Continued HPC Visualization Advancements via Intel® Rendering Framework

At SC18 in Dallas, Intel reaffirmed its commitment to creating solutions for some of the most demanding visual workloads. Our efforts have been building very strategically over the last few months as we continue to invest in a portfolio of open source rendering libraries to enable a number of industries, from Hollywood studios, world-renowned universities, and supercomputing data centers to deliver optimum visual impact and discovery.

As Intel’s Corporate Vice President and General Manager of the Open Source Technology Center Imad Sousou wrote during IBC 2018 in Amsterdam, Intel recently established the Intel® Rendering Framework as the umbrella name for Intel’s open source visualization libraries, including:

These libraries provide a strong foundation and catalyst for the Software Defined Visualization (SDVis) industry initiative Intel has collaborated on with institutions like Texas Advanced Computing Center (TACC), Kitware, Inc. (developers of ParaView and VTK), SCI at University of Utah, VISUS at University of Stuttgart, University of Tennessee, University of Oregon, and Stephen Hawking Centre for Theoretical Cosmology at University of Cambridge.

Also this year, Intel was proud to join the Academy Software Foundation, as I announced at SIGGRAPH 2018 in Vancouver, where we showed how Intel® technology was used in the rendering the Moana island scene that Disney Studios recently made publicly  available.* The 160+ billion object scene required over 100GB of system memory to create. Disney has publicly released the data set for others to research and establish best practices. Furthering our work in motion pictures, Intel artificial intelligence libraries along with Total Chaos VRay with Intel® Embree were used by FX studio Ziva to create the prehistoric, 75-foot long shark in this summer’s Warner Bros. action movie The Meg.*

Within high performance computing (HPC), one of the most exciting areas is in the realm of in situ visualization, the technique where visualizations are created together with the simulation code without utilizing storage. In situ visualization will be the norm in just a few years as more advanced supercomputers are built; visual analysis will be done right from the same memory used by the simulation with almost instant turn-around from compute to visualization without costly I/O. The recent launch of Intel® Optane™ DC Persistent Memory will be an incredibly powerful tool in these scenarios.

Another exciting area we are enabling is visualization “in the cloud.” With the growing emphasis on HPC availability in private and public cloud infrastructures, the Intel Rendering Framework based on open source software rendering is a natural fit for cloud solutions because it doesn’t require special GPU hardware instances. One exciting project that will be discussed at the Intel Booth Wednesday at 3:30pm is Tapestry from the University of Tennessee. Tapestry is a container application available on DockerHub that utilizes Intel OSPRay for scalable rendering across multiple nodes in a private or public cloud infrastructure. Tapestry provides cloud capable rendering particularly suited to scientific visualizations.  It also includes a movie editing and rendering suite. Come by the Intel SC18 Booth for tutorials and presentations from our SDVis community partners Wednesday from 2pm to 4:30pm.

SC18 Demos

Intel® Rendering Framework libraries and the SDVis applications that use it are the foundation that culminated in the release of our Intel® Selection Solutions for Professional Visualization, which harnesses SDVis for high performance, convenient and cost-effective deployment. At SC18 this year at the Intel booth, we will be showing four visualization uses showcasing the breadth of workloads that can be achieved with Intel® Select Solutions and Intel® Rendering Framework:

Auto CFD Virtual Windtunnel

  • OpenFOAM* Foundation v5.0* code1 + ParaView* + Intel® OSPRay
  • ~250GB of data—run on-site with an Intel® Select Solutions for Pro Visualization PLUS cluster at 10-30 frames per second

Intel SC18

Water Formation in the Universe

  • SC’18 Visualization and Data Analytics Showcase Finalist.
  • ParaView v5.6 + Intel® OSPRay and Intel® OpenSWR Animation.
  • 110 GBs of data, run remotely on Intel Select Solutions for Pro Visualization PLUS with AWS Nice DCV remote desktop app interactively at 10-30 frames per second.
  • It is believed that as much as half of the water in our solar system predates the formation of the Sun. In order to better understand how this early water formed, scientists at Los Alamos National Laboratory have coupled cosmological simulations with an inline, fully implicit, molecular chemistry solver. This combination allows scientists to observe the formation of water, methane, carbon dioxide, and other molecules in the early cosmos.
  • Using Intel® OSPRay and OpenSWR rendering libraries allows for the interactive rendering of these simulations on compute clusters, enabling interactive data exploration.
  • Research data used courtesy of Joseph Smidt, Los Alamos National Laboratory, and Brandon Wiggins, Southern Utah University.
Intel SC18
Image courtesy of Greg Abram from TACC at University of Texas, Austin

Argonne National Lab Theta supercomputer: Star Radiation Properties

  • October 2018 Nature cover story; SC’18 Vis and DA Showcase Finalist.
  • The geometry visualized for the Vis Showcase was generated on the Theta Supercomputer at Argonne National Laboratory using ParaView.
  • On the show flow, live rendering a ~350GB (in memory) version of the dataset run both on-site on Intel® Select Solutions for Pro Vis PLUS cluster and remotely on Argonne Nat’l Labs Cooley supercomputer interactively at 30+ frames per second.
  • 30TB+ data set simulated by a team from the University of California, Santa Barbara, led by Lars Bildsten.
  • This demo shows luminous blue variables: gigantic stars that are one hundred times bigger than the sun. Knowing the cause of the explosions of these luminous blue variables gives scientists a more complete picture of the life and death of the biggest stars in the universe.

Intel SC18

 

Energy / Oil and Gas Discovery: The Moroccan Ocean Shelf

  • Energy/Oil and Gas Discovery: The Moroccan Atlantic Margin
  • Beta “OSPRay Studio” Viewer showing 110GB data of detailed layering and geologic structures mapped by seismic geoscientists to identify oil/gas potential run remotely on Intel® Select Solution for Pro Visualization PLUS cluster with AWS NICE DCV remote desktop application at up to 50 frames per second
  • Research data used courtesy of the Bureau of Economic Geology at The University of Texas at Austin and the Moroccan National Office of Hydrocarbon and Mining

Intel SC18

Look for more Intel® Rendering Framework demos in partner booths including Atipa’s Intel® Select Solutions for Pro Visualization running LAMMPS in situ via SENSEI and Intel OSPRay on two platforms with different processor configurations, demonstrating Intel OSPRay scalability. And don’t miss the HPE booth running the latest in Einstein general relativity and gravity analysis via Cosmic String simulations using GR-CHOMBO and Catalyst/ParaView in situ on a large scale HPE SMP supercomputer.

If you’ll be attending SC18, I encourage you to come by the Intel booth, Atipa booth and HPE Booth to see all of the demos we have showcasing our expertise in HPC for visual discovery across various workloads.


1 Data collected with OpenFOAM* Foundation v5.0. This offering is not approved or endorsed by OpenCFD Limited, producer and distributor of the OpenFOAM software via https://www.openfoam.com/, and owner of the OPENFOAM* and OpenCFD* trademarks.

Published on Categories High Performance ComputingTags , , , , , ,
Jim Jeffers

About Jim Jeffers

Jim Jeffers leads Intel's Visualization Solutions Team developing open source CPU based rendering libraries used for generating animated movies, special effects, automobile design and scientific visualization working with companies like Disney and DreamWorks plus scientists like Stephen Hawking. Jim joined Intel in 2008 participating in the development of manycore parallel computing and the Intel® Xeon Phi™ product family, including co-authoring 4 Books on manycore parallel programming. Jim's experience includes software design and technical leadership in high performance computing, graphics, digital television, and data communications. Jim's notable work prior to Intel includes development for the Tech Emmy winning virtual 'First Down Line' technology seen on live American football TV broadcasts.