, IP, data transfer, software
licenses, performance, specific system requirements, data
security, interoperability, cost, etc.
4
5. How the Experiment works…
End-User joins the experiment
ISV joins
We select a Team Expert
We suggest a Resource Provider
Team is ready to go
… 25 steps on Basecamp virtual team office
Finally, writing the Case Study
6. Our recipe
Find the best matches and form the team
Step by step process
Collaborative tools
Engaged team participants
Experienced mentors
Exhibit – list of additional services to pick from
Community – get help when needed
7. Step by Step process
Basecamp project management platform for each team
8. Example: UberCloud Round 4
August – November 2013
Expecting 600 organizations, ready for 25 more teams
Extending applications: HPC, CAE, Life Sciences, Big Data
Guidance: 22-steps end-to-end process
Basecamp ‘team rooms’ for collaboration
Growing the UberCloud Exhibit services directory
3-level support: front line (within a team), 2nd level (UberCloud
Mentors), 3rd level (software & hardware providers)
The UberCloud Community Platform, an internal HPC Services
website with crowd-sourcing and social networking elements
9. How to participate
End-user: submit your project idea through info gathering
form, select a special service from the Exhibit
Provider (software, resource, expertise): invite one of your
customers, users, clients to register at the HPC Experiment
and jointly form a team
Provider: also, consider joining the Exhibit to expose your
service to the UberCloud community
HPC Expert: just provide your profile describing your
expertise so we can match you to a team which needs this
expertise
10. Example: HPC Experiment-Team 26
Simulating Stent Deployment using SIMULIA’s
Abaqus/Standard and RemoteViz Software from NICE to
run Abaqus/CAE on SGI Cyclone™
Assessment of a fictitious balloon expandable stent design from the
standpoint of device response to loading onto a delivery system,
deployment, physiological pulsatile loading and both axial and radial
compression.
11. Team 26- Members
End User: Anonymous. Global Designer and Manufacturer of Sterile
Medical Products
CAE Software Provider: Matt Dunbar, Chief Architect, SIMULIA
RemoteViz Software: NICE Desktop CloudVisualization (DCV)
HPC/CAE Expert: Scott Shaw, Senior Applications Engineer, SGI.
Resource Provider: SGI Cyclone. Tony DeVarco, Senior Manager for
Strategic Partners and Cloud Computing at SGI. Eugene Kremenetsky,
Systems EngineeringTechnical Lead at SGI
Team Mentor: Gregory Shirin from the HPC Experiment team
12. Transfer Only Pixels to Visualize
What questions were being asked?
User
Move Data; Submit; Run Job
1. Is running Abaqus production jobs in a HPC cloud feasible?
2. How hard is it to set up “cloud bursting” capability when local
resources are not available?
3. How well does remote visualization tools work?
Open up a viz
session
So what was to be tested by the End User in SGI Cyclone?
Transfer Data
When Finished
13. WhatWas Offered
For the experiment it was agreed the Simulia and SGI
would provide the end user with Abaqus licenses for up
to 128 cores in order to see if running a job on more
cores could reduce the time to finish the job.
Access to NICE DCV remote graphics software to view
the results with Abaqus/CAE in Northern California
before downloading them to the End User office on the
US East Coast.
15. Challenges, Barriers, Lessons Learned
Getting customer’s IT to open up firewall ports
Data transfer via the network was found to be slow. Final results
might be better transferred through an external USB hard drive via
FedEx.
Data security and privacy – assurance of the best possible data
stewardship is critical from the standpoint of protecting the
customer’s intellectual property.
For an Abaqus user using SGI Cyclone is a viable solution for both
compute and visualization.
The remote visualization aspect of the solution was impressive
16. Top benefits for the End User
1. Gained increased understanding of using a cloud-based solution
for doing Abaqus based FEA simulations.
2. Shifting computational work to the cloud during periods of full
utilization of in-house compute resources is a viable approach to
ensuring analysis throughput.
3. Participation in the experiment allowed direct assessment of speed
and integrity of remote viz of computational models (both pre-
and post-processing) for a variety of model and output database
sizes.
4. SGI/Nice DCV provides robust solution permitting fast and
accurate remote manipulation of the computational models used
in the study.
17. Compendium of case studies
25
of our case studies
are available for download:
http://tci.taborcommunications.com/UberCloud_HPC_Experiment
18. Thank you for your interest
For more questions and answers please see our
Q & A Session at:
http://www.hpcexperiment.com/q-and-a
And, if you are ready for registering:
http://www.hpcexperiment.com/