Modern computer processors contain many computing units, called cores, which share the same hardware resources. The on-chip interconnect is the component that enables these cores to communicate with each other. But when programs on multiple cores run simultaneously, there is a chance they can delay one another when they use the interconnect to send data across the chip at the same time.

By monitoring and measuring these delays, a malicious agent could conduct what is known as a "side-channel attack" and reconstruct secret information that is stored in a program, such as a cryptographic key or password.

MIT researchers reverse-engineered the on-chip interconnect to study how this kind of attack would be possible. Drawing on their discoveries, they built an analytical model of how traffic flows between the cores on a processor, which they used to design and launch surprisingly effective side-channel attacks. Then they developed two mitigation strategies that enable a user to improve security without making any physical changes to the computer chip.

"A lot of current side-channel defenses are ad hoc -- we see a little bit of leakage here and we patch it. We hope our approach with this analytical model pushes more systematic and robust defenses that eliminate whole classes of attacks at the same time," says co-lead author Miles Dai, MEng '21.

Dai wrote the paper with co-lead author Riccardo Paccagnella, a graduate student at the University of Illinois at Urbana-Champaign; Miguel Gomez-Garcia '22; John McCalpin, a research scientist at Texas Advanced Computing Center; and senior author Mengjia Yan, the Homer A. Burnell Career Development Assistant Professor of Electrical Engineering and Computer Science (EECS) and a member of the Computer Science and Artificial Intelligence Laboratory (CSAIL). The research is being presented at the USENIX Security Conference.

Probing processors

A modern processor is like a two-dimensional grid, with multiple cores laid out in rows and columns. Each core has its own cache where data are stored, and there is also a larger cache that is shared across the entire processor. When a program located on one core needs to access data in a cache that is on another core or in the shared cache, it must use the on-chip interconnect to send this request and retrieve the data.

Though it is a large component of the processor, the on-chip interconnect remains understudied because it is difficult to attack, Dai explains. A hacker needs to launch the attack when traffic from two cores is actually interfering with each other, but since traffic spends so little time in the interconnect, it is difficult to time the attack just right. The interconnect is also complex, and there are multiple paths traffic can take between cores.

To study how traffic flows on the interconnect, the MIT researchers created programs that would intentionally access memory caches located outside their local cores.

"By testing out different situations, trying different placements, and swapping out locations of these programs on the processor, we can understand what the rules are behind traffic flows on the interconnect," Dai says.

They discovered that the interconnect is like a highway, with multiple lanes going in every direction. When two traffic flows collide, the interconnect uses a priority arbitration policy to decide which traffic flow gets to go first. More "important" requests take precedence, like those from programs that are critical to a computer's operations.

Using this information, the researchers built an analytical model of the processor that summarizes how traffic can flow on the interconnect. The model shows which cores would be most vulnerable to a side-channel attack. A core would be more vulnerable if it can be accessed through many different lanes. An attacker could use this information to select the best core to monitor to steal information from a victim program.

"If the attacker understands how the interconnect works, they can set themselves up so the execution of some sensitive code would be observable through interconnect contention. Then they can extract, bit by bit, some secret information, like a cryptographic key," Paccagnella explains.

Effective attacks

When the researchers used this model to launch side-channel attacks, they were surprised by how quickly the attacks worked. They were able to recover full cryptographic keys from two different victim programs.

After studying these attacks, they used their analytical model to design two mitigation mechanisms.

In the first strategy, the system administrator would use the model to identify which cores are most vulnerable to attacks and then schedule sensitive software to run on less vulnerable cores. For the second mitigation strategy, the administrator could reserve cores located around a susceptible program and run only trusted software on those cores.

The researchers found that both mitigation strategies were able to significantly reduce the accuracy of side-channel attacks. Neither requires the user to make any changes to the physical hardware, so the mitigations would be relatively easy to implement, Dai says.

Ultimately, they hope their work inspires more researchers to study the security of on-chip interconnects, Paccagnella says.

"We hope this work highlights how the on-chip interconnect, which is such a large component of computer processors, remains an overlooked attack surface. In the future, as we build systems that have stronger isolation properties, we should not ignore the interconnect," he adds.

This work was funded, in part, by the National Science Foundation and the Air Force Office of Scientific Research.

Story Source:

Materials provided by Massachusetts Institute of Technology. Original written by Adam Zewe. Note: Content may be edited for style and length.

Materials with strong electronic interactions can have applications in energy-efficient electronics. When these materials are placed on a substrate, their electronic properties are changed by charge transfer, strain, and hybridisation.

The study also shows that electric fields and applied strain could be used to 'switch' interacting phases such as magnetism on and off, allowing potential applications in future energy-efficient electronics.

TURNING MAGNETISM ON AND OFF WITH SUBSTRATES

Strong interactions between electrons in materials gives rise to effects such as magnetism and superconductivity. These effects have uses in magnetic memory, spintronics, and quantum computing, making them appealing for emerging technologies.

Last year, another study at Monash discovered strong electronic interactions in a 2D metal-organic framework. The researchers found signatures of magnetism in this material. They showed that this magnetism arose due to strong interactions that were only present when the non-magnetic components were brought together.

This material was grown on a metallic substrate. The substrate was important for the growth and measurement of the material.

Explainer: Metal-organic framework

A crystalline material where organic molecules are connected by metal atoms. Metal-organic frameworks can show many different properties by changing the molecules or metal atoms. Understanding quasiparticle excitations and their interactions is crucial for efforts to control complex materials (such as high-temperature superconductors and topological insulators) that may form the basis of future low-energy electronics and quantum information processing.

"We observed this effect when the material was grown on silver, but not when it was grown on copper, despite them being very similar," says Bernard Field (Monash), co-author of the earlier study and lead author of the current study.

"So that begged the question: Why did the material behave so differently on different substrates?"

The researchers simulated the metal-organic framework on many different substrates to determine under what conditions magnetism could emerge.

They also created a simple model which accurately described the physical phenomena in their atomic-scale simulations. This model allowed the team to quickly and easily explore a wider range of systems with fine control over the important parameters.

Three key variables were found to determine the effect of substrates on electronic interactions: charge transfer, strain, and substrate hybridisation.

• Charge transfer is when a substrate gives or takes electrons from the 2D material. The effect of interactions was strongest when the material had one free electron per molecule.
• Strain is when a substrate stretches or squeezes the 2D material. When the material is stretched, electrons have difficulty moving between molecules and atoms, so they experience local interactions more strongly.
• Hybridisation is when the electronic character of the substrate and the 2D material are mixed due to coupling between them. Metallic substrates often have strong hybridisation, which can suppress magnetism. But insulating substrates, such as atomically-thin hexagonal boron nitride, have very weak hybridisation and preserve the electronic interactions in the material.

With this understanding of what the key variables are, it is possible to consider how to manipulate these variables to control the electronic interactions.

The study showed that an electric field could turn magnetism on and off by changing the charge transfer.

Electric fields are how existing transistors operate. Having electric control of magnetic phases is vital for using these materials in electronic devices.

The study also showed that applied strain could turn magnetism on and off. This could be achieved using piezoelectric materials. It is also an important consideration for flexible electronics.

"The team is continuing to investigate strong interactions in 2D metal-organic frameworks, which provide a rich platform to explore novel quantum physics applied for energy-efficient electronic devices," says corresponding author Prof Nikhil Medhekar (Monash Department of Materials Science and Engineering), who led the study, "We are investigating more advanced methods for simulating strong interactions between electrons."

"This work provides quantitative predictions, using diverse theoretical formalisms, on the electronic properties of low-dimensional nanomaterials on a wide range of substrates and conditions," says co-author A/Prof Agustin Schiffrin (Monash School of Physics and Astronomy), who leads experimental research on these materials, "This can guide future real-world experiments, which is extremely valuable for experimental researchers."

This study was supported through the Australian Research Council (Centre of Excellence and Future Fellowship programs). Resources for numerical calculations were provided by the National Computing Infrastructure (NCI) and Pawsey Supercomputing Centre.

Story Source:

Materials provided by ARC Centre of Excellence in Future Low-Energy Electronics Technologies. Note: Content may be edited for style and length.

Story Source:

Materials provided by Indiana University School of Medicine. Original written by Christina Griffiths. Note: Content may be edited for style and length.

The new measurements from Cygnus X-1, published online by the journal Science on Thursday, Nov. 3, represent the first observations of a mass-accreting black hole from the Imaging X-Ray Polarimetry Explorer (IXPE) mission, an international collaboration between NASA and the Italian Space Agency (ASI). Cygnus X-1 is one of the brightest X-ray sources in our galaxy, consisting of a 21 solar mass black hole in orbit with a 41 solar mass companion star.

"Previous X-ray observations of black holes only measured the arrival direction, arrival time and energy of the X-rays from hot plasma spiraling toward the black holes," said lead author Henric Krawczynski, the Wayman Crow Professor of Physics in Arts & Sciences at Washington University in St. Louis and a faculty fellow in the university's McDonnell Center for the Space Sciences. "IXPE also measures their linear polarization, which carries information about how the X-rays were emitted -- and if, and where, they scatter off material close to the black hole."

No light, not even light from X-rays, can escape from inside the event horizon of a black hole. The X-rays detected with IXPE are emitted by the hot matter, or plasma, in a 2,000-km diameter region surrounding the 60-km diameter event horizon of the black hole.

Combining the IXPE data with concurrent observations from NASA's NICER and NuSTAR X-ray observatories in May and June 2022 allowed the authors to constrain the geometry -- i.e., shape and location -- of the plasma.

The researchers found that the plasma extends perpendicular to a two-sided, pencil-shaped plasma outflow, or jet, imaged in earlier radio observations. The alignment of the direction of the X-ray polarization and the jet lends strong support to the hypothesis that the processes in the X-ray bright region close to the black hole play a crucial role in launching the jet.

The observations match models predicting that the corona of hot plasma either sandwiches the disk of matter spiraling toward the black hole or replaces the inner portion of that disk. The new polarization data rule out models in which the black hole's corona is a narrow plasma column or cone along the jet axis.

The scientists noted that a better understanding of the geometry of the plasma around a black hole can reveal much about the inner workings of black holes and how they accrete mass.

"These new insights will enable improved X-ray studies of how gravity curves space and time close to black holes," Krawczynski said.

Related to the Cygnus X-1 black hole specifically, "IXPE observations reveal that the accretion flow is seen more edge-on than previously thought," explained co-author Michal Dovčiak at the Astronomical Institute of the Czech Academy of Sciences.

"This may be a signature of a misalignment of the equatorial plane of the black hole and the orbital plane of the binary," or the paired duo of the black hole and its companion star, clarified co-author Alexandra Veledina from the University of Turku. "The system may have acquired that misalignment when the black hole progenitor star exploded."

"The IXPE mission uses X-ray mirrors fabricated at NASA's Marshall Space Flight Center and focal plane instrumentation provided by a collaboration of ASI, the National Institute for Astrophysics (INAF) and the National Institute for Nuclear Physics," said co-author Fabio Muleri of INAF-IAPS. "Beyond Cygnus X-1, IXPE is being used to study a wide range of extreme X-ray sources, including mass accreting neutron stars, pulsars and pulsar wind nebulae, supernova remnants, our galactic center and active galactic nuclei. We've found a lot of surprises, and we're having a lot of fun."

A second paper in the same issue of Science was led by Roberto Taverna at the University of Padova and describes the IXPE detection of highly polarized X-rays from the magnetar 4U 0142+61.

"We are thrilled to be part of this new wave of scientific discovery in astrophysics," Krawczynski said.

Using a combination of computational and synthetic approaches, a team from the Max Planck Institute for Terrestrial Microbiology in Marburg, Germany, in collaboration with the University of Singapore has now successfully resurrected and studied billion-year-old enzymes in the lab. In this process, which they describe as "molecular paleontology," the researchers found that instead of direct mutations in the active center, an entirely new component prepared photosynthesis to adapt to rising oxygen levels.

Rubisco's early confusion

Rubisco is ancient: it emerged approximately four billion years ago in primordial metabolism prior to the presence of oxygen on earth. However, with the invention of oxygen-producing photosynthesis and rise of oxygen in the atmosphere, the enzyme started catalyzing an undesired reaction, in which it mistakes O2 for CO2 and produces metabolites that are toxic to the cell. This confused substrate scope still scars Rubiscos to date and limits photosynthetic efficiency. Even though Rubiscos that evolved in oxygen-containing environments became more specific for CO2 over time, none of them could get completely rid of the oxygen capturing reaction.

The molecular determinants of increased CO2 specificity in Rubisco remain largely unknown. However, they are of great interest to researchers aiming to improve photosynthesis. Interestingly, those Rubiscos that show increased CO2 specificity recruited a novel protein component of unknown function. This component was suspected to be involved in increasing CO2 specificity, however, the true reason for its emergence remained difficult to determine because it already evolved billions of years ago.

Studying evolution by resurrecting ancient proteins in the lab

To understand this key event in the evolution of more specific Rubiscos, collaborators at the Max Planck Institute for Terrestrial Microbiology in Marburg and Nanyang Technological University in Singapore used a statistical algorithm to recreate forms of Rubiscos that existed billions of years ago, before oxygen levels began to rise. The team led by Max Planck researchers Tobias Erb and Georg Hochberg resurrected these ancient proteins in the lab to study their properties. In particular, the scientists wondered whether Rubisco's new component had anything to do with the evolution of higher specificity.

The answer was surprising, as doctoral researcher Luca Schulz explains: "We expected the new component to somehow directly exclude oxygen from Rubisco catalytic center. That is not what happened. Instead, this new subunit seems to act as a modulator for evolution: recruitment of the subunit changed the effect that subsequent mutations had on Rubisco's catalytic subunit. Previously inconsequential mutations suddenly had a huge effect on specificity when this new component was present. It seems that having this new subunit completely changed Rubisco's evolutionary potential."

An enzyme's addiction to its new subunit

This function as an "evolutionary modulator" also explains another mysterious aspect of the new protein component: Rubiscos that incorporated it are completely dependent on it, even though other forms of Rubisco can function perfectly well without. The same modulating effect explains why: When bound to this small protein component, Rubisco become tolerant to mutations that would otherwise be catastrophically detrimental. With the accumulation of such mutations, Rubisco effectively became addicted to its new subunit.

Altogether, the findings finally explain the reason why Rubisco kept this new protein component around ever since it encountered it. Max Planck Research Group Leader Georg Hochberg explains: "The fact that this connection was not understood until now highlights the importance of evolutionary analysis for understanding the biochemistry that drives life around us. The history of biomolecules like Rubisco can teach us so much about why they are the way they are today. And there are still so many biochemical phenomena whose evolutionary history we really have no idea about. So it's a very exciting time to be an evolutionary biochemist: almost the entire molecular history of the cell is still waiting to be discovered."

Scientific journeys back in time can provide invaluable insights for the future

The study also has important implications for how photosynthesis might be improved, says Max Planck Director Tobias Erb: "Our research taught us that traditional attempts to improve Rubisco might have been looking in the wrong place: for years, research focused solely on changing amino acids in Rubisco itself to improve it. Our work now suggests that adding entirely new protein components to the enzyme could be more productive and may open otherwise impossible evolutionary paths. This is uncharted land for enzyme engineering."

Story Source: Materials provided by Max-Planck-Gesellschaft. Note: Content may be edited for style and length.

Story Source:

Materials provided by Massachusetts Institute of Technology. Original written by Peter Dizikes. Note: Content may be edited for style and length.

You've probably heard the term machine learning before, but are you familiar with machine teaching? Machine learning is what happens behind the scenes when a computer uses input data to form models that can later be used to perform useful functions. But machine teaching is the somewhat less explored part of the process, of how the computer gets its input data to begin with. In the case of visual systems, for example ones that can recognize objects, people need to show objects to a computer so it can learn about them. But there are drawbacks to the ways this is typically done that researchers from the University of Tokyo's Interactive Intelligent Systems Laboratory sought to improve.

"In a typical object training scenario, people can hold an object up to a camera and move it around so a computer can analyze it from all angles to build up a model," said graduate student Zhongyi Zhou. "However, machines lack our evolved ability to isolate objects from their environments, so the models they make can inadvertently include unnecessary information from the backgrounds of the training images. This often means users must spend time refining the generated models, which can be a rather technical and time-consuming task. We thought there must be a better way of doing this that's better for both users and computers, and with our new system, LookHere, I believe we have found it."

Zhou, working with Associate Professor Koji Yatani, created LookHere to address two fundamental problems in machine teaching: firstly, the problem of teaching efficiency, aiming to minimize the users' time, and required technical knowledge. And secondly, of learning efficiency -- how to ensure better learning data for machines to create models from. LookHere achieves these by doing something novel and surprisingly intuitive. It incorporates the hand gestures of users into the way an image is processed before the machine incorporates it into its model, known as HuTics. For example, a user can point to or present an object to the camera in a way that emphasizes its significance compared to the other elements in the scene. This is exactly how people might show objects to each other. And by eliminating extraneous details, thanks to the added emphasis to what's actually important in the image, the computer gains better input data for its models.

"The idea is quite straightforward, but the implementation was very challenging," said Zhou. "Everyone is different and there is no standard set of hand gestures. So, we first collected 2,040 example videos of 170 people presenting objects to the camera into HuTics. These assets were annotated to mark what was part of the object and what parts of the image were just the person's hands. LookHere was trained with HuTics, and when compared to other object recognition approaches, can better determine what parts of an incoming image should be used to build its models. To make sure it's as accessible as possible, users can use their smartphones to work with LookHere and the actual processing is done on remote servers. We also released our source code and data set so that others can build upon it if they wish."

Factoring in the reduced demand on users' time that LookHere affords people, Zhou and Yatani found that it can build models up to 14 times faster than some existing systems. At present, LookHere deals with teaching machines about physical objects and it uses exclusively visual data for input. But in theory, the concept can be expanded to use other kinds of input data such as sound or scientific data. And models made from that data would benefit from similar improvements in accuracy too.