A moisture-absorbing polymer-based composite, or hydrogel, is set to enhance solar panel efficiency and lifespan, offering a scalable, low-cost solution for hot and humid environments. The material, developed at KAUST, absorbs ambient moisture overnight and provides cooling by slowly releasing water throughout the day[1].
Solar panels are central to the clean energy transition, accounting for most renewable additions worldwide and reducing nearly 1.5 billion tons of carbon dioxide emissions annually. Discoveries of new materials and improved manufacturing techniques are crucial for this continued progress, as they enhance solar panel performance.
A key challenge for solar panels is that, alongside converting sunlight into electricity, they also absorb light as heat. This thermal buildup raises panel temperatures, reducing power output and shortening operational lifespan. Existing cooling systems designed to manage these effects often rely on external power sources to circulate water or air — an approach that is both energy-intensive and costly. These systems also require frequent maintenance.
To design a cheaper and greener alternative, a team led by Qiaoqiang Gan and postdoc Saichao Dang has created a hydrogel layer that uses natural evaporative cooling. Attached to the rear of solar panels, the layer operates autonomously, requiring no electricity or maintenance.
The hydrogel consists of lithium chloride salt embedded in a cross-linked polymer network formed by sodium polyacrylate. Through this microporous polymer matrix, the salt helps pull in moisture from the air. The polymer network contains tiny pores that trap water molecules and includes hydrophilic carboxylate groups, which further enhance water storage capacity.
“Each component plays a role,” says Dang. “By adjusting their proportions, we found a sweet spot where the gel can hold enough water and release it slowly throughout the day.”
Researchers fabricated the hydrogel using a straightforward process. They stirred the polymer powder into the salt solution for three minutes, poured the mixture into a mold, flattened it, and allowed it to cure at room temperature for one hour.
The hydrogel demonstrated sustained cooling and strong water uptake. In a long-term outdoor test at temperatures ranging from 25 C to 41 C and relative humidities of 31 to 91 percent, it reversibly absorbed and released water over 21 days without failure. At 38 C, it achieved a record temperature drop of 14.1 C, boosting power conversion efficiency by 12.9 percent.
“We expected a slow release but were surprised by how steady and how long the cooling lasted — even under strong sunlight for 10 hours,” Dang says.
Ongoing performance assessments under extremely hot and arid conditions at King Abdulaziz City for Science and Technology (KACST) in Riyadh help demonstrate the hydrogel’s robustness in a real-world desert environment, which mirrors the conditions found at many solar farm sites worldwide, notes Gan.
The cooling system is expected to extend solar panel lifespan by more than 200 percent and reduce the levelized cost of electricity by 18 percent — a significant economic advantage for both residential and utility-scale installations. Gan adds: “We are exploring commercialization pathways to deploy this system in operational solar farms.”
The year is 2050. Sustainably sourced fish and seafood have replaced 70 percent of red meat consumption worldwide, and seaweed is a staple vegetable for millions. Food waste has been reduced by 75 percent thanks to ambitious new laws and behavioral change, and half of all land degraded by unsustainable agricultural practices has been successfully restored.
Is this a pipe dream or a feasible future reality?
These goals are reviewed by an international, interdisciplinary team of researchers, led by KAUST scientists. The team outlines how to meet the proposed targets and achieve the goals of the Rio Conventions by “bending the curve” of land degradation and transforming global food systems[1].
“Land lies at the heart of everything we depend on. It feeds our communities, supports rich ecosystems, and helps keep the climate in balance. But the way we’re degrading land today puts all of that at risk,” says KAUST’s Fernando Maestre. “This damage doesn’t happen in isolation. It fuels a chain reaction of growing global challenges — from food and water shortages to displacement, social unrest, and deepening inequality.”
Feeding a global population of more than eight billion is placing extreme pressure on Earth’s land and aquatic ecosystems. The three Rio Conventions, agreed at the Earth Summit in Rio de Janeiro in 1992, advocate for balanced global systems that protect and sustain the natural world.
While preventing and reversing land degradation is a key goal of the U.N. Convention to Combat Desertification (UNCCD), continued population growth means that global food production will need to increase by an estimated 35-56 percent by 2050.
“Tackling the land crisis while ensuring sustainable food for all demands a strategic, integrated approach that connects environmental health with how we grow and consume food,” Maestre notes. “We need clear targets and actionable solutions.”
The team calls for restoring half of all degraded farmland through sustainable land management practices to reverse decades of soil erosion, nutrient depletion, and the intensive use of machinery, pesticides, and fertilizers — particularly on large industrial farms.
However, smallholdings and family farms, which make up the majority of farms worldwide, often rely on local traditions to keep their land viable. Supporting these farms to diversify, build resilience, and adopt sustainable land management practices at scale is essential. “Empowering smallholder farmers with secure land rights, fair market access, and modern, sustainable agricultural tools can significantly boost their productivity and income,” says Maestre.
The team also proposes restoring half of all degraded non-agricultural land — 9.87 million square kilometers — by 2050. This will require equitable and inclusive engagement with all stakeholders, along with the integration of scientific, traditional, and local knowledge.
With global diets projected to require a shift away from carbon-intensive red meats and processed foods by 2050, the research team recommends substituting 70 percent of red meat intake with fish and seafood, and replacing 10 percent of vegetable consumption with seaweed.
Integrating marine and land-based food production requires careful consideration to avoid transferring environmental pressures from land to sea. Supporting this transition will require financial incentives, investment in infrastructure and transportation of marine foodstuffs, and technological support for local communities. The researchers acknowledge that meat-rich diets should be retained in lower-income countries and for specific population groups to ensure overall health and nutrition.
Currently, one-third of all food produced each year is wasted, representing US$1 trillion in losses and the use of 1.4 billion hectares of land. To reduce food waste by 75 percent by 2050, the KAUST-led team calls for coordinated efforts across both production and consumption. These include banning contracts that require aesthetically pleasing fruit and vegetables, investing in long-term storage solutions, and encouraging farmers to grow crops suited to local environmental conditions.
Maestre adds: “To tackle the land crisis and feed a growing population, we need a united, strategic approach that connects how we use land with how we produce food. Strengthening cooperation between global environmental agreements should drive bold action.”
Miniature LEDs called micro-LEDs have been shown to generate random numbers at gigabit-per-second speeds by a team of researchers from Saudi Arabia and the United States[1].
The generation of random numbers is vital for many tasks, including data security — where it is used to create encryption keys and passwords — and computer simulations of complex systems such as the weather and financial markets.
There is, therefore, a strong demand to develop cost-effective random number generators that are small enough for chip-scale integration while also offering a fast generation rate.
The most robust and reliable way to generate true random numbers is to sample and digitize a physical process underpinned by the intrinsic randomness of quantum mechanics. For example, the thermal noise, chaos, and jitter from electronic and optoelectronic devices have all been investigated in the past.
Now, Heming Lin, Boon Ooi, and coworkers from KAUST, King Abdulaziz City for Science and Technology (KACST), and the University of California at Santa Barbara report that intensity fluctuations in the spontaneous emission from blue GaN micro-LEDs, ranging in size from 5-100 μm, can serve as a quantum random number generator (QRNG) with an ultra-high generation rate of 9.375 Gbit/s.
“Micro-LEDs are compact, reliable, and cost-effective,” say Lin and Ooi. “They consume less power and require simpler electronic and photonic system architectures than other competing technologies.”
The idea of using LEDs to generate numbers is not new. Over the past decade, research teams have explored measuring photon number and arrival time. However, a major limitation of these previous schemes is that they have provided much slower generation rates, typically on the scale of no more than a few hundred megabits per second.
“Systems relying on single-photon detection typically extract only two bits per sampling cycle, whereas our system achieves six bits by leveraging intensity fluctuations,” explain Lin and Ooi.
Importantly, for any QRNG to be trusted, its output must be stringently tested to ensure it is sufficiently random. The tests developed by the U.S. National Institute of Standards and Technology (NIST) are the gold standard. The KAUST team tested a variety of micro-LEDs with different sizes — spanning from 5 × 5 μm² to 100 × 100 μm² — and drive currents ranging from 0.5 to 100 mA. All passed the NIST tests.
The team’s future work will focus on boosting generation rates by creating 2D arrays of micro-LEDs that enable parallel random number generation.
The researchers are also planning to create a fully integrated system, rather than using discrete components. At present, the KAUST system comprises a GaN micro-LED, which is temperature stabilized using a thermoelectric cooler and has its light emission fed to an avalanche photodetector. This, in turn, is connected to a sampling oscilloscope via an electronic amplifier.
Lin and Ooi add:“Our next step is to integrate an on-chip photodetector with the micro-LED and subsequently incorporate all the required electronic components to realize a fully integrated QRNG chip.”
From soft robotic fingers that can gently grasp and release delicate objects on demand to luminescent water quality monitors that dim to signal the presence of contaminants, numerous smart devices could be derived from a rapidly developing new family of stimuli-responsive molecular materials called porous cages.
“These stimuli-responsive molecular entities enable us to make smart materials that can automatically respond to cues or changes in their environment,” says Niveen Khashab from the Smart Hybrid Materials Lab at KAUST. As a pioneer in stimuli-responsive porous cage research, Khashab was invited to share her expert perspective on this rapidly emerging field[1].
“Porous cages are discrete, hollow molecular constructs that can accept small molecule ‘guests’ within the cavity at their core,” she explains. Hosting a guest molecule triggers structural and property changes in the porous cage, enabling useful functions such as movement or color change. “Our key contribution to the field has been the design and synthesis of molecular hosts capable of recognizing a wide range of guest molecules, and thus responding to changes at the molecular level.”
Khashab adds that this work has led to smart gels and pastes with applications ranging from smart agriculture to wound healing.
Unlike other guest-hosting materials such as metal-organic frameworks (MOFs), a key advantage of porous cages is that they dissolve easily in organic solvents, which enables their ready incorporation into various devices. In the dissolved state, porous cages can be coated onto surfaces, mixed into other materials such as plastics, or directly cross-linked to create smart materials with a dynamic, stimuli-responsive coating or core.
The potential to develop sophisticated devices from such materials drew Khashab’s two co-authors, postdoctoral researchers Peiren Liu and Fang Fang, to join her lab. Fang says: “What intrigued both of us was the possibility of integrating molecular cage host–guest systems with polymers, thereby translating molecular-level responsiveness into tangible, macroscopic functional behaviors.”
Liu and Fang co-led the lab work on the team’s latest contribution to the field, novel urea porous cages, which they combined with a polymer to create a stimuli-responsive film. When exposed to organic vapors, the two ends of the film curled up toward each other as guest molecules from the vapor were absorbed by the film. The team used this material to build soft robotic fingers that grasp and release objects in response to specific vapor cues.
The KAUST team’s perspective article highlighted advances from across the field. These developments included porous cage membranes with switchable permeability for chemical purification applications, as well as porous cages decorated with a light-emitting sidechain that underwent a shift in light emission along with the absorption of specific guest molecules — such as common contaminants in water supplies.
“The potential applications of porous cages are diverse,” Liu says. “I think the most exciting real-world uses for such materials lie in soft robotics — particularly in developing artificial muscles — and in wearable devices for environmental monitoring and healthcare.”
When water drains from the bottom of a vertical tube, it is followed by a thin film of liquid that can adopt complex and beautiful shapes. KAUST researchers have now studied exactly how these “fluted films” form and break up, developing a mathematical model of their behavior that could help improve the performance, safety, and efficiency of industrial processes[1]
“At first glance, water draining from a tube seems like an everyday process driven by gravity,” says Abhijit Kushwaha, a member of the team behind the work. “It is only with high-speed imaging that we can slow down time enough to capture the hidden choreography of this process.”
For the study, the team used hollow tubes of varying diameters, filled with water to different heights. As the researchers allowed the water to flow out, a high-speed camera captured the shapes formed over a period of about a hundred milliseconds.
This revealed a curious effect for certain combinations of tube diameter and water height. As the liquid fell, a thin film of water dragged against the tube walls and descended more slowly. Once the main water column exited the tube, this film emerged and formed a fleeting, tulip-shaped bubble. In some cases, the fluted film quickly retracted into the tube; in others, it stretched until the water column broke away from it.
The formation of fluted films depends on a delicate balance of gravity, surface tension, inertia, and viscosity, explains Kushwaha. If the water column is too short or the tube is too narrow, the film does not form. Conversely, the widest tubes produce a cylindrical film that breaks away from the tube to create a crown shape.
The researchers created a mathematical model to predict the behavior of these films based on a few simple parameters, such as tube radius and water height. “This can inform better design and control strategies in any system where thin liquid films play a vital role — from industrial reactors to microelectronics to biological systems, such as the lungs,” explains Tadd Truscott, who leads the research.
For example, devices called falling-film evaporators are widely used in industries like food processing, pharmaceuticals, and power generation to concentrate liquids or remove solvents. These systems feature thin films of liquid that evaporate as they flow down the walls of heated tubes. If these films break or become uneven, heat transfer efficiency can be reduced, or equipment can be damaged.
“Our research helps improve understanding of when and how such films might rupture or behave unexpectedly, offering insights that could be used to design more reliable systems,” Truscott says. “This could also be relevant to cooling rocket engines or applying protective coatings to surfaces.”
The team plans to study how other fluids behave in a broader range of tubes. “Ultimately, our goal is to develop a predictive framework that helps scientists and engineers understand, design, and optimize systems where thin films play a hidden but crucial role,” Kushwaha adds.
KAUST bioengineers have developed a flexible optoelectronic patch, or ePatch, that is worn on a patient’s skin and can continuously monitor blood pressure without the need for compressible cuffs or wired devices [1].
Continuous blood pressure monitoring could offer significant benefits across multiple health conditions, helping both patients and medical professionals. Deterioration in conditions such as cardiovascular disease and diabetes could be caught and treated more quickly if there were a comfortable, real-time method of monitoring blood pressure throughout patients’ daily lives.
“Wearable electronics enable 24/7 monitoring and deliver comprehensive data for health analysis without patients needing to attend multiple medical appointments,” says Yizhou Zhong, who worked on the project under the supervision of KAUST’s Sahika Inal as part of an international team from KAUST that included scientists from the United Kingdom, United States, and Spain.
The thin-film patch integrates multiple components to provide accurate blood pressure readings. Its robust polymer substrate houses various functional elements, including an organic electrochemical transistor, an organic photodiode, LEDs, and biosensor electrodes.
“Our breakthrough moment was achieving simultaneous measurement of two critical types of physiological information in a combined hybrid signal,” says Zhong.
Existing blood pressure monitoring relies on electrocardiography (ECG), which measures heart rhythm and rate. At the same time, blood volume changes in microvascular tissues are determined using photoplethysmography (PPG), which involves shining light into the skin and measuring the amount of light transmitted or reflected back. This enables calculation of blood transit time from the aortic valve to peripheral sites. The new ePatch integrates these two signals into a single hybrid signal, known as an electrocardio-photoplethysmogram (EC-PPG).
“The transistor collects signals from both the PPG-recording photodiodes and the ECG electrodes, combines them, and then amplifies the output so that we can resolve the signals properly. This is unique to our design,” says Zhong. “The EC-PPG data is then analyzed externally via a deep learning model to estimate blood pressure.”
“This is a tool with very exciting potential,” adds Inal. “Our approach enables continuous monitoring at 10-second intervals and outperforms existing dual-signal approaches. Separately, biosensor electrodes in the ePatch also measure sweat ion and glucose concentrations — vital signs unrelated to blood pressure but useful metrics in cardiovascular health assessment.”
The ePatch also reduces hardware expenses, and computational demands are low because the deep learning models are trained on a single hybrid signal. However, the manufacturing process is currently time-consuming, so the researchers need to optimize it before the patch can be mass-produced.
The team is planning several ePatch improvements as it moves into prototype trials. Further miniaturization and encapsulation will enhance user comfort and durability, while integrating a compact, efficient power source will support long-term, self-powered operation.
“We also hope to incorporate a reliable wireless communication interface that will improve real-time data transmission and offer consistent remote monitoring capabilities,” concludes Zhong.
Stylolites — irregular seams that occur in limestone — have been found to affect how acoustic waves move through rock samples. Laboratory-based insights from KAUST researchers offer an improved understanding of how these features impact acoustic imaging techniques, which are used to analyze induced microseismic events during hydraulic fracturing[1].
Carbonate-based sedimentary rocks like limestone often hold gas and oil reserves within their layers. Researchers commonly use sound (acoustic) waves to interrogate subsurface rocks and identify rock types, reservoir size, and internal sedimentary or structural features that influence fluid flow.
“Sedimentary rock layers are rarely uniform. Stylolites, for example, are serrated discontinuities that run through carbonate rock and result in visible, jagged ‘boundary layers,’ often at oblique angles to bedding,” says Thomas Finkbeiner, who led the study in collaboration with colleagues and former KAUST postdoc Bing Yang from Three Gorges University in Yichang, China.
Stylolites mark dissolution surfaces where minerals from the host rock have been dissolved by large overburden stresses. The resulting boundary consists of reprecipitated, insoluble material, such as clay. Due to their mechanical contrast with the host rock, these discontinuities may disrupt sound waves as they pass through.
The finding came from a stroke of luck for the researchers. “We were using limestone blocks for another experimental lab study when we noticed that stylolites were present in our samples,” says Finkbeiner. “This inspired us to investigate their physical properties in more detail and find out how they influence acoustic wave propagation at the lab scale. Few studies have explored stylolites from this angle before.”
The team imaged the stylolites using X-ray tomography equipment to gather data on their three-dimensional morphologies and characterize their dimensions.
“Imaging these stylolites was tricky because they were rather thin and had geometrically very irregular surfaces,” notes Finkbeiner. “Also, to better understand how their mechanical properties contrast with the ambient host rock, we had to open up our rock specimens with a saw, chisel, and hammer to access the stylolites and measure their hardness.”
The researchers recorded acoustic wave velocities and amplitudes passing through the rock samples. They fed the acquired data into a computer model that simulated sound wave propagation through the rocks at frequencies appropriate for lab-scale specimens.
The results showed that stylolites are weak discontinuities that exhibit minimal influence on the first arrivals of transmitted acoustic waveforms. However, they significantly affect coda waves — secondary waves that form due to scattering from small-scale variations. This impacts the overall soundwave energy transmission through the rock.
“With increasing stylolite thickness, acoustic waves scatter more strongly and introduce more noise into the wavefield,” says Finkbeiner. “In laboratory experiments, this has implications for monitoring hydraulic fracture propagation in rock samples that contain stylolites. Our results will help determine the best way to locate acoustic emissions inside lab-scale rock samples.”
The researchers are now conducting larger rock block tests. They use advanced fiber optics detection and refined data processing techniques to see whether these findings can be scaled up and repeated.
Hydrogen peroxide (H2O2) packs so much chemical energy into a small space that it is powerful enough to fuel rockets. But this same ability to concentrate energy also makes hydrogen peroxide useful for more Earthly energy applications, such as powering fuel cells. It also holds promise as a green and sustainable energy source: when hydrogen peroxide releases its stored energy, the main byproduct is simply water.
Huabin Zhang and his team at KAUST, together with colleagues from China and the United States, have developed a material that significantly boosts hydrogen peroxide production[1]. “We proposed a convenient strategy to regulate the chemical state of a catalyst at the atomic level,” says Chengyang Feng from the KAUST team.
The conventional approach to producing hydrogen peroxide at an industrial scale requires several organic solvents that are toxic to the environment. A far more environmentally friendly approach is to use a sunlight-powered photocatalytic system.
These systems generate hydrogen peroxide directly from water and oxygen in the air, using a semiconductor photocatalyst that harvests solar energy. When exposed to sunlight, the photocatalyst absorbs photons and generates charge carriers — electrons and holes. The photogenerated electrons participate in the oxygen reduction reaction (ORR) at the catalyst surface, reducing oxygen molecules to hydrogen peroxide.
The efficiency of this process is significantly enhanced by optimizing the surface properties of the photocatalyst and its interaction with light.
One major challenge to improving photocatalytic efficiency is that multiple oxygen reduction reactions can occur. The pathway that leads to hydrogen peroxide involves two electrons. But there are other, less helpful reactions. The four-electron process generates water, while one-electron reactions form unstable superoxides. Among these possibilities, water formation is thermodynamically the most favorable, so catalysts need to be engineered to kinetically favor the desired two-electron pathway to hydrogen peroxide.
The photocatalyst developed by the researchers was made using tungsten trioxide (WO₃), which they modified by adding isolated copper atoms. While tungsten trioxide is already a well-known photocatalyst, the team added copper atoms to capture and activate oxygen molecules and to guide the reactions toward the two-electron pathway.
“Compared to previously demonstrated catalysts, our catalyst features well-defined single-atom catalytic sites where the electronic states that drive the chemical reactions are tunable,” explains Feng “This can be readily achieved by adjusting the interaction between the metal sites and the support.”
The team studied several different compositions, but their best-performing photocatalyst produced 102 micromoles of hydrogen peroxide per hour when irradiated by visible light. This is much higher than any previously reported photocatalyst and 17.3 times more than a copper-free tungsten trioxide catalyst.
“The next step in this research is to further optimize the catalyst system under real-world conditions, explore its scalability and long-term operational stability, and investigate its integration into practical devices or processes,” Feng adds.
“The first global-level assessment of the number, size and properties of shallow-water tropical coral reefs suggest that restoration is more achievable than previously thought,” suggests KAUST faculty and marine ecologist, Carlos M. Duarte
Tropical coral reefs form some of the largest living structures on Earth, offering shelter and sustenance to numerous marine creatures, and providing livelihoods and food to coastal communities. However, these essential marine ecosystems are facing significant damage and degradation due to climate change and human activities.
The Kunming-Montreal Global Biodiversity Framework aims to halt and reverse global biodiversity loss by 2030, but scientists have been missing vital fundamental data on coral reefs that would enable viable restoration and protection projects to get underway.
“Where exactly are tropical coral reefs located? How large are they? What spatial patterns do they exhibit and why?” asks Duarte.
To answer these questions, Duarte has teamed up with scientists Alex Giménez-Romero and Manuel Matias at the Institute of Cross-Disciplinary Physics and Complex Systems (IFISC, UIB-SCIC) in Spain, to conduct an AI-based analysis of the vast Allen Coral Atlas database[1]. The Allen Coral Atlas comprises hundreds of thousands of satellite images, gathered by the Planet satellites, which are then classified by AI to identify precise coral reef habitats from space.
However, it has been cumbersome to extract reef-specific information from the images. So Duarte and co-workers have used AI to identify individual reefs across the entire Atlas, and retrieve location, size, and shape for each of the reefs to examine their collective properties.
The team reports on 1.5 million coral reefs covering a total area of more than 50,000 km2. Their analysis reveals that the spatial geometries of reefs follow three universal scaling laws, or ‘power laws’; in other words, reef growth and distribution follow the scaling and fractal patterns of the Fibonacci sequence.
“Fibonacci series are prevalent across nature, such as fern fronds that form spiral-like features, and even in the shape of the bubble ‘curtains’ that humpback whales produce during hunting,” says Duarte. “To see these scaling patterns replicated in coral reef geometries as seen from space is really exciting.”
The team showed that the size-frequency distribution, the inter-reef distance distribution and the area-perimeter relationship of each reef all follow power laws. Reefs tend to evolve from simple rounded shapes to become more complex, elongated and less compact as they grow. Their fractal geometries emerge with age, and these same patterns are replicated across the globe, regardless of location.
“These universal patterns hint at the nature of the biological and chemical processes driving the shape of coral reefs,” says Duarte. “To support marine biodiversity, the more complex and intricate the structures the better. These are important insights for guiding the morphology of restored reefs; they also provide specific constraints to test reef models.”
The team found that the characteristic (median) size of a coral reef is 0.3 hectares. This suggests that, to achieve global biodiversity goals, the area that needs to be restored in each reef is a maximum of a few hundred square meters.
“Such relatively small areas could be restored, even with limited resources and time, by custodians and citizens committed to individual reefs” says Duarte. “Our paper brings coral reef conservation and restoration to the human scale. It may be feasible for us to reverse the damage we’ve done to these truly beautiful, vital ecosystems.”
A programming framework could streamline chip design by bridging the gap between conceptual design and practical execution[1]. This may help address the ongoing challenge in the computer hardware industry of meeting the demand for high-performance, energy-efficient electronic devices at ever smaller scales.
This demand has long posed a challenge: it requires not only smaller transistors and microchips but also innovative new hardware architectures that provide the optimum arrangements of components for rapid data flow and processing. However, transforming a high-level design — exactly what we want a chip to do — into the low-level details of practical hardware is a lengthy, complex process requiring multiple iterations and collaboration across multiple teams.
In developing the framework, called Assassyn (ASynchronous Semantics for Architectural Simulation and SYNthesis), researchers—including Jian Weng from KAUST—incorporated both architectural simulation and real-world hardware implementation.
Digital computer systems contain interconnected information-storing modules called registers. Register-transfer-level (RTL) languages like Verilog represent the circuit connections between registers, while the designers focus on the overall behaviors of the circuit. This imposes a gap between the design process and the real implementation (the actual wiring of a chip).
“Most chip design research adopts two separate styles: performance is simulated, while power-per-area characteristics are estimated through a separate RTL implementation,” says Weng. “This means there is a mindset misalignment between design and implementation. When designing, you think about doing jobs at one stage and ‘pushing’ data to the next. When implementing in RTL, you need to translate it into a ‘pull’ style in your mind — the latter stage listens for the signal to pull data in.”
Most efficient chip designs include ‘pipelining’ architectures, which allow multiple tasks to be executed simultaneously, keeping many transistors active and saving time — analogous to a factory assembly line. A key challenge in pipeline implementation is the coordination between stages.
“Our insight is that the behavior of pipeline stages cannot take effect immediately,” says Weng. He and his co-workers found that the secret to unifying the design and implementation stages was that events do not all have to happen at the same time.
“In Assassyn we use asynchronous event handling, a widely adopted programming paradigm in website development,” says Weng. “When functions are called, they are not executed all at once. This asynchronicity is the key innovation that makes Assassyn work.”
The team’s evaluation showed that Assassyn generated accurate, high-performance RTL simulations, achieving an order of magnitude speedup over previous models. To avoid reinventing the wheel, Assassyn-generated RTL can directly fit into existing design tools. The results showed that the generated RTL matched the quality of laboriously handcrafted designs in terms of power, chip area, and performance.
“Many hardware design concepts have already converged to their optimal points over the past few decades,” says Weng. “Instead of manually programming every detail of the hardware implementation, some common behaviors can be abstracted in a high-level manner.”
“We are now actively working in several directions, including building new hardware using Assassyn, building better ecosystem support for Assassyn, and extending the abstraction of Assassyn for the next best hardware design and implementation,” concludes Weng.
