SUPERCOMPUTING NEWS SUPERCOMPUTING NEWS
    • MEDIA KIT
    • MOST READ
    • RSS FEED
    • ACADEMIA
    • AEROSPACE
    • APPLICATIONS
    • ASTRONOMY
    • AUTOMOTIVE
    • BIG DATA
    • BIOLOGY
    • CHEMISTRY
    • CLIENTS
    • CLOUD
    • DEFENSE
    • DEVELOPER TOOLS
    • EARTH SCIENCES
    • ECONOMICS
    • ENGINEERING
    • ENTERTAINMENT
    • GAMING
    • GOVERNMENT
    • HEALTH
    • OIL & GAS
    • INDUSTRY
    • INTERCONNECTS
    • MANUFACTURING
    • MIDDLEWARE
    • MOVIES
    • NETWORKS
    • PHYSICS
    • PROCESSORS
    • RETAIL
    • SCIENCE
    • STORAGE
    • SYSTEMS
    • VISUALIZATION
    • AcyMailing subscription form

    • ADD YOUR VIDEOS
    • MANAGE VIDEOS
    • CONVERSATION INBOX
    • SOCIAL ADVERTISER
    • SOCIAL NETWORK VIDEOS
    • SURVEYS
    • GROUPS
    • PAGES
    • MARKETPLACE LISTINGS
    • APPLICATIONS BROWSER
    • PRIVACY CONFIRM REQUEST
    • PRIVACY CREATE REQUEST
    • LEADERBOARD
    • POINTS LISTING
      • BADGES
    • TRADE SHOWS
Sign In
Supercomputing reveals why some black hole flares fade away
Supercomputing reveals why some black hole flares fade away
The next supercomputing breakthrough may come from memory, not compute
The next supercomputing breakthrough may come from memory, not compute
Supercomputing rewrites the timeline of planet formation at cosmic dawn
Supercomputing rewrites the timeline of planet formation at cosmic dawn
Supercomputers reveal four regimes of radiation damage in tungsten
Supercomputers reveal four regimes of radiation damage in tungsten
Computational radiative transfer reveals a gas-ensheathed black hole at cosmic dawn
Computational radiative transfer reveals a gas-ensheathed black hole at cosmic dawn
Supercomputers scan 165 years of weather data to find the ‘Snow-Eater’ heat waves behind Western US snowmelt
Supercomputers scan 165 years of weather data to find the ‘Snow-Eater’ heat waves behind Western US snowmelt
previous arrow
previous arrow
next arrow
next arrow
 
Shadow
Supercomputing reveals why some black hole flares fade away
Featured

Supercomputing reveals why some black hole flares fade away

CHRIS O'NEAL, PUBLISHER August 24, 2026, 8:00 am

Hydrodynamical simulations show that a rapidly spinning star can survive repeated encounters with a supermassive black hole while producing progressively weaker flares, potentially revealing how the star was captured in the first place.

Some black holes exhibit a particularly destructive mechanism for tracking time. When a star ventures too close, gravitational forces strip away its outer layers; the resulting debris falls toward the black hole, generating a brilliant flare. Months or years later, the surviving star may return, initiating the cycle anew.
 
Astronomers have observed a perplexing trend in several of these systems: each successive flare often diminishes in intensity. Recently, researchers at Syracuse University and their collaborators utilized hydrodynamical simulations to identify a potential contributing factor: the star may have been rotating at a high velocity prior to its initial encounter with the black hole.
 
For SuperComputing News, the primary significance of this study extends beyond the potential explanation of an astronomical mystery. It underscores how researchers have employed computational hydrodynamics to simulate an extreme gravitational experiment, one impossible to replicate in a laboratory setting, to discover that a star's evolutionary history may be encoded within the attenuation of its repeated flares.

When a Star Survives the Impossible

A conventional tidal disruption event occurs when a star ventures sufficiently close to a supermassive black hole that the difference in gravitational force across the star overwhelms its self-gravity.

The star is stretched and ultimately destroyed.

Its debris begins falling back toward the black hole, releasing enormous amounts of energy and producing a transient flare that allows astronomers to study an otherwise invisible black hole.

But some stars survive.

In a repeating partial tidal disruption event, or rpTDE, the star loses only part of its mass during each close passage. Its surviving core remains gravitationally bound and returns for another encounter months or years later. 

That makes rpTDEs extraordinarily valuable.

Astronomers effectively get multiple observations of the same star-black-hole interaction.

And that is where the mystery begins.

The Fading-Flare Problem

There are roughly ten known repeating systems of this general type, and about four have displayed progressively dimmer flares. 

At first glance, the explanation seems obvious.

If the star loses less material during each encounter, there should be less material available to produce the next flare.

Less fuel should mean less light.

But previous hydrodynamical simulations produced an unexpected result.

Although the amount of stripped material decreased, the predicted peak flare brightness could remain approximately constant.

Why?

Because the black hole does more than remove mass.

It also spins the star up.

The black hole's tidal field exerts a torque on the surviving stellar core. As the star's rotation increases, stripped material can return toward the black hole on a shorter timescale.

That faster fallback can compensate for the declining amount of material.

The result is surprisingly persistent flare brightness.

The simulations therefore produced a prediction that did not match the progressively fading flares observed in some real systems. 

The researchers needed another variable.

They found it in the star's initial spin.

A Computational Experiment in Stellar Spin

The new study, published in The Astrophysical Journal, tests high-mass main-sequence stars repeatedly disrupted by a 10-million-solar-mass black hole.

Actually, the simulations use a (10^6)-solar-mass supermassive black hole, one million times the mass of the Sun. 

That distinction matters because the computational experiment is deliberately controlled.

The researchers vary the star's initial rotation and examine what happens as it repeatedly passes the black hole.

The simulations show that rapidly rotating, prograde stars, stars whose spin is aligned with their orbital angular momentum, can produce weaker outbursts successively.

The required initial rotation is on the order of tens of percent of the star's breakup speed, the point at which centrifugal forces become strong enough to approach gravitational binding at the stellar surface. 

This is the crucial computational result.

The model finally reproduces the qualitative behavior astronomers have been seeing:

less mass lost → similar fallback timescale → lower peak fallback rate → dimmer flare.

Why Spin Changes the Calculation

The physics is subtle.

Consider a slowly rotating star.

During its first close encounter, the black hole's tidal forces strip material from the star and transfer angular momentum into the surviving core.

The star begins spinning faster.

On subsequent encounters, that additional spin changes the dynamics of the stripped material.

The fallback timescale decreases.

Consequently, even though the star is losing less mass, the material returns more rapidly.

That can preserve the peak fallback rate, and therefore preserve the brightness of subsequent flares.

Now start the experiment with a star that is already rapidly rotating.

There is less room for the black hole to spin it up significantly.

The fallback timescale therefore changes much less from one encounter to the next.

As the star loses progressively less mass, the peak fallback rate declines.

And the flare gets dimmer.

The computational model has effectively identified the missing initial condition required to reproduce the astronomical observations. 

Hydrodynamics at the Extreme

This is precisely the kind of problem for which numerical astrophysics becomes indispensable.

There is no laboratory capable of reproducing a stellar interior being repeatedly distorted by the tidal field of a million-solar-mass black hole.

The researchers instead solve the underlying fluid-dynamical problem computationally.

Their simulations follow the interaction of:

  • stellar structure;
  • self-gravity;
  • the black hole's tidal field;
  • orbital motion;
  • stellar rotation;
  • angular-momentum transfer;
  • mass stripping; and
  • the subsequent fallback of stellar debris.

The current study builds on a broader research program using hydrodynamical simulations to understand repeated stellar mass loss in rpTDEs. Previous work demonstrated that the survivability of a star depends strongly on its internal structure and that high-mass, centrally concentrated stars can survive repeated encounters. 

But mapping every possible combination of stellar mass, structure, orbit and encounter parameters through full hydrodynamic calculations is itself computationally prohibitive.

The researchers have therefore also developed intermediate analytical and hybrid models to explore regions of parameter space that would be impractical to simulate directly. 

That is an important HPC lesson:

The challenge isn't merely running one enormous simulation. It is efficiently exploring the space of possible universes.

The Black Hole Is Also a Stellar Spin-Up Machine

The simulations reveal something counterintuitive.

The black hole is not simply destroying the star.

It is changing the star's internal rotational state.

Every close passage transfers angular momentum.

That means the history of previous encounters affects the outcome of future encounters.

In computational terms, the system has memory.

The initial conditions matter.

The state of the star after encounter one becomes the initial condition for encounter two.

Encounter two changes the state used for encounter three.

And so on.

This is precisely why simple static models are inadequate.

The researchers need a dynamic, evolving computational representation of the star.

A Million-Solar-Mass Laboratory

The simulated black hole has a mass of approximately one million Suns.

The stellar models include main-sequence stars of at least one solar mass, and the calculations examine repeated partial disruptions under different stellar-spin conditions. 

The computational experiment effectively asks:

What happens if we change only the star's rotational state?

That controlled numerical experiment is enormously powerful.

The researchers found that high, prograde initial spins naturally generate the progressively dimmer outbursts seen in observations.

By contrast, the previously modeled spin-up of initially slower stars tends to counteract the declining mass loss.

This provides a physical explanation for why seemingly similar stellar encounters can generate very different flare histories.

The Star's Spin May Reveal Its Past

The story becomes even more interesting when the researchers ask a second question:

Why was the star spinning so rapidly before it ever met the black hole?

The proposed answer is the Hills mechanism.

Imagine two stars orbiting each other in a very tight binary.

The binary wanders too close to a supermassive black hole.

The black hole's enormous tidal field tears the binary apart.

One star is ejected at high velocity.

The other becomes gravitationally captured by the black hole.

This is known as Hills capture.

And there is a crucial consequence.

A close binary can become tidally locked, meaning each star rotates at approximately the same rate that it orbits its companion.

The tighter the binary, the faster that rotation.

Therefore, when the black hole destroys the binary and captures one member, the captured star can enter its new orbit already spinning rapidly. 

The same event could therefore explain two otherwise puzzling properties:

Why is the star spinning so rapidly?

Why is it on such a tight orbit around the black hole?

Supercomputing Connects the Clues

This is where the study becomes particularly compelling from a computational-science perspective.

The simulation isn't simply producing a prettier visualization of a tidal disruption event.

It is connecting multiple physical phenomena:

binary dynamics → stellar rotation → black-hole capture → repeated tidal stripping → angular-momentum transfer → fallback dynamics → flare luminosity.

That is a complex chain of causality.

And numerical modeling makes it possible to follow that chain.

The computer effectively lets researchers rewind the system and ask what initial conditions could have produced the behavior astronomers see today.

From Stellar Spin to Observable Light

One of the most useful aspects of the calculation is the connection between an internal property of a star and an observable astronomical signal.

Astronomers cannot easily measure the star's initial rotation directly.

But they can observe its flares.

That means the computational model creates a bridge:

Initial stellar spin → hydrodynamic interaction → mass stripping → fallback rate → flare brightness.

If the simulated relationship is correct, the light curve itself becomes an indirect probe of stellar rotation.

A fading sequence of flares could therefore reveal something about a star's history long before it encountered the black hole.

The Computational Challenge of Repeating Encounters

A single tidal encounter is already an extreme hydrodynamic problem.

A repeating event is harder.

The star must be evolved through one encounter, allowed to respond internally, placed back onto its orbit and then brought through another close passage.

Its mass, density profile, rotation and internal structure are no longer identical to the previous encounter.

That makes the calculation inherently time-dependent.

The researchers' previous simulations showed that high-mass, centrally concentrated stars can survive relatively small amounts of mass loss and continue through multiple encounters. 

This creates a computational feedback loop:

tidal stripping changes the star → the changed star responds differently to the next tidal encounter.

That is precisely the sort of nonlinear behavior that numerical hydrodynamics is designed to capture.

Why This Matters Beyond One Black Hole

The implications may extend into the center of our own galaxy.

Syracuse researchers point out that Hills capture may also have produced some of the stars orbiting Sagittarius A*, the supermassive black hole at the center of the Milky Way. 

If so, the same dynamical process could help explain both distant repeating tidal-disruption events and some unusual stellar populations in the Galactic Center.

That makes the computational model potentially relevant far beyond the specific systems that motivated the study.

A New Kind of Astronomical Forensics

There is a broader scientific idea here that deserves attention.

Astronomers often think of observations as snapshots of the Universe.

Computational astrophysics can turn those snapshots into forensic evidence.

A fading flare isn't simply a measurement of brightness.

It contains information about:

  • how much stellar material was removed;
  • how quickly that material returned;
  • how the star was rotating;
  • how angular momentum was transferred;
  • how the star's structure changed;
  • and potentially how the star arrived in its orbit.

The simulation allows researchers to decode those clues.

The Supercomputing Lesson

This research illustrates an increasingly important role for HPC in astrophysics.

The breakthrough isn't necessarily a new telescope or a larger detector.

It is the ability to construct a numerical experiment complicated enough to connect microscopic stellar dynamics with macroscopic astronomical observations.

The Universe supplies the event.

The telescope records the light.

The supercomputer works out what had to happen in between.

And in this case, the answer may be that the star was already spinning rapidly when it entered the black hole's deadly orbit.

A Black Hole's Flare as a Computational Fingerprint

The researchers' result offers a striking new interpretation of fading rpTDEs.

The progressively weaker flares may not simply mean that the star is running out of material.

They may be telling us something about the star's rotational history.

A rapidly spinning, prograde star produces the right combination of mass loss and fallback behavior to reproduce the observed decline. 

And that rapid rotation may itself be evidence of a much earlier encounter with a binary companion.

In other words, a black hole flare could carry a fingerprint of a star's life before the star ever met the black hole.

The Universe's Most Extreme Computer Experiment

Recent research from Syracuse University provides a compelling explanation for the phenomenon of fading black hole flares during repeating partial tidal disruption events. While standard models previously suggested that flare brightness should remain relatively constant due to angular momentum transfer, which offsets mass loss by accelerating debris fallback, new hydrodynamical simulations indicate that a star's initial rotation is the decisive factor. 

The study demonstrates that stars beginning their orbit with rapid, prograde rotation possess limited capacity for further spin-up during gravitational encounters. Consequently, as these stars lose mass over successive passages, the lack of an accelerated fallback mechanism leads to a measurable decline in peak flare brightness. These findings suggest that the initial high-speed rotation is likely a byproduct of the Hills mechanism, where a captured star retains the rotational momentum from its former binary companion. By utilizing these advanced computational models, scientists can now effectively bridge the gap between observed light patterns and a star's evolutionary history, using the cadence of fading flares to decode the conditions surrounding the star's initial capture.

The next supercomputing breakthrough may come from memory, not compute
Featured

The next supercomputing breakthrough may come from memory, not compute

CHRIS O'NEAL, PUBLISHER August 20, 2026, 10:00 am

Micron's new U.S. research initiative highlights a fundamental shift in AI and HPC: as accelerators become extraordinarily powerful, the ability to move, store, and feed data efficiently is becoming just as important as raw compute.

The race to build faster supercomputers has traditionally been measured in familiar numbers: FLOPS, accelerator counts, memory capacity, interconnect bandwidth and power consumption.

But the next major performance breakthrough may come from somewhere less glamorous.

Memory.

As artificial intelligence and high-performance computing workloads become increasingly data-intensive, the limiting factor is no longer necessarily how many calculations a processor can perform. Increasingly, it is whether the system can deliver the right data to the processor quickly enough to keep those calculations running.

That makes Micron Technology's announcement of Micron Research Labs, a U.S.-based long-horizon innovation hub, particularly relevant to the future of supercomputing. The initiative is designed to pursue research beyond today's memory products, including new memory devices and materials, advanced architectures, three-dimensional integration, heterogeneous systems, memory-centric computing and storage-class memory. 

For SuperComputing News, the important story isn't simply that Micron is opening another research operation.

The architecture of future supercomputers may increasingly be determined by what happens between the processor and the data.

The memory wall Is becoming a supercomputing problem

A modern accelerator can perform an extraordinary number of operations every second.

But computational throughput is useful only when the processor has data to work on.

This creates one of the fundamental challenges in computer architecture: the memory wall.

Processor performance has historically increased faster than memory latency. Meanwhile, AI workloads have introduced enormous quantities of parameters, activations, intermediate results and cached context that must constantly move through the system.

Micron itself now describes AI system performance as increasingly dependent on memory subsystem performance and capacity, elevating memory from a supporting component to a strategic element of the architecture. 

That shift has profound implications for HPC.

A supercomputer can contain thousands of GPUs, but if those GPUs spend too much time waiting for data, theoretical compute performance becomes increasingly disconnected from delivered application performance.

The question changes from: How many FLOPS can we build?

to: How efficiently can we feed those FLOPS?

AI has made the problem much bigger

Artificial intelligence has accelerated the memory challenge.

Training increasingly large models requires enormous amounts of compute and data.

Inference introduces a different problem: models must respond continuously to users and applications, often while maintaining increasingly large contexts.

Agentic AI pushes the requirements further by maintaining state and performing multiple operations over extended periods.

Micron says that as AI workloads evolve from training toward large-scale inference and agentic systems, memory capacity and bandwidth are becoming increasingly important. 

That matters to supercomputing because many of the same architectural pressures are appearing in scientific AI.

A climate model enhanced by machine learning.

A molecular simulation coupled with an AI surrogate.

A scientific foundation model analyzing astronomical observations.

A digital twin running continuously against real-time sensor data.

All of these workloads depend on moving enormous amounts of information efficiently.

From FLOPS to data movement

The conventional supercomputing race has often centered on floating-point performance.

But real applications rarely achieve theoretical peak performance.

Memory bandwidth, latency, cache behavior, interconnect performance, synchronization, and data locality can determine how much of the processor's theoretical capability actually reaches the scientific application.

This makes memory hierarchy increasingly important.

At one level are registers and caches.

Then comes high-bandwidth memory.

Then system DRAM.

Then increasingly sophisticated storage and data-management layers.

The challenge is to place the right data at the right level at the right time.

That sounds straightforward.

At exascale, it isn't.

HBM is only the beginning

High-bandwidth memory, or HBM, has become a critical technology for AI accelerators because it places large amounts of extremely high-bandwidth memory close to the processor.

Micron's current AI portfolio includes HBM3E and HBM4, alongside DRAM, LPDDR, GDDR and high-performance SSD technologies. 

The company's current HBM4 technology is positioned for next-generation AI data centers, with Micron citing up to 2.8 TB/s of bandwidth per stack. 

But the bigger question is what comes after today's HBM architectures.

That is where long-horizon research becomes important.

Micron Research Labs is intended to investigate technologies beyond current generations, including new materials and devices, advanced architectures, and three-dimensional integration. 

Why 3D memory matters

The physical distance between compute and memory matters.

The farther data must travel, the greater the latency and energy cost.

Three-dimensional integration offers one potential answer by allowing memory and compute technologies to be stacked or integrated more tightly.

Instead of treating the processor and memory as physically separate components communicating across a board, future architectures can increasingly bring them together.

For HPC, that could mean:

  • higher effective bandwidth;
  • lower data-movement latency;
  • improved energy efficiency;
  • greater memory density; and
  • potentially new ways of distributing computation.

The important point is that future performance may come not just from making transistors faster, but from shortening the distance between computation and information.

Memory-centric computing changes the architecture

Micron's research agenda explicitly includes memory-centric computing. 

That phrase deserves attention.

Traditional computer architecture is fundamentally compute-centric.

Data is moved to the processor.

The processor performs an operation.

The result is moved somewhere else.

But moving data can consume substantial energy and bandwidth.

Memory-centric approaches explore architectures in which computation occurs closer to where the data resides, reducing unnecessary movement.

For data-intensive scientific workloads, this could be transformative.

Imagine a simulation processing enormous arrays of data.

Instead of repeatedly moving those arrays between memory and distant processing units, some operations could potentially occur closer to the memory itself.

The result could be less traffic, lower energy consumption and greater effective application performance.

Supercomputing has an energy problem, too

Performance isn't the only issue.

Data movement consumes energy.

As HPC systems scale, energy efficiency becomes increasingly important because operating a massive supercomputer is ultimately constrained by power, cooling and facility infrastructure.

That creates a three-way optimization problem: Compute performance + memory performance + energy efficiency.

A processor that delivers twice the theoretical performance isn't necessarily twice as useful if feeding it requires disproportionately more energy.

Memory technologies therefore have the potential to improve computing efficiency without simply increasing the number of processors.

That could be especially important for future exascale and post-exascale systems.

Storage is moving closer to the compute conversation

The memory hierarchy also extends beyond DRAM and HBM.

Modern AI systems increasingly depend on fast storage for data ingestion, checkpointing, model loading and inference.

Micron's AI portfolio includes high-performance data-center NVMe SSDs designed for these workloads. 

That matters because the distinction between "memory" and "storage" is increasingly becoming an architectural question rather than a simple hardware category.

Large AI models may not fit entirely into the fastest memory.

Scientific datasets can be vastly larger than system memory.

Checkpointing enormous simulations can create substantial I/O loads.

Future systems therefore need intelligent movement of information across the entire hierarchy.

The HPC memory hierarchy of the future

The supercomputer of the future may look less like a collection of CPUs and GPUs connected to memory and more like an integrated data-processing fabric.

At the accelerator:

HBM → extremely high bandwidth

At the node:

DRAM → larger working capacity

Across the system:

network fabric → distributed memory and communication

Below the compute layer:

NVMe and emerging storage → massive datasets and persistent state

And surrounding all of it:

software → deciding where data should live and when it should move.

That final element is critical.

Hardware alone cannot solve the memory problem.

Compilers, runtimes, operating systems and application frameworks will have to understand increasingly complex memory hierarchies.

Micron itself identifies software-driven optimization as an important part of the future memory and storage landscape. 

Research today for systems that may not exist yet

This is where Micron's long-horizon strategy becomes particularly interesting.

The company says the new research organization will focus on technologies that could take years or even decades to reach commercial impact.

That is exactly the kind of research needed for next-generation supercomputing.

Today's systems were shaped by research decisions made years ago.

The architecture of tomorrow's exascale and post-exascale machines is being influenced by research happening now.

Materials scientists, device engineers, computer architects and software researchers are therefore working on problems whose eventual importance may not be obvious from today's products.

The memory system inside a future supercomputer may depend on ideas that are still laboratory experiments today.

The supercomputer is becoming a system of systems

There is a broader lesson here for the HPC community.

The processor can no longer be viewed in isolation.

Neither can memory.

Neither can networking.

Neither can storage.

The performance of a scientific application emerges from the interaction among all of them.

That is why the industry's attention is shifting toward system-level optimization.

Micron has described this explicitly, arguing that AI requires memory and compute to be designed together rather than treated as independent technologies. 

That principle applies equally to HPC.

A different definition of supercomputing performance

Suppose two systems have identical GPUs.

One has significantly better memory bandwidth and data locality.

The other has more powerful theoretical compute but spends more time waiting for data.

Which is the faster supercomputer?

For a real scientific application, the answer may be the first.

This is why benchmarks based solely on peak FLOPS can tell only part of the story.

Researchers increasingly care about time to solution, energy to solution and cost to solution.

Memory performance directly affects all three.

A better memory architecture can therefore make a system effectively more powerful without increasing its nominal compute capability.

Micron's Research bet fits a larger industry shift

Micron is not alone in recognizing the importance of memory.

The broader semiconductor industry is moving toward increasingly heterogeneous architectures in which CPUs, GPUs, specialized accelerators, HBM, networking and storage are engineered together.

Micron's recent work with AI infrastructure partners reflects the same trend. In June, the company announced a strategic agreement with Anthropic spanning memory and storage architecture design, supply and AI infrastructure. 

And Micron's current research agenda includes not only memory devices but architectures capable of supporting future AI and data-intensive computing. 

The direction is unmistakable.

Memory is becoming an architectural differentiator.

The next supercomputing race may be about moving less data

There is an intriguing possibility emerging from all of this.

The next generation of supercomputers may not win primarily by moving data faster.

They may win by moving less data in the first place.

That could mean:

  • computation closer to memory;
  • larger local memory pools;
  • smarter caching;
  • 3D integration;
  • compressed representations;
  • intelligent data placement;
  • memory-aware algorithms;
  • processing-in-memory techniques; and
  • tighter integration between compute, memory and storage.

The objective is simple:

Keep the computation close to the information it needs.

That could become one of the defining principles of post-exascale computing.

From more FLOPS to more useful FLOPS

The history of supercomputing is filled with breathtaking increases in theoretical performance.

But the ultimate goal has never been FLOPS for their own sake.

It is solving scientific problems faster.

If better memory architecture allows a climate simulation, molecular model or AI workload to complete in half the time while consuming less energy, that may be more valuable than simply adding another layer of compute.

This is why Micron's research initiative deserves attention from the HPC community.

It points toward a future in which memory is treated as part of the computing engine itself.

The road ahead

Micron's new research initiative is ultimately a bet on technologies that may define computing long after today's GPUs and accelerators have been replaced.

The company is investing in research spanning new materials and devices, advanced architectures, 3D integration, heterogeneous systems, memory-centric computing and storage-class memory. 

Not all of those technologies will necessarily become mainstream.

Some will fail.

Some will evolve into entirely different technologies.

But that is what long-horizon research is supposed to do: explore possibilities before the market knows which ones it will need.

And the need is becoming increasingly clear.

AI and HPC systems are producing extraordinary amounts of computation.

The next challenge is getting information to that computation efficiently enough to matter.

The future of supercomputing may depend on what happens between the FLOPS

The race for supercomputing supremacy has entered a transformative new phase. For years, the industry’s primary metric was raw computational capacity, how many operations a system could perform per second. Today, however, the focus has shifted toward efficiency: how much useful work can be accomplished per byte moved, per watt consumed, and per dollar invested. This transition places memory directly at the heart of the architectural conversation. 

Micron’s investment in long-horizon memory research is more than just a semiconductor story; it is a fundamental bet on the future of computing architecture. While next-generation supercomputers will undoubtedly feature an unprecedented number of accelerators, those processors will only achieve their true potential if the underlying architecture can reliably supply them with data. In the emerging era of AI and post-exaFLOPS computing, the next major performance breakthrough may not come from building a faster engine, but from building a better, more efficient road to deliver data to that engine. Ultimately, memory is that road.

 

Supercomputing rewrites the timeline of planet formation at cosmic dawn
Featured

Supercomputing rewrites the timeline of planet formation at cosmic dawn

Tyler O'Neal, Staff Editor August 18, 2026, 8:00 am

High-resolution simulations on the Austrian Scientific Cluster show that water-rich planetesimals could have formed around low-mass stars only about 100 million years after the Big Bang, far earlier than the conventional picture of planet formation might suggest.

The first stars did not merely illuminate the young Universe.

According to a new computational study, they may also have begun building planets almost immediately afterward.

Using detailed hydrodynamic simulations, researchers have modeled the evolution of a protoplanetary disk around a low-mass star formed from gas enriched by an earlier Population III pair-instability supernova. The calculations follow the transformation of primordial material into dust, the growth of that dust, the emergence of gravitational structures within the disk and, ultimately, the formation of planetesimals, the building blocks of planets.

The simulations were performed using the FEOSAD numerical framework on the Austrian Scientific Cluster (ASC). Rather than observing an ancient planetary system directly, the researchers have effectively reconstructed its formation computationally.

The result is remarkable: the models produce approximately 6 Earth masses of planetesimals over the simulated disk evolution, with a substantial fraction of the material water-rich.

For SuperComputing News, however, the most compelling aspect is not simply that the early Universe may have formed planets.

It is that high-resolution computation allows scientists to experiment on an era of cosmic history that no telescope can directly revisit.

A Planetary System Before the Solar System Had a Chance to Exist

The Universe was initially dominated by hydrogen and helium.

The heavier elements required for rocky planets, carbon, oxygen, silicon, iron, and others, were manufactured inside stars and distributed into space when those stars died.

That creates an obvious question: How quickly could planet formation begin?

The new simulations investigate one possible pathway.

A massive first-generation star undergoes a pair-instability supernova, enriching its surrounding environment with heavy elements. That material subsequently collapses to form a low-mass protostar and its surrounding disk.

The researchers then follow what happens inside that disk.

The simulation places this process at approximately 100 million years after the Big Bang.

That is extraordinarily early.

Yet the computation suggests that once even a modest amount of heavy elements becomes available, the basic machinery of planet formation may begin operating surprisingly quickly.

The Computer Becomes a Laboratory for Cosmic Dawn

There is no possibility of observing the formation of these particular systems directly.

They existed more than 13 billion years ago.

Instead, researchers must construct a numerical representation of the physical environment and allow the equations governing gas, dust, gravity, and chemistry to determine what happens.

The simulations use FEOSAD, a two-dimensional radiation-hydrodynamics code designed to model the evolution of protoplanetary disks.

The calculation simultaneously follows gas and dust while incorporating gravitational dynamics, heating and cooling, dust evolution, and the conversion of dust into planetesimals.

That combination makes the calculation substantially more than a simple orbital simulation.

It is an evolving multiphysics system.

Gas changes the gravitational environment.

Temperature influences the disk.

Dust grows and migrates.

Dust concentration changes the conditions for gravitational and aerodynamic instabilities.

And those instabilities can ultimately produce planetesimals.

The computer has to keep all of these processes interacting consistently.

Modeling a Cosmic Dawn Disk

The researchers simulate the disk using a 400 × 256 polar grid, following approximately 100,000 years of evolution.

Near the inner boundary, the spatial resolution reaches roughly 0.01 astronomical units.

That resolution is significant because the interesting physics occurs across vastly different spatial scales.

The disk itself extends across astronomical distances, while dust concentration and planetesimal formation involve much smaller structures.

A computational model therefore has to balance physical detail against the enormous cost of resolving the system.

This is one reason high-performance computing is so important to the work.

The simulation was performed on the Austrian Scientific Cluster, providing the computational resources needed to evolve the disk and its coupled physical processes.

From Supernova Debris to a Protostar

The simulation begins with material enriched by a Population III pair-instability supernova.

That material undergoes gravitational collapse.

Approximately 24,000 years after the beginning of the simulated collapse, a protostar forms, followed roughly 1,000 years later by the emergence of its disk.

This sequence is important.

The simulation isn’t simply inserting a mature protoplanetary disk into the early Universe.

It follows the transition toward the disk itself.

Once the disk develops, gravity begins shaping its structure.

By approximately 21,000 years after protostar formation, the model produces prominent spiral structures associated with gravitational instability.

Those spirals become part of the mechanism by which material moves through the disk.

The Chemistry of a Young Planetary System

The simulation also incorporates a chemical network specifically designed for low-metallicity environments.

The model includes 27 reactions involving primordial species such as hydrogen, molecular hydrogen, ionized hydrogen, negative hydrogen ions, deuterium, HD, and electrons.

That chemistry matters because the thermal evolution of the gas affects the dynamics of the disk.

Temperature influences pressure.

Pressure influences gravitational stability.

And temperature and density also influence how dust behaves.

The researchers use numerical root-finding procedures, including Newton-Raphson iteration with bisection fallback, to solve the energy equation within the simulation.

This is a useful reminder that a modern astrophysical simulation is not one equation running on a computer.

It is a tightly coupled numerical system in which chemistry, thermodynamics, radiation, and gravity continually interact.

The Critical Transition: Dust Becomes Planetary Building Material

Planets don’t form directly from a diffuse gas disk.

Small solid particles first have to grow.

Those particles can collide and stick, becoming progressively larger grains.

Eventually, however, another problem emerges.

If particles simply grow and drift inward toward the star, much of the solid material could disappear before becoming planets.

One of the mechanisms that can overcome this problem is the streaming instability.

When solids become sufficiently concentrated relative to the gas, aerodynamic interactions can amplify those concentrations.

The resulting dense regions can collapse into much larger solid body planetesimals.

The simulation explicitly follows the dust evolution and evaluates the conditions under which streaming instability can occur.

This is the computational bridge between microscopic dust grains and the first genuine planetary building blocks.

Six Earth Masses of Planetesimals

The most dramatic result emerges during the later stages of the calculation.

The modeled disk produces approximately six Earth masses of planetesimals before luminosity bursts terminate the planetesimal-formation phase at roughly 37,000 years.

Put another way, the simulation does not merely show that dust can survive around an early low-mass star.

It demonstrates a pathway by which that dust can become a substantial reservoir of solid planetary material.

And it happens astonishingly quickly on cosmic timescales.

The Universe has barely begun its evolution when the computational model is already producing the ingredients for planetary systems.

These May Have Been Water-Rich Worlds

Perhaps the most intriguing aspect is the composition.

The modeled disk is substantially enriched in oxygen-bearing material, and its H₂O mass fraction is only a few times lower than that of the present-day Solar System.

That opens an extraordinary possibility.

Some of the first planetary building blocks in the Universe may not have been dry, primitive rocks.

They could have contained significant amounts of water.

Of course, the simulation does not demonstrate that habitable planets actually formed.

It demonstrates something more fundamental: the physical conditions necessary for producing water-rich planetesimals may have existed remarkably early.

A Computationally Visible Planet-Formation Factory

The simulation provides researchers with something impossible to obtain observationally: a detailed movie of the formation process.

The calculation can be inspected at different times to determine:

  • where gas accumulates;
  • where spiral structures emerge;
  • where dust concentrates;
  • how dust migrates;
  • when gravitational instability develops;
  • where streaming instability becomes possible; and
  • how much planetesimal material ultimately forms.

The simulation therefore acts as a kind of virtual laboratory for planetary formation at cosmic dawn.

Researchers can ask “what if?” questions that cannot be posed observationally.

What happens if the metallicity changes?

What happens if the stellar mass changes?

What happens if the initial disk conditions differ?

What happens to the water fraction?

What happens to the planetesimal mass?

Those experiments can be performed numerically.

Why High-Performance Computing Changes the Question

The important distinction is that the researchers are not using computation merely to process observations.

The simulation itself is producing new scientific knowledge.

Without numerical modeling, there is no direct way to watch a low-metallicity disk evolve for tens of thousands of years while simultaneously tracking its gas, dust, chemistry, gravitational instability, and planetesimal formation.

That makes this a particularly strong supercomputing story.

The computer isn’t an accessory.

It is the experimental apparatus.

A Multiscale Problem in Space and Time

Planet formation is inherently multiscale.

A star forms from material distributed across astronomical distances.

A disk develops spiral structures on scales of fractions of astronomical units to many AU.

Dust grains are microscopic.

The streaming instability concentrates those grains into dense regions.

Planetesimals eventually become kilometer-scale bodies.

The simulation has to represent these processes within one computational framework.

The 400 × 256 grid and approximately 0.01-AU inner resolution provide the numerical resolution needed to follow the relevant disk dynamics while maintaining a computationally manageable domain.

This is exactly the kind of compromise that defines computational astrophysics: enough resolution to capture the physics, enough scale to capture the system.

The First Planetary Systems May Have Been Surprisingly Fast

The findings challenge an intuitive assumption about cosmic evolution.

It is tempting to imagine the early Universe as a chemically primitive place in which planets could not emerge until much later generations of stars had enriched the cosmos.

The simulation presents a more complicated picture.

Once a first generation of massive stars has produced heavy elements, subsequent star-forming environments may acquire enough material for dust and planetesimal formation surprisingly quickly.

The study therefore suggests that the Universe may have begun producing planetary building blocks far earlier than conventional expectations based on later-generation planetary systems might imply.

And the computation puts a timescale on that possibility.

From the First Stars to the First Worlds

There is a beautiful sequence hidden inside the numerical experiment:

First stars → supernova → heavy elements → gravitational collapse → low-mass star → disk → dust → instability → planetesimals.

Each step is connected to the next.

The supernova provides the raw ingredients.

Gravity concentrates them.

The disk organizes them.

Dust evolution converts atomic material into solids.

Instabilities concentrate those solids.

And eventually, planetesimals emerge.

The simulation allows researchers to watch that chain unfold.

The Bigger Supercomputing Story

This research demonstrates why astrophysical simulations are becoming increasingly important as telescopes push farther back toward the beginning of cosmic history.

Observatories such as JWST can reveal ancient galaxies and stars.

But they cannot rewind the Universe and watch those systems form.

Numerical models can.

They allow scientists to reconstruct plausible histories and test whether the laws of physics permit particular structures to emerge under early-Universe conditions.

In this case, the answer appears to be yes.

Planetary building blocks may have emerged almost as soon as the Universe became chemically capable of making them.

Computing a Planetary Future in the Young Universe

There is an inspiring irony to this study: researchers are using humanity’s most advanced computing technology to investigate a period when the universe possessed almost none of the complexity we associate with the modern era. High-performance simulations act as a virtual laboratory, reconstructing the moment when simple atoms began transitioning into stars, protoplanetary disks, and the raw materials for worlds. 

These findings challenge the assumption that planet formation was a late development in cosmic history; instead, it may have been one of the universe’s earliest acts of chemical complexity. The simulations demonstrate that a low-mass star could form just 100 million years after the Big Bang, spawning a disk capable of producing substantial quantities of planetesimals. Furthermore, because these disks were surprisingly rich in water-bearing material, these early building blocks may have been far more dynamic than the dry, primitive rocks one might expect from the early universe. Ultimately, while the first worlds may have formed in the silence of the cosmic dawn, it took the power of a supercomputer to finally bring that process to light.

POPULAR RIGHT NOW
  • AI supercharges the hunt for stronger magnets: Iowa State researchers launch a new era of intelligent materials discovery
    AI supercharges the hunt for stronger magnets: Iowa State researchers launch a new era of intelligent materials discovery
  • IBM's Historic stock collapse raises questions for the future of enterprise supercomputing
    IBM's Historic stock collapse raises questions for the future of enterprise supercomputing
  • Could a novel dark matter theory simultaneously resolve multiple cosmic enigmas? Supercomputer simulations provide a compelling, albeit currently unverified, potential solution
    Could a novel dark matter theory simultaneously resolve multiple cosmic enigmas? Supercomputer simulations provide a compelling, albeit currently unverified, potential solution
  • Melting icebergs may be reshaping Earth’s greatest ocean current
    Melting icebergs may be reshaping Earth’s greatest ocean current
  • Supercomputers replace ‘bathtub’ flood maps with physics-based digital twins of Britain’s coastline
    Supercomputers replace ‘bathtub’ flood maps with physics-based digital twins of Britain’s coastline
  • Supercomputers push neural quantum simulation beyond previous limits
    Supercomputers push neural quantum simulation beyond previous limits
  • Intel’s AI supercomputing revival: Q2 financial surge signals new era for CPU-powered HPC infrastructure
    Intel’s AI supercomputing revival: Q2 financial surge signals new era for CPU-powered HPC infrastructure
  • AI infrastructure financing fears shake semiconductor sector
    AI infrastructure financing fears shake semiconductor sector
  • AWS supercomputing investment reaches historic scale as Amazon’s AI strategy powers record financial results
    AWS supercomputing investment reaches historic scale as Amazon’s AI strategy powers record financial results
  • Catching the wave of the future: Supercomputers unlock the hidden dynamics of Venus
    Catching the wave of the future: Supercomputers unlock the hidden dynamics of Venus
THIS YEAR'S MOST READ
  • Wall Street wants to trade supercomputing power like oil
    Wall Street wants to trade supercomputing power like oil
  • Beamforming the future: BeammWave's 6G push signals the rise of orbital-terrestrial wireless networks
    Joakim Axmon
    Joakim Axmon
  • Cosmic ambition at scale: UK’s supercomputer unlocks a 2.5 petabytes universe
    Cosmic ambition at scale: UK’s supercomputer unlocks a 2.5 petabytes universe
  • Hidden order, revealed at scale: Supercomputing, electron ptychography uncover the inner workings of relaxor ferroelectrics
    Hidden order, revealed at scale: Supercomputing, electron ptychography uncover the inner workings of relaxor ferroelectrics
  • Intel's Q1 results signal supercomputing surge driving Xeon momentum
    Intel's Q1 results signal supercomputing surge driving Xeon momentum
  • When stars fall apart: Supercomputing reveals the hidden physics of black holes
    When stars fall apart: Supercomputing reveals the hidden physics of black holes
  • MIT develops computational framework to probe dark matter via gravitational waves
    MIT develops computational framework to probe dark matter via gravitational waves
  • Explainable AI moves into the watershed: FAMU-FSU engineers build predictive framework for real-time E. coli forecasting
    FAMU-FSU College of Engineering Assistant Professor Nasrin Alamdari. (Scott Holstein/FAMU-FSU College of Engineering)
    FAMU-FSU College of Engineering Assistant Professor Nasrin Alamdari. (Scott Holstein/FAMU-FSU College of Engineering)
  • Japanese scientists decode dolphin speed with supercomputing: Turbulence, vortices, and the hidden physics of propulsion
    Japanese scientists decode dolphin speed with supercomputing: Turbulence, vortices, and the hidden physics of propulsion
  • Multi-layer simulations reveal the hidden supply chain of solar prominences
    Multi-layer simulations reveal the hidden supply chain of solar prominences
MOST READ OF ALL-TIME
  • Largest Computational Biology Simulation Mimics The Ribosome
    Details
    112135
    The amino acid (green) slithers into the chemical reaction center, moving through an evolutionarily ancient corridor of the ribosome (purple). The amino acid is delivered to the reaction core by the transfer RNA molecule (yellow).
    The amino acid (green) slithers into the chemical reaction center, moving through an evolutionarily ancient corridor of the ribosome (purple). The amino acid is delivered to the reaction core by the transfer RNA molecule (yellow).
  • Silicon 'neurons' may add a new dimension to chips
    Details
    81063
    Silicon 'neurons' may add a new dimension to chips
  • Linux Networx Accelerators Expected to Drive up to 4x Price/Performance
    Details
    75572
  • Complex Concepts That Really Add Up
    Details
    73716
    Complex Concepts That Really Add Up
  • Blue Sky Studios Donates Animation SuperComputer to Wesleyan
    Details
    68176
    Each rack holds 52 Angstrom Microsystem-brand “blades,” with a memory footprint of 12 or 24 gigabytes each. (Photos by Olivia Bartlett Drake)
    Each rack holds 52 Angstrom Microsystem-brand “blades,” with a memory footprint of 12 or 24 gigabytes each. (Photos by Olivia Bartlett Drake)
  • Humanities, HPC connect at NERSC
    Details
    57982
  • TeraGrid ’09 'Call for Participation'
    Details
    54986
  • Turbulence responsible for black holes' balancing act
    Details
    52348
  • Cray Wins $52 Million SuperComputer Contract
    Details
    50173
  • SDSC Researchers Accurately Predict Protein Docking
    Details
    46124
  • FRONTPAGE
  • LATEST
  • POPULAR
  • REGISTER
  • SOCIAL
  • VIDEO
  • SUBSCRIPTION
  • RSS
  • GUIDELINES
  • PRIVACY
  • TOS
  • ABOUT
  • +1 (816) 799-4488
  • editorial@supercomputingonline.com
© 2001 - 2026 SuperComputingOnline.com, LLC. All rights reserved. This material may not be published, broadcast, rewritten or redistributed without permission.
Sign In
  • FRONT PAGE
  • LATEST
    • MEDIA KIT
    • MOST READ
    • RSS FEED
    • ACADEMIA
    • AEROSPACE
    • APPLICATIONS
    • ASTRONOMY
    • AUTOMOTIVE
    • BIG DATA
    • BIOLOGY
    • CHEMISTRY
    • CLIENTS
    • CLOUD
    • DEFENSE
    • DEVELOPER TOOLS
    • EARTH SCIENCES
    • ECONOMICS
    • ENGINEERING
    • ENTERTAINMENT
    • HEALTH
    • INDUSTRY
    • INTERCONNECTS
    • GAMING
    • GOVERNMENT
    • MANUFACTURING
    • MIDDLEWARE
    • MOVIES
    • NETWORKS
    • OIL & GAS
    • PHYSICS
    • PROCESSORS
    • RETAIL
    • SCIENCE
    • STORAGE
    • SYSTEMS
    • VISUALIZATION
  • VIDEOS
    • ADD YOUR VIDEOS
    • MANAGE VIDEOS
  • COMMUNITY
    • TRADE SHOWS
    • SOCIAL NETWORK VIDEOS
    • SURVEYS
    • APPLICATIONS BROWSER
    • CONVERSATION INBOX
    • SOCIAL ADVERTISER
    • GROUPS
    • MARKETPLACE LISTINGS
    • PAGES
    • LEADERBOARD
    • POINTS LISTING
      • BADGES
    • PRIVACY CONFIRM REQUEST
    • PRIVACY CREATE REQUEST

Hey there! We noticed you’re using an ad blocker.