In Australia, HPC Illuminates the Early Universe

By Oliver Peckham

May 11, 2020

Many billions of years ago, the universe was a swirling pool of gas. Unraveling the story of how we got from there to here isn’t an easy task, with many simulations of large swaths of the universe taking years to complete on powerful supercomputers. In a talk for the ICM Seminars series (hosted by the Interdisciplinary Centre for Mathematical and Computational Modelling University of Warsaw), Dr. Simon Mutch highlighted how Australian research organizations are working around the computational requirements to deliver insights into the origins of the universe as we know it.

Dr. Simon Mutch

The reionization of the universe – and why we should care about it

“Immediately after the Big Bang, the universe was filled mostly just with neutral hydrogen – there wasn’t really much else there,” said Mutch, who is a postdoctoral fellow at the University of Melbourne’s ASTRO 3D Centre of Excellence and a senior research data specialist for the Melbourne Data Analytics Platform. “But then there were gravitational perturbations that caused gas to collapse in on itself, and eventually stars and galaxies began to form, and those stars gave out light, which was of high enough energy to start to ionize the surrounding neutral gas, so it stripped electrons off that neutral gas and changed its properties.”

This ionization, he explained, spread into bubbles, and as galaxies grew in number and size, the bubbles began to overlap, eventually resulting in the total reionization of the universe – some 12.5 billion years ago. The relationship between these ionized bubbles and the galaxies that birthed them a major focus for Mutch and his colleagues. 

Over the course of billions of years, ionization bubbles grew to fill the entire universe. Image courtesy of Simon Mutch.

 “That’s really interesting, because it means that if we can observe this reionization signal … then we can infer something about the galaxies which are driving this reionization process.” Mutch said. “What’s even more interesting is that the reionization signal is sensitive to all galaxies.” In terms of the galaxies populating the universe, he explained, relatively faint galaxies are the most common – but also the most difficult to see. “By studying this reionization structure,” he said, “we can actually learn something about the very smallest, very faintest galaxies that we can’t actually see.” 

Mutch compared the process to dropping stones in a pond and studying the ripples to understand the shapes and sizes of each stone. “The main problem is: how do we connect the properties of galaxies to the signature of the bubbles that we see during reionization?” he said. “For that, we use cosmological simulations.”

The trouble with simulations

Cosmological simulations of galaxy formation build a chunk of the universe from the ground up, accounting for elements like gravity, dark matter, heating, cooling, turbulence, chemistry, supernovae, black holes, magnetic fields and more, which are all woven into hydrodynamical or mesh models.

“While these are incredibly powerful, they are extremely computationally intensive,” Mutch said. “That’s because there is a large dynamic range, both in terms of temporal and spatial resolution.” By way of examples, he discussed IllustrisTNG, a galaxy formation simulation one billion light years across that required 35 million CPU hours on the Hazel Hen supercomputer at the High Performance Computing Center (HLRS) in Stuttgart. Similarly, he said, the larger BlueTides simulation took 20 million CPU hours on the Blue Waters system at NCSA, nearly taking up the entire machine.

The necessary scale of simulating reionization compounded the high computational needs. “What we’re always doing is making this tradeoff between the amount of resolution we get in the simulation and the size of the simulation,” Mutch said. “This problem is particularly acute, though, if you’re talking about the early universe and reionization.” The reionization bubbles were tens of millions of light years across, so in order to produce a statistically relevant sample of them, you would need many bubbles – and a massive simulation.

Normally, researchers adjust parameters so they can match to the known universe nearby. But not much is known about the early universe – so instead, Mutch and his colleagues needed to run “many, many different realizations of the simulations” to test different models, feedback processes and other variables to see how they affected the ionized bubbles.

Finding a path through the cosmos

Tackling this uphill battle was the goal for the University of Melbourne’s Dark-ages, Reionization And Galaxy-formation Observables Numerical Simulation, or “DRAGONS,” program. (“We love our acronyms in astronomy,” Mutch said. “Everything needs to have a good acronym.”)

Thankfully, he said, the universe gave them a helping hand. Overall, the universe consists of around 70% dark energy, 25% dark matter and only 5% normal matter – what we interact with in our daily lives. “What this actually means is that we can do a pretty good job of simulating the position and the large-scale distribution of the matter by simply ignoring normal matter, and that makes things much easier,” Mutch said, explaining that they could ignore gas, shocks, star formation and more. “All we care about is getting the large-scale distribution of matter correct.”

The matter composition of the universe. Image courtesy of Simon Mutch.

So the researchers developed a N-body (particle) simulation that treated all the matter in the universe as collisionless. “We can pour all our computing power into doing this problem of gravity, essentially, and doing as big a simulation to as high a resolution as we possibly can,” Mutch said. They ran a large N-body simulation – about 300 million light years on each side – with billions of particles, each corresponding to a mass about 400 million times that of the sun.

The telltale “knots” circled in the N-body simulation. Image courtesy of Simon Mutch.

Looking at the simulation, the researchers then identified “knots” in the images – formations called “dark matter halos” where galaxies would start to form. The researchers tracked these halos through the simulation, building hierarchical merger trees that described how the halos coalesced over time. Using a semi-analytic galaxy formation model, they then “painted on” galaxies over the halos. 

“What we also did, which was unique at the time with the DRAGONS program,” Mutch said, “is that we use the information of these galaxies to calculate how many ionizing photons they were producing and then fed that into another code, called a seminumerical model, that then was able to give us what the ionization state of the volume was a function of position. So basically, it allowed us to figure out where these ionized bubbles were.”

With that process in hand, the researchers would then evolve the galaxies again, run the seminumerical model again to get the ionization results and repeat the process until the ionization was complete.

The computational implications

“What this allows us to do is to run one really expensive N-body simulation, on the order of tens or hundreds of millions of CPU hours, and just do that once,” Mutch explained. “And then we can keep running our semi-analytic model over the top of that.” The semi-analytic model, he said, took only on the order of ten CPU hours. “And that’s where things start to get really powerful.”

“What that means is we’re no longer restricted to running one really big cosmological hydrodynamic simulation once every few years and needing a large grant and a whole supercomputer to do it,” Mutch said. “Instead, we can start to explore what happens when we change different parameters in our galaxy model, and we can then see how that changes the signal from reionization.”

This capacity for rapid iteration leaves the researchers well-positioned to be able to interpret ionization results as near-future high-power telescopes like the Square Kilometre Array (which is under construction in South Africa and Australia) begin to provide large amounts of data on the radio signals produced by reionization. “So that way, when we measure the ionized bubbles,” Mutch said, “we can infer something about the galaxies.”

Mutch is also taking part in the Genesis simulations under the government-funded ASTRO 3D program. The 30 researchers under Genesis are preparing to run a “big box” simulation – half a billion light years on each side – with 80003 particles inside of it. They expect the model to be “competitive on an international scale,” Mutch said.

To conduct the reionization simulations and the Genesis simulations, Mutch and his colleagues turned to homegrown supercomputing power. Initially, they were using NCI Australia’s Raijin supercomputer, an Intel-based system delivering 1.7 Linpack petaflops that barely squeaked into the most recent Top500 list. Raijin, however, is now being decommissioned, and the researchers are helping to stress test its replacement: Gadi. 

The Gadi supercomputer. Image courtesy of NCI.

While not yet complete, Gadi will boast 3,000 Cascade Lake nodes with two 24-core CPUs and 192 GB of memory, 160 nodes with four Nvidia V100 GPUs and 50 large memory nodes with 1.5 TB of memory. Already, Gadi’s first phase – installed in 2019 – is delivering 4.4 Linpack petaflops, placing it 47th in its first appearance on the Top500 list.

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!

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…

Royalty-free stock illustration ID: 1938746143

MosaicML, Led by Naveen Rao, Comes Out of Stealth Aiming to Ease Model Training

October 15, 2021

With more and more enterprises turning to AI for a myriad of tasks, companies quickly find out that training AI models is expensive, difficult and time-consuming. Finding a new approach to deal with those cascading challenges is the aim of a new startup, MosaicML, that just came out of stealth... Read more…

NSF Awards $11M to SDSC, MIT and Univ. of Oregon to Secure the Internet

October 14, 2021

From a security standpoint, the internet is a problem. The infrastructure developed decades ago has cracked, leaked and been patched up innumerable times, leaving vulnerabilities that are difficult to address due to cost Read more…

SC21 Announces Science and Beyond Plenary: the Intersection of Ethics and HPC

October 13, 2021

The Intersection of Ethics and HPC will be the guiding topic of SC21's Science & Beyond plenary, inspired by the event tagline of the same name. The evening event will be moderated by Daniel Reed with panelists Crist Read more…

Quantum Workforce – NSTC Report Highlights Need for International Talent

October 13, 2021

Attracting and training the needed quantum workforce to fuel the ongoing quantum information sciences (QIS) revolution is a hot topic these days. Last week, the U.S. National Science and Technology Council issued a report – The Role of International Talent in Quantum Information Science... Read more…

AWS Solution Channel

Cost optimizing Ansys LS-Dyna on AWS

Organizations migrate their high performance computing (HPC) workloads from on-premises infrastructure to Amazon Web Services (AWS) for advantages such as high availability, elastic capacity, latest processors, storage, and networking technologies; Read more…

Eni Returns to HPE for ‘HPC4’ Refresh via GreenLake

October 13, 2021

Italian energy company Eni is upgrading its HPC4 system with new gear from HPE that will be installed in Eni’s Green Data Center in Ferrera Erbognone (a province in Pavia, Italy), and delivered “as-a-service” via H 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…

Royalty-free stock illustration ID: 1938746143

MosaicML, Led by Naveen Rao, Comes Out of Stealth Aiming to Ease Model Training

October 15, 2021

With more and more enterprises turning to AI for a myriad of tasks, companies quickly find out that training AI models is expensive, difficult and time-consuming. Finding a new approach to deal with those cascading challenges is the aim of a new startup, MosaicML, that just came out of stealth... Read more…

Quantum Workforce – NSTC Report Highlights Need for International Talent

October 13, 2021

Attracting and training the needed quantum workforce to fuel the ongoing quantum information sciences (QIS) revolution is a hot topic these days. Last week, the U.S. National Science and Technology Council issued a report – The Role of International Talent in Quantum Information Science... Read more…

Eni Returns to HPE for ‘HPC4’ Refresh via GreenLake

October 13, 2021

Italian energy company Eni is upgrading its HPC4 system with new gear from HPE that will be installed in Eni’s Green Data Center in Ferrera Erbognone (a provi Read more…

The Blueprint for the National Strategic Computing Reserve

October 12, 2021

Over the last year, the HPC community has been buzzing with the possibility of a National Strategic Computing Reserve (NSCR). An in-utero brainchild of the COVID-19 High-Performance Computing Consortium, an NSCR would serve as a Merchant Marine for urgent computing... Read more…

UCLA Researchers Report Largest Chiplet Design and Early Prototyping

October 12, 2021

What’s the best path forward for large-scale chip/system integration? Good question. Cerebras has set a high bar with its wafer scale engine 2 (WSE-2); it has 2.6 trillion transistors, including 850,000 cores, and was fabricated using TSMC’s 7nm process on a roughly 8” x 8” silicon footprint. Read more…

What’s Next for EuroHPC: an Interview with EuroHPC Exec. Dir. Anders Dam Jensen

October 7, 2021

One year after taking the post as executive director of the EuroHPC JU, Anders Dam Jensen reviews the project's accomplishments and details what's ahead as EuroHPC's operating period has now been extended out to the year 2027. Read more…

University of Bath Unveils Janus, an Azure-Based Cloud HPC Environment

October 6, 2021

The University of Bath is upgrading its HPC infrastructure, which it says “supports a growing and wide range of research activities across the University.” Read more…

Ahead of ‘Dojo,’ Tesla Reveals Its Massive Precursor Supercomputer

June 22, 2021

In spring 2019, Tesla made cryptic reference to a project called Dojo, a “super-powerful training computer” for video data processing. Then, in summer 2020, Tesla CEO Elon Musk tweeted: “Tesla is developing a [neural network] training computer... Read more…

Enter Dojo: Tesla Reveals Design for Modular Supercomputer & D1 Chip

August 20, 2021

Two months ago, Tesla revealed a massive GPU cluster that it said was “roughly the number five supercomputer in the world,” and which was just a precursor to Tesla’s real supercomputing moonshot: the long-rumored, little-detailed Dojo system. Read more…

Esperanto, Silicon in Hand, Champions the Efficiency of Its 1,092-Core RISC-V Chip

August 27, 2021

Esperanto Technologies made waves last December when it announced ET-SoC-1, a new RISC-V-based chip aimed at machine learning that packed nearly 1,100 cores onto a package small enough to fit six times over on a single PCIe card. Now, Esperanto is back, silicon in-hand and taking aim... Read more…

CentOS Replacement Rocky Linux Is Now in GA and Under Independent Control

June 21, 2021

The Rocky Enterprise Software Foundation (RESF) is announcing the general availability of Rocky Linux, release 8.4, designed as a drop-in replacement for the soon-to-be discontinued CentOS. The GA release is launching six-and-a-half months... 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…

Intel Completes LLVM Adoption; Will End Updates to Classic C/C++ Compilers in Future

August 10, 2021

Intel reported in a blog this week that its adoption of the open source LLVM architecture for Intel’s C/C++ compiler is complete. The transition is part of In Read more…

Hot Chips: Here Come the DPUs and IPUs from Arm, Nvidia and Intel

August 25, 2021

The emergence of data processing units (DPU) and infrastructure processing units (IPU) as potentially important pieces in cloud and datacenter architectures was Read more…

AMD-Xilinx Deal Gains UK, EU Approvals — China’s Decision Still Pending

July 1, 2021

AMD’s planned acquisition of FPGA maker Xilinx is now in the hands of Chinese regulators after needed antitrust approvals for the $35 billion deal were receiv Read more…

Leading Solution Providers

Contributors

HPE Wins $2B GreenLake HPC-as-a-Service Deal with NSA

September 1, 2021

In the heated, oft-contentious, government IT space, HPE has won a massive $2 billion contract to provide HPC and AI services to the United States’ National Security Agency (NSA). Following on the heels of the now-canceled $10 billion JEDI contract (reissued as JWCC) and a $10 billion... 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…

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…

Quantum Roundup: IBM, Rigetti, Phasecraft, Oxford QC, China, and More

July 13, 2021

IBM yesterday announced a proof for a quantum ML algorithm. A week ago, it unveiled a new topology for its quantum processors. Last Friday, the Technical Univer 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…

Frontier to Meet 20MW Exascale Power Target Set by DARPA in 2008

July 14, 2021

After more than a decade of planning, the United States’ first exascale computer, Frontier, is set to arrive at Oak Ridge National Laboratory (ORNL) later this year. Crossing this “1,000x” horizon required overcoming four major challenges: power demand, reliability, extreme parallelism and data movement. Read more…

Intel Unveils New Node Names; Sapphire Rapids Is Now an ‘Intel 7’ CPU

July 27, 2021

What's a preeminent chip company to do when its process node technology lags the competition by (roughly) one generation, but outmoded naming conventions make i Read more…

Intel Launches 10nm ‘Ice Lake’ Datacenter CPU with Up to 40 Cores

April 6, 2021

The wait is over. Today Intel officially launched its 10nm datacenter CPU, the third-generation Intel Xeon Scalable processor, codenamed Ice Lake. With up to 40 Read more…

  • arrow
  • Click Here for More Headlines
  • arrow
HPCwire