Articles tagged with Computer Architecture
A new AI algorithm, APOLLO, accurately predicts microprocessor power consumption by analyzing just 100 signals out of millions, offering potential to improve efficiency and develop new processors. The technique has been validated on high-performance microprocessors and could help designers inform future chip design.
Washington University researchers have designed a new processing-in-memory (PIM) circuit that can increase PIM computing's performance by orders of magnitude. The circuit uses resistive random-access memory PIM, allowing for analog computations and eliminating the need for digital conversions.
The University of Texas at Arlington has received a $1.5 million grant from the US Department of Education to support domestic doctoral students specializing in Internet-of-Things study and research. The project aims to increase diversity among computer science researchers, with a focus on women and minorities.
Fred Martin, a UMass Lowell computer scientist, is being recognized for developing an educational framework that introduces robotics and AI to K-12 students. His work has guided the development of primary and secondary AI education initiatives worldwide.
A new brain-inspired computing device harnesses randomness to improve AI and machine learning performance. The device, called hetero-memristor, emulates the stochastic properties of neurons, allowing for more efficient optimization and reduced power consumption.
Scientists at Aarhus University are working on a nano-sized brain-inspired computer that can harvest its own energy, making it the smallest and most efficient AI system yet. The project aims to reduce power consumption by 12 orders of magnitude compared to modern supercomputers.
A new molecule discovered by researchers at University of Limerick in Ireland enables fast decision-making in computers, breaking the von Neumann bottleneck. The device can solve problems even if individual components fail, providing smaller, faster, and more energy-efficient computing.
Researchers at George Washington University have created a nanophotonic analog processor capable of solving partial differential equations. The processor can process arbitrary inputs at the speed of light and is integrated at chip-scale.
Researchers at Tokyo Institute of Technology developed a tunable neural network framework that achieves high accuracy and efficiency for sparse CNNs. The new architecture employs a Cartesian-product MAC array and pipelined activation aligners to enable dense computing of sparse convolution, resulting in better resource utilization.
Quantum engineers at the University of New South Wales have discovered a new technique to control millions of spin qubits, a critical step towards building a practical quantum computer. This breakthrough uses a novel component called a dielectric resonator to focus microwave power and deliver uniform magnetic fields across the chip.
Lehigh University researcher Roberto Palmieri aims to make RDMA technology even faster by revisiting a long-held theory. His goal is to enable all machines to interact with local memory, reducing latency and improving performance.
Columbia Engineering researchers have developed new techniques to bolster memory safety, creating a processor for the Air Force Research Lab. The novel solution, ZeRØ, protects code and data pointers without affecting system performance, while No-FAT speeds up fuzz testing and makes security checks faster with minimal impact on speed.
Margaret Martonosi's work on power-aware microarchitectures and narrow bit-widths has led to new fields of research. She has authored over 175 publications, including papers on dynamic thermal management and formal control-theoretic approaches for balancing power and performance.
A team of University of Virginia researchers has uncovered a new line of attack that breaks all Spectre defenses, leaving billions of computers and devices vulnerable to hackers. The attackers can steal data by exploiting micro-op caches, which have been built into Intel processors since 2011.
A new system called SpAtten enables more streamlined NLP with less computing power, achieving speeds of over 100 times faster than competing general-purpose processors and reducing energy consumption by over 1,000 times.
Researchers study the role of memristors in neuromorphic computing to mimic biological brain architectures. Memristor devices can memorize current to reduce device size and increase processing speed.
Researchers at EPFL's LANES laboratory have created a revolutionary circuit that combines logic operations and data storage into a single architecture, reducing energy losses and increasing efficiency. This breakthrough technology uses a 2D material called MoS2 to enable smaller, more powerful devices with improved performance.
A recent study proposes a collaborative system architecture that enables MNO-CSP collaboration to improve the performance of advanced 5G applications. By sharing information and resources, CSPs and MNOs can optimize their systems for better agility, flexibility, and efficiency.
Rice University researchers have demonstrated methods to optimize data-centric processing, improving energy efficiency by up to two orders of magnitude. Two complementary approaches, TIMELY and SmartExchange, were presented at ISCA 2020, leveraging innovative hardware architecture and co-designing with machine-learning algorithms.
Barroso's designs paired inexpensive hardware with powerful distributed systems software to dramatically change system design. His work led to the development of hyperscale datacenters and has had a significant impact on industry, resulting in reduced hardware and operating costs.
The ACM Transactions on Internet of Things (TIOT) journal chronicles the transformative development of the IoT. The journal covers a broad range of topics, including low-power wireless networking, embedded systems, data analytics, and security, with a strong emphasis on experimental validation.
Scientists at Tokyo Institute of Technology have developed a new processor architecture, STATICA, that can efficiently solve combinatorial optimization problems. The proposed system is fully connected and considers all spin-to-spin interactions, enabling parallel updating using stochastic cell automata, which reduces calculation time.
Researchers propose hybrid architectures to advance brain-inspired neuromorphic computing, focusing on operational functions needed to process information efficiently. The future of computing will not be about scaling components, but rethinking processor architecture to emulate brain efficiency.
The University of North Carolina at Chapel Hill is leading a $20 million project to create a platform for testing novel internet architectures. FABRIC, funded by the National Science Foundation, will allow scientists to explore what a new internet could look like at scale.
Researchers have developed a brain-inspired, analog neural network that provides probabilistic responses for complex decision-making. The device is more energy efficient and produces less heat than current computing architectures.
A new deep neural network architecture can identify manipulated images at the pixel level, detecting blurred boundaries and unnatural transitions between regions. This technology aims to improve photo editing tool security and detect deepfakes with high precision.
KAUST researchers develop a universal framework for querying big data, allowing researchers to focus on advancing the query engine rather than coding for specific platforms. The approach uses sparse-matrix algebra and achieves performance comparable to existing specialized approaches.
A Massachusetts Institute of Technology-developed system automatically generates optimized cryptography code for Google Chrome browsers and web applications, matching performance of handwritten code but faster. The system, known as Fiat Cryptography, is being widely used by tech firms and has populated Google's BoringSSL library.
A new method developed by researchers eliminates speculative memory side-channel attacks, exposing security vulnerabilities in microprocessors. The proposed solution reduces performance costs by 11% and energy usage by 7%, without compromising system security.
MORPHEUS, a new processor architecture, blocks potential attacks by encrypting and randomizing key bits 20 times per second. This makes vulnerabilities virtually impossible to exploit, providing a future-proof secure system.
Researchers have developed a new detection technique that uses power anomalies to identify micro-architectural attacks in embedded systems. This approach can reduce the effects of malware even if it evades traditional detection methods, by slowing down data transfer rates by up to 97 percent.
Researchers have designed a new flash-storage system called LightStore that modifies SSDs to connect directly to a data center's network, enabling more efficient data storage operations. In experiments, the system was found to run twice as efficiently as traditional storage servers and required less than half the physical space.
Researchers at UNSW's CQC2T have shown that they can build atomic precision qubits in a 3D device, achieving critical components of the 3D chip architecture. They demonstrated the feasibility of an architecture using atomic-scale qubits aligned to control lines inside a 3D design.
A team of researchers is using a $820,000 DARPA grant to design a new kind of chip that can automatically match software with hardware, significantly reducing the time it takes. The chip, called DASH-SoC, aims to achieve this in just five nanoseconds.
KAUST researchers create a distributed architecture allowing drones to coordinate based on local information and peer-to-peer communications. The algorithm enables rapid reaction times without excessive computation, making it effective in real-time applications.
Researchers at U.S. Army Research Laboratory have developed a way to factor large composite integers using brain-inspired computer architectures, moving away from traditional computing methods. This breakthrough could break many modern-day internet security protocols, including public key encryption.
Researchers found that initial bone parameters significantly correlate with the risk of fracture in postmenopausal women. Peak bone mass and density appear to play a more crucial role in fragility fractures than previously thought.
Researchers at the University of Groningen have developed a new metal-semiconductor interface that combines storage, memory and processing in one unit, paving the way for brain-inspired computing architectures. The device uses a spin-memristor with tunability, enabling non-volatile storage and operation at room temperature.
CONIX Research Center aims to develop novel architectures for large-scale, distributed computing systems with implications for social interaction, smart buildings and infrastructure. The center will create a substrate for networked computing that connects edge devices to the cloud, enabling performance, security, and privacy guarantees.
Argonne is collaborating with HPE to evaluate early versions of Cavium ARM ThunderX2 processors for the ARM ecosystem, seeking a cost-effective alternative to x86 architectures. The Comanche Wave prototype ARM64 server platform will be installed at Argonne's testing environment to support this work.
Two ORNL-led research teams will assess the feasibility of quantum architectures in addressing big science problems and develop algorithms to harness massive power predicted by quantum computing systems. Researchers aim to create quantum computers capable of simulating phenomena at unprecedented scales and speeds.
Researchers at North Carolina State University developed a tool called CIF-CK to create more reactive and adaptable NPCs in games. The tool uses social behavior models to make individual NPCs respond differently to player behavior, leading to a more immersive gaming experience.
Researchers at MIT and Stanford developed a 3D chip that integrates computing and data storage, overcoming communication bottlenecks. The chip uses carbon nanotubes and RRAM cells, enabling dense and fine-grained integration of computating and data storage.
Researchers at UTA have developed a mathematical solution to optimize computing and network resource allocation in data centers, ensuring better user experiences. The new grant will allow for the development of a user-context-aware joint computing-and-networking resource allocation framework.
A new architecture for 3D stacked memory increases rendering speeds by up to 65% by processing data in memory, reducing traffic between GPU cores. This innovation benefits both science, particularly complex modeling and virtual reality, and the gaming industry.
Researchers at Princeton University discovered over 100 errors in RISC-V instruction specification that could cause software running on chips to perform calculations using incorrect values. The findings highlight the importance of correcting these issues to ensure the reliability and safety of systems relying on RISC-V processors.
Researchers at Oak Ridge National Laboratory are developing three new approaches to improve deep learning technologies. They are bringing together quantum, high-performance, and neuromorphic computing architectures to optimize complexity in a low-power environment. Additionally, scientists have created an approach to get a better look ...
EPFL scientists developed a new computer model to predict allosteric pathways for enzymes and other proteins, enabling more efficient drug design. The model proposes a new hypothesis for allosteric architectures, introducing the concept of 'levers' that amplify responses at a distance.
Researchers are developing novel approaches to process and store data in energy-efficient ways, using nanoscale devices and innovative architectures. The goal is to create computers that can learn from data and operate like the human brain.
The Computational Science Initiative at Brookhaven Lab will develop modeling and simulation applications for nuclear and high-energy physics, focusing on lattice quantum chromodynamics calculations and computational chemistry code NWChem. The projects aim to optimize societal impact in fields like climate science and materials science.
Researchers at Princeton University have developed a new microchip called Piton that significantly increases processing speed and slashes energy consumption for data centers. The chip's scalable architecture enables thousands of cores to be connected together, leading to increased efficiency and cost savings.
Researchers at University of Innsbruck propose new quantum computer architecture that detaches logical qubit from physical implementation, overcoming challenges in adiabatic quantum computation. This approach enables scalable and fault-tolerant quantum computing.
Researchers at MIT developed a new system that makes servers using flash memory as efficient as those using conventional RAM while preserving power and cost savings. The system uses preprocessed data on flash drives to improve distributed computation efficiency.
Researchers are exploring software-defined cellular networking to provide next-generation mobile broadband with speeds of up to 10 Gbits/s. The proposed end-to-end architecture offers flexibility, scalability, agility, and efficiency, while overcoming bandwidth shortages and improving quality of service.
Researchers at UCSB's Martinis Lab successfully demonstrated a quantum version of Gauss's law using superconducting qubits. The team achieved full control over a two-qubit system, enabling precise measurement of local curvature through movement, showcasing the power of arbitrary control in quantum simulation.
The NSF has awarded $10 million to create a cloud computing testbed called Chameleon, enabling researchers to develop and experiment with novel cloud architectures. The testbed will provide persistent infrastructure clouds, 'bare metal' provisioning of hardware, and pre-configured software stacks for users to build cloud services.
The NSF awards two $10 million projects to create cloud computing testbeds, enabling the academic research community to develop and experiment with novel cloud architectures. Chameleon and CloudLab will support real-time and safety-critical applications in medical devices, power grids, and transportation systems.
Thomas Colbert proposes innovative solutions to protect Galveston Bay from hurricane storm surge and climate change through thoughtful design. His concepts include functional urban amenities, leveraging lush wetlands as tourist attractions, to address the challenges faced by this coastal community.
The National Science Foundation has awarded $15 million to three multi-institutional projects to further develop, deploy and test future Internet architectures. The projects aim to enhance security, respond to emerging service challenges and enable the scalability of the information infrastructure.
Carnegie Mellon researchers will test the eXpressive Internet Architecture (XIA) in vehicular networks and for delivering online video. The deployment aims to eliminate bottlenecks in video transmission, which accounts for most internet traffic.