Exascale Readiness Key to Solving High Energy Physics Mysteries

By Linda Barney

April 13, 2022

Scientists at Brookhaven National Laboratory, Columbia University, the University of Connecticut, University of Edinburgh, Regensburg University, and the University of Southampton are seeking answers to physics mysteries at the highest energies and shortest distances. The team is devising new methods and enhancing their code in order to exploit the huge potential of the forthcoming Intel-HPE Aurora exascale supercomputer.

Dr. Norman Christ, Ephraim Gildor Professor of Computational Theoretical Physics at Columbia University explains, “Supercomputers are required because we are studying phenomena that involve quarks and gluons whose interactions require enormous computer resources to accurately predict results. Using the Aurora exascale supercomputer will allow us to do high energy physics research that isn’t possible on existing supercomputers. Our research focuses on rare processes that are sensitive to new phenomena, which can appear directly only in experiments performed at energies far higher than those accessible in present particle accelerators.”

Some of the questions the team hopes to answer include:

  • Can the 0.0000000000001% difference between the masses of the two species of neutral K meson be predicted by the Standard Model of particle physics?
  • What are the contributions of the up, down, strange and charm quarks at low energy to the asymmetries between particles and anti-particles found in the decays of these K mesons?
  • Does the new measurement of a very rare decay of a K+ meson into a p+ meson plus two neutrinos that occurs once in every ten billion decays, agree with the prediction of the Standard Model?

Physicists hope the answers to questions such as these will provide clues to underlying laws of nature that governed the origins of our universe.

The team is using early versions of the Intel Xe GPUs and anticipates a 50X processing speed-up in running calculations compared to those done on the ALCF Mira supercomputer. The team’s improved codes targeting Aurora make extensive use of the oneAPI multi-platform framework.

Discovering the nature of sub-atomic particles

Physics research seeks to discover information about the fundamental elements of matter and the laws, which govern their interactions. There are important phenomena observed on a cosmic scale such as the dominance of matter over anti-matter and the observations that suggest the existence of dark matter and dark energy that hint at new particles and interactions not present in the Standard Model.

Central to the Standard Model are elementary particles called quarks and gluons that combine to form particles known as hadrons which include the protons and neutrons that make up the atomic nucleus. Quarks have properties such as mass, electrical charge, color charge and spin. There are six known types of quark named up, down, charm, strange, top, and bottom.

For every type of quark, an antiquark exists with the same properties as the quark but with the opposite sign (negative versus positive) for the electric and color charges. The four heavier types of quark can change into up and down quarks through a process called weak decay. Extensive experiments performed over the past 70 years revealed how these quarks combine to form hundreds of different types of stable and unstable particles. In the 1970’s, scientists discovered that the large array of different hadron particles were actually different combinations of these six different types of quarks. The forces causing the quarks to combine could be described by quantum chromo-dynamics (QCD) equations that describe the strong nuclear force. However, solving these QCD equations was impossible before the development of supercomputers.

The realistic solution of QCD equations is only possible on modern supercomputers. The approach using supercomputers is dramatically different from the usual approach to quantum problems at the atomic scale. Richard Feynman developed an approach that requires carefully generating samples or snapshots of the fluctuating fields that describe the forces between the quarks. These samples indicate points in a space of 5 billion dimensions and require powerful computers to run the simulations.

Lattice QCD, developed in the 1970s, replaces the space-time continuum by a regular grid or lattice of locations, each corresponding to a specific point in space at a specific time. Research accomplishments using these methods evolved with each new generation of supercomputers enabling exciting new results which were previously out of reach. The large step in computer power represented by the class of exascale computers such as Aurora will again propel lattice QCD research to a new level.

Lattice QCD case study

The study of fundamental questions in particle physics using lattice QCD is a large international effort actively pursued on the fastest supercomputers around the world. Christ states, “In preparing to use an important new supercomputer such as Aurora, our research groups work together as a single Aurora Early Science project so that what is learned can benefit everyone. Our project is one part of the effort to be ready to use Aurora as soon as it begins to operate.”

“The weak decays in which the heavier four quarks (top, bottom, charm, and strange) transform into the two lighter up and down quarks are caused by processes that happen at very high energies. This energy is approximately one hundred times higher than the energy stored in the mass of the proton or neutron—the constituents from which the atomic nucleus is built. These high energies can accurately be treated by a complementary method called QCD perturbation theory. Lattice QCD can be used to solve the low energy part of the problem while QCD perturbation theory can handle this high-energy part.”

“The charm quark is too heavy to treat accurately with lattice QCD but too light to allow QCD perturbation theory to be used. Fortunately, this is a problem that Aurora can solve. The underlying difficulty created by the large mass of the charm quark is caused by its small quantum size. As a particle becomes more massive, its properties become more sensitive to what happens at shorter distances. If lattice QCD is to accurately describe the physics of charm quarks, the artificial distance between the lattice points must be made smaller than that needed to treat the less massive up, down, and strange quarks. If the lattice points are to be placed closer together, far more points will be needed to fill a volume of an appropriate physical size. These issues make the problem much more difficult. Typically, if the spacing between lattice points is reduced by a factor of two, the computational time required to solve this new problem will by one hundred times greater. Thus, this entire class of questions can for the first time be effectively tackled with Aurora.”

Figure 1. – Sketch of the quantum process in which a K0 meson transforms into its anti-particle. The wavy lines represent gluons, the red lines light up and down quarks and the green line indicates a strange quark. The yellow squares show Standard Model weak interactions in which a strange quark transforms into three light quarks. The underlying grid structure indicates the process as computed using lattice QCD. Courtesy of Dr. Christ, Columbia University.

Software used in the high energy research

A large component of lattice QCD research is devoted to developing powerful algorithms and highly efficient computer code. The change from supercomputers constructed of arrays of interconnected CPUs (central processing units) to arrays of interconnected clusters of GPUs (graphics processing units) created huge challenges and huge opportunities for those developing lattice QCD codes and algorithms. The software used for the team’s ESP project is based on the open-source Grid system, software targeting lattice QCD developed by Peter Boyle of Brookhaven and the University of Edinburgh and his collaborators.

Grid was developed to replace the low-level but high-performance assembly language code used in early lattice QCD calculations. Grid exploits the features of the C++11 language and the significant advances in compiler technology to achieve assembly-language performance even for highly parallel CPU architectures from a high-level language. In fact, the abstractions used in Grid can also provide platform portability. Grid code can now be compiled both for CPUs and GPUs giving high performance on both architectures.

Preparing to run on the Aurora supercomputer

Aurora is taking advantage of the oneAPI industry initiative that will free researchers from the proprietary CUDA software that is widely used on the Summit supercomputer. This provides for code development on a broad-based open standard allowing developers to program once and run on different hardware from different vendors. To do so, Aurora supports the SYCL C++ abstraction layer that is part of the oneAPI industry initiative. As described above, running on Aurora required generalizations of Grid since lattice QCD code must be able to compile on both CPU and GPU systems.

Christ states, “Perhaps the greatest difficulty for lattice QCD is caused by the substantial increase in GPU computing power which is not matched by the performance of the communications network that interconnects the Aurora nodes, which are themselves clusters of well-connected GPUs. Since lattice QCD must exchange data between these nodes, such an imbalance between the processing power of the nodes and the bandwidth of the data communication between the nodes can result in a ‘network-bound’ application in which the slower communications network limits the overall performance.” The Intel solution slated for Aurora includes two Next Generation Intel Xeon Scalable processors (codename Sapphire Rapids) paired with six Intel Xe GPUs offering significant computing power for the available network performance. Application algorithms can be modified like lattice QCD has done to adjust to this architectural change and achieve optimal performance.

The Aurora supercomputer will also use Intel’s Next Gen Xeon Scalable Processor plus High Bandwidth Memory (HBM). It offers 64GB of high-bandwidth memory configurable in three modes: HBM-only, Flat, and Cache. End users running on Aurora can determine the mode in which to run the workload. It is anticipated that applications and problems that can fit within the available 64GB of HBM can run in HBM-only mode without changes. Similarly for those applications requiring more than 64GB, cache mode can be utilized without changes. Flat memory mode offers users greater control over tuning for optimal performance if developers are willing and able to make some algorithmic changes.

“A major goal of the team’s Early Science preparations is to overcome these network challenges by exploiting the size and power of the individual nodes. To avert the QCD communication problem, the team is reorganizing the calculation. A larger fraction of the calculation is divided among the powerful nodes and carried out with no communication between the nodes, using only infrequent bursts of communication between the nodes. Fortunately, there are earlier domain-decomposed algorithms that have this character, and which are now being actively reconfigured for Aurora’s architecture,” states Christ.  This work will also benefit QCD on future technologies expected to continue this trend.

Future Lattice QCD Research

“Having the power of the exascale Aurora supercomputer will open new vistas for lattice QCD research. For those processes in which the charm quark plays a central role, we expect to carry out calculations that were not previously possible, obtaining results with a level of precision and accuracy with the potential to reveal new particles and interactions not present in the Standard Model. The same physical laws determine the behavior of matter at the smallest and largest scales. We may dream that these discoveries will also help to explain unanswered questions raised by the large-scale structure of the Universe such as nature of dark matter, dark energy, and the preponderance of matter over anti-matter, a dominance necessary for our existence,” says Christ.

The ALCF is a DOE Office of Science User Facility.

Linda Barney is the founder and owner of Barney and Associates, a technical/marketing writing, training, and web design firm in Beaverton, OR.

Subscribe to HPCwire's Weekly Update!

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

MLPerf Inference 4.0 Results Showcase GenAI; Nvidia Still Dominates

March 28, 2024

There were no startling surprises in the latest MLPerf Inference benchmark (4.0) results released yesterday. Two new workloads — Llama 2 and Stable Diffusion XL — were added to the benchmark suite as MLPerf continues Read more…

Q&A with Nvidia’s Chief of DGX Systems on the DGX-GB200 Rack-scale System

March 27, 2024

Pictures of Nvidia's new flagship mega-server, the DGX GB200, on the GTC show floor got favorable reactions on social media for the sheer amount of computing power it brings to artificial intelligence.  Nvidia's DGX Read more…

Call for Participation in Workshop on Potential NSF CISE Quantum Initiative

March 26, 2024

Editor’s Note: Next month there will be a workshop to discuss what a quantum initiative led by NSF’s Computer, Information Science and Engineering (CISE) directorate could entail. The details are posted below in a Ca Read more…

Waseda U. Researchers Reports New Quantum Algorithm for Speeding Optimization

March 25, 2024

Optimization problems cover a wide range of applications and are often cited as good candidates for quantum computing. However, the execution time for constrained combinatorial optimization applications on quantum device Read more…

NVLink: Faster Interconnects and Switches to Help Relieve Data Bottlenecks

March 25, 2024

Nvidia’s new Blackwell architecture may have stolen the show this week at the GPU Technology Conference in San Jose, California. But an emerging bottleneck at the network layer threatens to make bigger and brawnier pro Read more…

Who is David Blackwell?

March 22, 2024

During GTC24, co-founder and president of NVIDIA Jensen Huang unveiled the Blackwell GPU. This GPU itself is heavily optimized for AI work, boasting 192GB of HBM3E memory as well as the the ability to train 1 trillion pa Read more…

MLPerf Inference 4.0 Results Showcase GenAI; Nvidia Still Dominates

March 28, 2024

There were no startling surprises in the latest MLPerf Inference benchmark (4.0) results released yesterday. Two new workloads — Llama 2 and Stable Diffusion Read more…

Q&A with Nvidia’s Chief of DGX Systems on the DGX-GB200 Rack-scale System

March 27, 2024

Pictures of Nvidia's new flagship mega-server, the DGX GB200, on the GTC show floor got favorable reactions on social media for the sheer amount of computing po Read more…

NVLink: Faster Interconnects and Switches to Help Relieve Data Bottlenecks

March 25, 2024

Nvidia’s new Blackwell architecture may have stolen the show this week at the GPU Technology Conference in San Jose, California. But an emerging bottleneck at Read more…

Who is David Blackwell?

March 22, 2024

During GTC24, co-founder and president of NVIDIA Jensen Huang unveiled the Blackwell GPU. This GPU itself is heavily optimized for AI work, boasting 192GB of HB Read more…

Nvidia Looks to Accelerate GenAI Adoption with NIM

March 19, 2024

Today at the GPU Technology Conference, Nvidia launched a new offering aimed at helping customers quickly deploy their generative AI applications in a secure, s Read more…

The Generative AI Future Is Now, Nvidia’s Huang Says

March 19, 2024

We are in the early days of a transformative shift in how business gets done thanks to the advent of generative AI, according to Nvidia CEO and cofounder Jensen Read more…

Nvidia’s New Blackwell GPU Can Train AI Models with Trillions of Parameters

March 18, 2024

Nvidia's latest and fastest GPU, codenamed Blackwell, is here and will underpin the company's AI plans this year. The chip offers performance improvements from Read more…

Nvidia Showcases Quantum Cloud, Expanding Quantum Portfolio at GTC24

March 18, 2024

Nvidia’s barrage of quantum news at GTC24 this week includes new products, signature collaborations, and a new Nvidia Quantum Cloud for quantum developers. Wh Read more…

Alibaba Shuts Down its Quantum Computing Effort

November 30, 2023

In case you missed it, China’s e-commerce giant Alibaba has shut down its quantum computing research effort. It’s not entirely clear what drove the change. Read more…

Nvidia H100: Are 550,000 GPUs Enough for This Year?

August 17, 2023

The GPU Squeeze continues to place a premium on Nvidia H100 GPUs. In a recent Financial Times article, Nvidia reports that it expects to ship 550,000 of its lat Read more…

Shutterstock 1285747942

AMD’s Horsepower-packed MI300X GPU Beats Nvidia’s Upcoming H200

December 7, 2023

AMD and Nvidia are locked in an AI performance battle – much like the gaming GPU performance clash the companies have waged for decades. AMD has claimed it Read more…

DoD Takes a Long View of Quantum Computing

December 19, 2023

Given the large sums tied to expensive weapon systems – think $100-million-plus per F-35 fighter – it’s easy to forget the U.S. Department of Defense is a Read more…

Synopsys Eats Ansys: Does HPC Get Indigestion?

February 8, 2024

Recently, it was announced that Synopsys is buying HPC tool developer Ansys. Started in Pittsburgh, Pa., in 1970 as Swanson Analysis Systems, Inc. (SASI) by John Swanson (and eventually renamed), Ansys serves the CAE (Computer Aided Engineering)/multiphysics engineering simulation market. Read more…

Choosing the Right GPU for LLM Inference and Training

December 11, 2023

Accelerating the training and inference processes of deep learning models is crucial for unleashing their true potential and NVIDIA GPUs have emerged as a game- Read more…

Intel’s Server and PC Chip Development Will Blur After 2025

January 15, 2024

Intel's dealing with much more than chip rivals breathing down its neck; it is simultaneously integrating a bevy of new technologies such as chiplets, artificia Read more…

Baidu Exits Quantum, Closely Following Alibaba’s Earlier Move

January 5, 2024

Reuters reported this week that Baidu, China’s giant e-commerce and services provider, is exiting the quantum computing development arena. Reuters reported � Read more…

Leading Solution Providers

Contributors

Comparing NVIDIA A100 and NVIDIA L40S: Which GPU is Ideal for AI and Graphics-Intensive Workloads?

October 30, 2023

With long lead times for the NVIDIA H100 and A100 GPUs, many organizations are looking at the new NVIDIA L40S GPU, which it’s a new GPU optimized for AI and g Read more…

Shutterstock 1179408610

Google Addresses the Mysteries of Its Hypercomputer 

December 28, 2023

When Google launched its Hypercomputer earlier this month (December 2023), the first reaction was, "Say what?" It turns out that the Hypercomputer is Google's t Read more…

AMD MI3000A

How AMD May Get Across the CUDA Moat

October 5, 2023

When discussing GenAI, the term "GPU" almost always enters the conversation and the topic often moves toward performance and access. Interestingly, the word "GPU" is assumed to mean "Nvidia" products. (As an aside, the popular Nvidia hardware used in GenAI are not technically... Read more…

Shutterstock 1606064203

Meta’s Zuckerberg Puts Its AI Future in the Hands of 600,000 GPUs

January 25, 2024

In under two minutes, Meta's CEO, Mark Zuckerberg, laid out the company's AI plans, which included a plan to build an artificial intelligence system with the eq Read more…

Google Introduces ‘Hypercomputer’ to Its AI Infrastructure

December 11, 2023

Google ran out of monikers to describe its new AI system released on December 7. Supercomputer perhaps wasn't an apt description, so it settled on Hypercomputer Read more…

China Is All In on a RISC-V Future

January 8, 2024

The state of RISC-V in China was discussed in a recent report released by the Jamestown Foundation, a Washington, D.C.-based think tank. The report, entitled "E Read more…

Intel Won’t Have a Xeon Max Chip with New Emerald Rapids CPU

December 14, 2023

As expected, Intel officially announced its 5th generation Xeon server chips codenamed Emerald Rapids at an event in New York City, where the focus was really o Read more…

IBM Quantum Summit: Two New QPUs, Upgraded Qiskit, 10-year Roadmap and More

December 4, 2023

IBM kicks off its annual Quantum Summit today and will announce a broad range of advances including its much-anticipated 1121-qubit Condor QPU, a smaller 133-qu Read more…

  • arrow
  • Click Here for More Headlines
  • arrow
HPCwire