Visualization and Filesystem Use Cases Show Value of Large Memory Fat Nodes on Frontera

By Rob Farber

February 2, 2021

Frontera, the world’s largest academic supercomputer housed at the Texas Advanced Computing Center (TACC), is big both in terms of number of computational nodes and the capabilities of the large memory “fat” compute nodes. A couple of recent use cases demonstrate how academic researchers are using the quad-socket, 112-core, 2.1 TB persistent memory to support Frontera’s large memory nodes to advance a wide variety of research topics including visualization and filesystems.

Visualization

The advent of Software Defined Visualization (SDVis) is a seismic event in the visualization community because it permits interactive, high-resolution, photorealistic visualization of large data without having to move the data off the compute nodes. In transit and in situ visualization are two techniques that allow SDVis libraries such as Embree and OSPRay to render data on the same nodes that generate the data. In situ visualization renders data for visualization on the same computational nodes that perform the simulation. In transit visualization lets users tailor the render vs simulation workload by using a subset of the computation nodes for rendering.

“The HPC community is entering a new era in photorealistic, interactive visualization using SDVis,” said Dr. Paul Navrátil, director of visualization at TACC.  “For many users, the Optane-enhanced nodes on Frontera provide their first opportunity to experience the performance and visualization benefits of SDVis to analyze data requiring a very large shared memory resource.”

The quad socket Intel Xeon Platinum 8280M large memory Frontera nodes give scientists the ability to interactively render and see important events (due to CPU-based rendering) and – again interactively – jump back in the data to examine what caused the important event to happen. This interactive “instant replay” capability is enabled by the high core count, high-bandwidth (six memory channels per socket or 24 memory channels total) of the TACC large memory 2.1 TB nodes.

Jim Jeffers (senior principal engineer and senior director of advanced rendering and visualization at Intel) has been a central mover and shaker in HPC visualization with his work on SDVis and the Intel Embree and Intel OSPRay libraries. He explains, “Optane Persistent Memory provides scientists with the memory capacity, bandwidth, and persistence features to enable a new level of control and capability to interactively visualize large data sets in real time and with up to film-quality fidelity. Scientists are able to recognize or more easily identify key occurrences and interactively step forward and backward in time to see and understand the scientific importance. Because of the cost of moving large data from memory to storage or over the internet can be hours, days, and beyond; this is a capability that will be critical to fully take advantage of Exascale and beyond computing.”

Use Case: Visualizing OpenFOAM Fluid Flows

David DeMarle (Intel computer graphics software engineer) points out that the 2.1 TB memory capacity in the Frontera large memory nodes gives users the ability to keep extensive histories of their OpenFOAM simulations in memory. Using software, scientists can trigger on an event, receive an alert that the event has happened, and then review the causes of the event. Collisions, defined as an event where multiple particles are contained in a voxel or 3D block in space, are one example of an important fluid flow event. Alternatives include triggers that occur when the pressure exceeds or drops below a threshold in a voxel.

Memory capacity is important to preserving the simulation histories that help scientists understand physical phenomena as modern systems can simulate larger, more complex systems with higher fidelity.  Keeping data in the persistent memory devices delivers a performance boost. DeMarle observes, “The runtime savings is highly correlated to amount of memory, which implies that the savings will scale to large runs both in terms of size and resolution.” Scalable approaches are important as we move into the exascale computing era.

DeMarle and his collaborators used in situ methods to create their OpenFOAM visualizations and histories so the data does not have to move off the computational nodes. They called the Catalyst library to perform the in situ rendering. Alternatively, users can also perform in situ visualization using the OpenFOAM Catalyst adapter. ParaView was used as the visualization tool.

To control resource utilization, Catalyst calls the open-source Intel memkind library. This provides two advantages: (1) the persistent memory capacity could be allocated for use by the simulation (using Memory Mode) and (2) data could be directly written to the persistent memory devices using App Direct mode.

In Memory Mode, the DRAM acts as a cache for the most frequently-accessed data, while Intel Optane persistent memory provides the large memory capacity. No program modifications are required for an application to use Memory Mode.

With App Direct mode, the software directly reads and writes the persistent memory media via the DAX (data analysis expressions library) instructions that are available on select Intel Xeon Scalable processors. [i] These instructions let the CPU directly access the persistent memory media via transactional operations that keep user data safe, thus bypassing the traditional operating systems bottlenecks such as the page cache and block layer. [ii]

DeMarle notes that multi-threaded applications are required to realize the performance benefits due to the high memory bandwidth of the Intel Optane memory. “Single threaded performance gains are negligible while the performance differences are clear when running multithreaded.”

Overall, DeMarle believes the TACC large memory nodes are a good transition for people to get used to having a terabyte of memory on a workstation. In particular, in-situ visualization is a good use of the increased memory capacity due to the Intel Optane persistent memory tier.

Figure 1: Intel Optane devices can be added at many locations in the HPC storage memory hierarchy

180-degree High Resolution Climate Simulations

Scientists are also visualizing photorealistic, high-resolution climate data on the TACC Frontera large memory nodes. The German Climate Computing Centre (Deutsches Klimarechenzentrum, DKRZ) in collaboration with Intel is investigating the capability of the persistent memory capacity of the TACC nodes. In this investigation, the Intel Optane persistent memory devices are operating in Memory Mode, so no code modifications or adapters were required.

Carson Brownlee (graphics software engineer, researcher and part of the OneAPI Render Kit team at Intel) is part of the team that integrated a path tracer in the popular ParaView visualization tool. The path tracer allows photorealistic visualizations of the DKRZ high resolution data.

Brownlee notes, “The TACC visualization work was an extension of the simulations and visualizations conducted at DKRZ (the Eurovision/supercomputing submissions were done at DKRZ). We were able to view many timesteps of this data during our demos due to the large memory space.” The amazing fidelity of the DKRZ simulations can be seen in the Eurovision video at https://www.youtube.com/watch?v=gQl_RQLw2Mw. This research project was a finalist in the scientific visualization track at Supercomputing 2020.

Of course, everything in the visualization relies on the simulated climate data. The DKRZ climate simulations are remarkable in that they can model local climate systems to 150m of detail. Further, DKRZ is experimenting with 1 kilometer global data modeled for the years 1850 to 2300. Statistics such as mean temperature, precipitation can be used to validate the simulation against data recorded during this time period and to date.  The effects of CO2 can also be evaluated by comparison between simulated and historical data.

Providing a high-level summary, Niklas Röber (DKRZ visualization staff) said, “Visualization is analysis of the data in visual form. You can make annotations, debug, and even steer the model using this technology.” He continues, “Without OSPRay and pathtracing this video would not have been possible. It is really exciting to see the visualization on a 180-degree display.”

The simulated data can be rendered on a 180-degree display. Röber notes, “We hope to make this available for viewing in places such as planetariums.” The following image conveys the impact of photorealistic climate model visualization in print format. The difference compared to ray casting is stunning. You can also view the SC20 presentation and video at https://nextcloud.dkrz.de/s/cBTBso8w5T8bpSq.

Figure 2: A representative comparison of ray casting and volumetric path tracing with Intel OSpRay renders of high-resolution climate data (Images courtesy Niklas Rober (DKRZ), Michael Migliore (Kitware) and Carson Browning (Intel)

Size is a significant challenge with such high-resolution data. According to Brownlee, the demo data consumes 1.3 TB. The 2.1 TB persistent memory capacity on the Frontera large memory nodes allows loading of 150 time steps. These numbers can help the reader get a sense of why terabyte storage capacity is so important in preserving interactive response time.

Use Case: Order of Magnitude Faster Filesystem Research

In another use case, Simon Peter (Professor of Computer Science at U.T. Austin) is leading the effort on the Assise, an open source, client-local distributed filesystem. [i] Assise is a French name that describes a geographic stratum of rock.

Recent innovations in nonvolatile storage media – like Optane persistent memory devices that operate off the system memory bus – have made filesystem research a hot research topic. The performance characteristics are notable. For example, the persistent memory devices have already been shown to deliver latency measured in nanoseconds and bandwidth measured in tens of gigabytes per second per node.[ii]

Peter’s research effort at TACC looks to incorporate the benefits of NVM technologies in a distributed environment by creating a topology-aware filesystem. “It’s all about locality,” he observes.

Assise breaks that model and instead uses a NUMA like model where the thread that creates the data owns it as well as the lock that allows coherent operations on the data in a parallel/distributed environment.  The software does not use the Linux VFS layer. Rather it intercepts filesystem calls from the application. Direct access to storage from user space bypasses all the overhead of the OS system call, page cache, VFS (Virtual File System) layer, block cache, PCIe driver and PCIe bus.

These are some of the key reasons why the Assise github page states, “Assise accelerates POSIX file IO by orders of magnitude by leveraging client-local NVM without kernel involvement, block amplification, or unnecessary coherence overheads.” Please see the OSDI paper for more detail.

Rob Farber is a global technology consultant and author with an extensive background in HPC and in developing machine learning technology that he applies at national labs and commercial organizations. Rob can be reached at info@techenablement.com.

[i] https://www.intel.com/content/dam/support/us/en/documents/memory-and-storage/data-center-persistent-mem/Intel-Optane-DC-Persistent-Memory-Quick-Start-Guide.pdf

[ii] https://software.intel.com/content/www/us/en/develop/articles/introduction-to-programming-with-persistent-memory-from-intel.html

[iii] https://wreda.github.io/papers/assise-osdi20.pdf

[iv] https://lenovopress.com/lp1083.pdf

Subscribe to HPCwire's Weekly Update!

Be the most informed person in the room! Stay ahead of the tech trends with industy updates delivered to you every week!

Multiverse Targets ‘Quantum Computing for the Masses’

January 19, 2022

The race to deliver quantum computing solutions that shield users from the underlying complexity of quantum computing is heating up quickly. One example is Multiverse Computing, a European company, which today launched the second financial services product in its Singularity product group. The new offering, Fair Price, “delivers a higher accuracy in fair price calculations for financial... Read more…

Students at SC21: Out in Front, Alongside and Behind the Scenes

January 19, 2022

The Supercomputing Conference (SC) is one of the biggest international conferences dedicated to high-performance computing, networking, storage and analysis. SC21 was a true ‘hybrid’ conference, with a total of 380 o Read more…

New Algorithm Overcomes Hurdle in Fusion Energy Simulation

January 15, 2022

The exascale era has brought with it a bevy of fusion energy simulation projects, aiming to stabilize the notoriously delicate—and so far, unmastered—clean energy source that would transform the world virtually overn Read more…

Summit Powers Novel Protein Function Prediction Work

January 13, 2022

There are hundreds of millions of sequenced proteins and counting—but only 170,000 have had their structures solved by researchers, bottlenecking our understanding of proteins and their functions across organisms’ ge Read more…

Q-Ctrl – Tackling Quantum Hardware’s Noise Problems with Software

January 13, 2022

Implementing effective error mitigation and correction is a critical next step in advancing quantum computing. While a lot of attention has been given to efforts to improve the underlying ‘noisy’ hardware, there's be Read more…

AWS Solution Channel

shutterstock 377963800

New – Amazon EC2 Hpc6a Instance Optimized for High Performance Computing

High Performance Computing (HPC) allows scientists and engineers to solve complex, compute-intensive problems such as computational fluid dynamics (CFD), weather forecasting, and genomics. Read more…

Nvidia Defends Arm Acquisition Deal: a ‘Once-in-a-Generation Opportunity’

January 13, 2022

GPU-maker Nvidia is continuing to try to keep its proposed acquisition of British chip IP vendor Arm Ltd. alive, despite continuing concerns from several governments around the world. In its latest action, Nvidia filed a 29-page response to the U.K. government to point out a list of potential benefits of the proposed $40 billion deal. Read more…

Multiverse Targets ‘Quantum Computing for the Masses’

January 19, 2022

The race to deliver quantum computing solutions that shield users from the underlying complexity of quantum computing is heating up quickly. One example is Multiverse Computing, a European company, which today launched the second financial services product in its Singularity product group. The new offering, Fair Price, “delivers a higher accuracy in fair price calculations for financial... Read more…

Students at SC21: Out in Front, Alongside and Behind the Scenes

January 19, 2022

The Supercomputing Conference (SC) is one of the biggest international conferences dedicated to high-performance computing, networking, storage and analysis. SC Read more…

Q-Ctrl – Tackling Quantum Hardware’s Noise Problems with Software

January 13, 2022

Implementing effective error mitigation and correction is a critical next step in advancing quantum computing. While a lot of attention has been given to effort Read more…

Nvidia Defends Arm Acquisition Deal: a ‘Once-in-a-Generation Opportunity’

January 13, 2022

GPU-maker Nvidia is continuing to try to keep its proposed acquisition of British chip IP vendor Arm Ltd. alive, despite continuing concerns from several governments around the world. In its latest action, Nvidia filed a 29-page response to the U.K. government to point out a list of potential benefits of the proposed $40 billion deal. Read more…

Nvidia Buys HPC Cluster Management Company Bright Computing

January 10, 2022

Graphics chip powerhouse Nvidia today announced that it has acquired HPC cluster management company Bright Computing for an undisclosed sum. Unlike Nvidia’s bid to purchase semiconductor IP company Arm, which has been stymied by regulatory challenges, the Bright deal is a straightforward acquisition that aims to expand... Read more…

SC21 Panel on Programming Models – Tackling Data Movement, DSLs, More

January 6, 2022

How will programming future systems differ from current practice? This is an ever-present question in computing. Yet it has, perhaps, never been more pressing g Read more…

Edge to Exascale: A Trend to Watch in 2022

January 5, 2022

Edge computing is an approach in which the data is processed and analyzed at the point of origin – the place where the data is generated. This is done to make data more accessible to end-point devices, or users, and to reduce the response time for data requests. HPC-class computing and networking technologies are critical to many edge use cases, and the intersection of HPC and ‘edge’ promises to be a hot topic in 2022. Read more…

Citing ‘Shortfalls,’ NOAA Targets Hundred-Fold HPC Increase Over Next Decade

January 5, 2022

From upgrading the Global Forecast System (GFS) to acquiring new supercomputers, the National Oceanic and Atmospheric Administration (NOAA) has been making big moves in the HPC sphere over the last few years—but now it’s setting the bar even higher. In a new report, NOAA’s Science Advisory Board (SAB) highlighted... Read more…

IonQ Is First Quantum Startup to Go Public; Will It be First to Deliver Profits?

November 3, 2021

On October 1 of this year, IonQ became the first pure-play quantum computing start-up to go public. At this writing, the stock (NYSE: IONQ) was around $15 and its market capitalization was roughly $2.89 billion. Co-founder and chief scientist Chris Monroe says it was fun to have a few of the company’s roughly 100 employees travel to New York to ring the opening bell of the New York Stock... Read more…

US Closes in on Exascale: Frontier Installation Is Underway

September 29, 2021

At the Advanced Scientific Computing Advisory Committee (ASCAC) meeting, held by Zoom this week (Sept. 29-30), it was revealed that the Frontier supercomputer is currently being installed at Oak Ridge National Laboratory in Oak Ridge, Tenn. The staff at the Oak Ridge Leadership... Read more…

AMD Launches Milan-X CPU with 3D V-Cache and Multichip Instinct MI200 GPU

November 8, 2021

At a virtual event this morning, AMD CEO Lisa Su unveiled the company’s latest and much-anticipated server products: the new Milan-X CPU, which leverages AMD’s new 3D V-Cache technology; and its new Instinct MI200 GPU, which provides up to 220 compute units across two Infinity Fabric-connected dies, delivering an astounding 47.9 peak double-precision teraflops. “We're in a high-performance computing megacycle, driven by the growing need to deploy additional compute performance... Read more…

Intel Reorgs HPC Group, Creates Two ‘Super Compute’ Groups

October 15, 2021

Following on changes made in June that moved Intel’s HPC unit out of the Data Platform Group and into the newly created Accelerated Computing Systems and Graphics (AXG) business unit, led by Raja Koduri, Intel is making further updates to the HPC group and announcing... Read more…

Nvidia Buys HPC Cluster Management Company Bright Computing

January 10, 2022

Graphics chip powerhouse Nvidia today announced that it has acquired HPC cluster management company Bright Computing for an undisclosed sum. Unlike Nvidia’s bid to purchase semiconductor IP company Arm, which has been stymied by regulatory challenges, the Bright deal is a straightforward acquisition that aims to expand... Read more…

D-Wave Embraces Gate-Based Quantum Computing; Charts Path Forward

October 21, 2021

Earlier this month D-Wave Systems, the quantum computing pioneer that has long championed quantum annealing-based quantum computing (and sometimes taken heat fo Read more…

Killer Instinct: AMD’s Multi-Chip MI200 GPU Readies for a Major Global Debut

October 21, 2021

AMD’s next-generation supercomputer GPU is on its way – and by all appearances, it’s about to make a name for itself. The AMD Radeon Instinct MI200 GPU (a successor to the MI100) will, over the next year, begin to power three massive systems on three continents: the United States’ exascale Frontier system; the European Union’s pre-exascale LUMI system; and Australia’s petascale Setonix system. Read more…

Three Chinese Exascale Systems Detailed at SC21: Two Operational and One Delayed

November 24, 2021

Details about two previously rumored Chinese exascale systems came to light during last week’s SC21 proceedings. Asked about these systems during the Top500 media briefing on Monday, Nov. 15, list author and co-founder Jack Dongarra indicated he was aware of some very impressive results, but withheld comment when asked directly if he had... Read more…

Leading Solution Providers

Contributors

Lessons from LLVM: An SC21 Fireside Chat with Chris Lattner

December 27, 2021

Today, the LLVM compiler infrastructure world is essentially inescapable in HPC. But back in the 2000 timeframe, LLVM (low level virtual machine) was just getting its start as a new way of thinking about how to overcome shortcomings in the Java Virtual Machine. At the time, Chris Lattner was a graduate student of... Read more…

2021 Gordon Bell Prize Goes to Exascale-Powered Quantum Supremacy Challenge

November 18, 2021

Today at the hybrid virtual/in-person SC21 conference, the organizers announced the winners of the 2021 ACM Gordon Bell Prize: a team of Chinese researchers leveraging the new exascale Sunway system to simulate quantum circuits. The Gordon Bell Prize, which comes with an award of $10,000 courtesy of HPC pioneer Gordon Bell, is awarded annually... Read more…

The Latest MLPerf Inference Results: Nvidia GPUs Hold Sway but Here Come CPUs and Intel

September 22, 2021

The latest round of MLPerf inference benchmark (v 1.1) results was released today and Nvidia again dominated, sweeping the top spots in the closed (apples-to-ap Read more…

Three Universities Team for NSF-Funded ‘ACES’ Reconfigurable Supercomputer Prototype

September 23, 2021

As Moore’s law slows, HPC developers are increasingly looking for speed gains in specialized code and specialized hardware – but this specialization, in turn, can make testing and deploying code trickier than ever. Now, researchers from Texas A&M University, the University of Illinois at Urbana... Read more…

Julia Update: Adoption Keeps Climbing; Is It a Python Challenger?

January 13, 2021

The rapid adoption of Julia, the open source, high level programing language with roots at MIT, shows no sign of slowing according to data from Julialang.org. I Read more…

Top500: No Exascale, Fugaku Still Reigns, Polaris Debuts at #12

November 15, 2021

No exascale for you* -- at least, not within the High-Performance Linpack (HPL) territory of the latest Top500 list, issued today from the 33rd annual Supercomputing Conference (SC21), held in-person in St. Louis, Mo., and virtually, from Nov. 14–19. "We were hoping to have the first exascale system on this list but that didn’t happen," said Top500 co-author... Read more…

TACC Unveils Lonestar6 Supercomputer

November 1, 2021

The Texas Advanced Computing Center (TACC) is unveiling its latest supercomputer: Lonestar6, a three peak petaflops Dell system aimed at supporting researchers Read more…

10nm, 7nm, 5nm…. Should the Chip Nanometer Metric Be Replaced?

June 1, 2020

The biggest cool factor in server chips is the nanometer. AMD beating Intel to a CPU built on a 7nm process node* – with 5nm and 3nm on the way – has been i Read more…

  • arrow
  • Click Here for More Headlines
  • arrow
HPCwire