Publikační činnost IT4Innovations / Publications of IT4Innovations (9600)

Permanent URI for this collectionhttp://hdl.handle.net/10084/96109

Browse

Now showing 1 - 20 out of 892 results

Rule-based profit taxation in dynamic Cournot oligopoly: Transmission, stability and welfare
(Elsevier, 2026) Nálepová, Veronika; Lampart, Marek
This study develops a dynamic Cournot model to examine whether profit taxation can stabilise oligopolistic markets hit by demand shocks. The tax rate is updated each period by a simple welfare rule, allowing fiscal policy to respond automatically to changing market conditions. The analysis connects the effectiveness with which taxes influence firms' decision-making (the transmission strength) to market stability. Simulations and chaos analysis show that when the tax signal is strong, firms adjust smoothly, volatility falls and competition is preserved. In contrast, when transmission is weak, feedback effects magnify shocks, increasing exit risk and market concentration. Moderate shocks are absorbed through temporary tax changes, while stronger demand shocks in the model mainly threaten the high-cost firm. Overall, transparent and predictable profit taxation serves as a practical stabiliser in concentrated industries, limiting volatility without ad hoc measures and providing a scalable framework for future fiscal design.
Classification enhanced machine learning model for energetic stability of binary compounds
(Elsevier, 2024) Liu, Y. K.; Liu, Z. R.; Xu, T. F.; Legut, Dominik; Yin, X.; Zhang, R. F.
As contemporary computational technologies and machine learning methodologies rapidly evolve, machine learning (ML) models for predicting formation enthalpies of materials exhibited convincible numerical precision and remarkable predictive efficiency, thus establishing a solid foundation for materials thermodynamic design. Despite achieving numerically high global probability accuracy, current ML models for formation enthalpy nonetheless exhibit suboptimal local accuracy within specific physical domain, which can be attributed to the misalignment between the physical constraints of chemical bonds and the critical descriptors capturing classspecific traits. Herein, we propose a novel approach to improve the local precision of the ML model for predicting formation enthalpy by utilizing Miedema theory-based classification, which segments data into distinct categories according to the electronegativity difference, electron density discontinuity and atomic size difference. Utilizing ML algorithms to build surrogate models guided by the classification strategy significantly improves the local predictive accuracy of formation enthalpy for specific binary compounds, significantly raising the R2 value from 0.4-0.9 to 0.8-0.9 compared to an unclassified method. Furthermore, feature importance analysis demonstrates that the pivotal factors for each category vary in some manner, highlighting the insufficiency of a sole ML model in classifying large-dimensional data, which can be addressed by adopting a physicsinformed classification strategy. Our results suggest that employing physical-informed classification scheme into ML equips the models with broad applicability and local accuracy, which also shed light for other material properties predication.
Novel numerical approach toward hybrid nanofluid flow subject to Lorentz force and homogenous/heterogeneous chemical reaction across coaxial cylinders
(AIP Publishing, 2024) Janjua, Khuram Hina; Bilal, Muhammad; Riaz, Muhammad Bilal; Saqib, Abdul Baseer; Ismail, Emad A. A.; Awwad, Fuad A.
The combination of AA7075 and Ti6Al4V aluminum alloys provides an effective balance of endurance, corrosion resistance, and lightness. Some potential applications include aviation components, marine structures with anti-corrosion characteristics, surgical instruments, and athletic apparel. Therefore, the hybrid nanofluid (Hnf) consists of aluminum alloys (AA7075-Ti6Al4V), water (50%), and ethylene glycol (EG-50%) in the current analysis. The Hnf flow subject to heat radiation and Lorentz force is studied through coaxial cylinders. In addition, the flow has been observed under the impacts of homogeneous-heterogeneous (HH) chemical reaction and exponential heat source/sink. The modeled equations (continuity, momentum, HH, and heat equations) are renovated into the non-dimensional form through the similarity approach, which are further numerically computed by employing the ND-solve technique coupling with the shooting method. It can be noticed from the graphical results that the flow rate of Hnf drops with the rising effect of porosity and magnetic field parameters. The addition of AA7075-Ti6Al4V nanoparticles (NPs) also reduces the fluid temperature and velocity profile. Furthermore, the concentration distribution diminishes with the flourishing effect of HH parameters.
Graphene acid: A potent carbocatalyst for the friedel-crafts arylation of aldehydes with indoles
(Wiley, 2025) Galathri, Eirini M.; Hrubý, Vítězslav; Mountanea, Olga G.; Mantzourani, Christiana; Chronopoulos, Demetrios D.; Otyepka, Michal; Kokotos, Christoforos G.
Carbocatalysis represents a highly attractive and effective field within the realm of metal-free nanocatalysis, significantly advancing sustainability in synthetic chemistry. Graphene acid (GA) emerges as a well-defined graphene derivative, characterized by a high density of homogeneously distributed carboxylic groups over graphene lattice. This unique and uniform structure positions GA as an elegant alternative to other 2D carbocatalysts, namely graphene oxide. GA was successfully employed as the catalyst in a Friedel-Crafts-type reaction between indoles and aldehydes, facilitating the synthesis of bis(indolyl)methanes, organic compounds exhibiting interesting biological properties and significant pharmaceutical potential. The metal-free nature of GA, combined with the performance of the reaction "on water" under mild conditions, highlight the green credentials of the developed protocol. Comprehensive substrate screening, including a plethora of aliphatic or aromatic aldehydes and various 1- or 2-substituted indoles, resulted in moderate to high yields of variously functionalized bis(indolyl)methanes. Mechanistic and recovering studies showed that GA acts catalytically as a Br & oslash;nsted acid, maintaining its catalytic activity at a high rate for six subsequent cycles.
Analytical solutions and dynamical behaviors of the extended Bogoyavlensky-Konopelchenko equation in deep water dynamics
(IOP Publishing, 2025) Jhangeer, Adil; Beenish, Abdallah M.; Talafha, Abdallah M.; Ansari, Ali R.
In this study, we delve into the mathematical intricacies of the novel Bogoyavlensky-Konopelchenko equation, which finds practical applications in understanding the dynamics of internal waves in deep water. This equation holds significance across scientific fields such as plasma physics, nonlinear optics, and fluid dynamics. The equation extends the (2+1)-dimensional Bogoyavlensky-Konopelchenko equation by adding the second-order derivative terms B mu x mu x and B mu y mu y due to second-order dissipative elements. The generalized exponential rational function method, crucial in mechanical engineering, analyzes analytical solutions featuring symmetric waveform representations. The planar dynamical system, derived via Galilean transformation with mathematical models and parameter values, enhances problem comprehension. Sensitivity analysis and phase portraits of equilibrium points highlight symmetrical properties. The global analysis identifies periodic, quasi-periodic, and chaotic behaviors, corroborated by Poincar & eacute; maps, attractor, power spectrum, return map, and a symmetric basin of the largest Lyapunov exponent.
Modeling and simulations for the mitigation of atmospheric carbon dioxide through forest management programs
(AIMS Press, 2024) Riaz, Muhammad Bilal; Raza, Nauman; Martinovič, Jan; Bakar, Abu; Tunç, Osman
The growing global population causes more anthropogenic carbon dioxide (CO2) 2 ) emissions and raises the need for forest products, which in turn causes deforestation and elevated CO2 2 levels. A rise in the concentration of carbon dioxide in the atmosphere is the major reason for global warming. Carbon dioxide concentrations must be reduced soon to achieve the mitigation of climate change. Forest management programs accommodate a way to manage atmospheric CO2 2 levels. For this purpose, we considered a nonlinear fractional model to analyze the impact of forest management policies on mitigating atmospheric CO2 2 concentration. In this investigation, fractional differential equations were solved by utilizing the Atangana Baleanu Caputo derivative operator. It captures memory effects and shows resilience and efficiency in collecting system dynamics with less processing power. This model consists of four compartments, the concentration of carbon dioxide C (t), human population N (t), forest biomass B (t), and forest management programs P (t) at any time t. The existence and uniqueness of the solution for the fractional model are shown. Physical properties of the solution, non-negativity, and boundedness are also proven. The equilibrium points of the model were computed and further analyzed for local and global asymptotic stability. For the numerical solution of the suggested model, the Atangana-Toufik numerical scheme was employed. The acquired results validate analytical results and show the significance of arbitrary order delta . The effect of deforestation activities and forest management strategies were also analyzed on the dynamics of atmospheric carbon dioxide and forest biomass under the suggested technique. The illustrated results describe that the concentration of CO2 2 can be minimized if deforestation activities are controlled and proper forest management policies are developed and implemented. Furthermore, it is determined that switching to low-carbon energy sources, and developing and implementing more effective mitigation measures will result in a decrease in the mitigation of CO 2 .
Applying fractional calculus to malware spread: A fractal-based approach to threat analysis
(PLOS, 2025) Razi, Nausheen; Riaz, Muhammad Bilal; Kamran, Tayyab; Ishtiaq, Umar; Shafiq, Anum
Malware is a common word in modern era. Everyone using computer is aware of it. Some users have to face the problem known as Cyber crimes. Nobody can survive without use of modern technologies based on computer networking. To avoid threat of malware, different companies provide antivirus strategies on a high cost. To prevent the data and keep privacy, companies using computers have to buy these antivirus programs (software). Software varies due to types of malware and is developed on structure of malware with a deep insight on behavior of nodes. We selected a mathematical malware propagation model having variable infection rate. We were interested in examining the impact of memory effects in this dynamical system in the sense of fractal fractional (FF) derivatives. In this paper, theoretical analysis is performed by concepts of fixed point theory. Existence, uniqueness and stability conditions are investigated for FF model. Numerical algorithm based on Lagrange two points interpolation polynomial is formed and simulation is done using Matlab R2016a on the deterministic model. We see the impact of different FF orders using power law kernel. Sensitivity analysis of different parameters such as initial infection rate, variable adjustment to sensitivity of infected nodes, immune rate of antivirus strategies and loss rate of immunity of removed nodes is investigated under FF model and is compared with classical. On investigation, we find that FF model describes the effects of memory on nodes in detail. Antivirus software can be developed considering the effect of FF orders and parameters to reduce persistence and eradication of infection. Small changes cause significant perturbation in infected nodes and malware can be driven into passive mode by understanding its propagation by FF derivatives and may take necessary actions to prevent the disaster caused by cyber crimes.
Analyzing optical soliton solutions in Kairat-X equation via new auxiliary equation method
(Springer Nature, 2024) Faridi, Waqas Ali; Tipu, Ghulam Hussain; Riaz, Muhammad Bilal; Mostafa, Almetwally M.; Alqahtani, Salman A.; Myrzakulov, Ratbay; Umurzakhova, Zhanar
The paper introduce a novel auxiliary equation method for the successful derivation of traveling wave solutions for the non-linear Kairat-X (K-X) equation. Along with other novel results, soliton, singular, triangular periodic, and doubly periodic topological solutions are among the solutions obtained. The study revisits the concept of optical solitary waves, enhancing our understanding of the model. Previous studies have already derived analytical solutions using diverse approaches, contributing to the discovery of new soliton solutions within this framework. These solutions are characterized through three-dimensional, contour plot, and two-dimensional profile analyses. Additionally, the impact of time on the propagation of wave patterns is explored. The outcomes show how well our suggested approach works to solve non-linear evolution equations by producing fresh, more thorough solutions, making it a powerful mathematical tool for doing so. Through this article, we elucidate how leveraging NAEM with the Kairat-X equation can lead to optimized optical systems, improved data transmission rates, and the evolution of nonlinear optics towards more efficient and reliable communication technologies.
Reactive power compensation during the convergence of grid system with FACTS devices
(Elsevier, 2024) Nigam, Akhil; Sharma, Kamal Kant; Riaz, Muhammad Bilal; Yaseen, Moh; Shafiq, Anum; Sindhu, Tabassum Naz
From the last few years power demanding is increasing as per load requirement. The condition of power demand is not accompanied by only generation and transmission of electricity. Many industrial power plants are running on higher capacity to fulfill load requirement. There are many issues such as voltage instability and power quality which led to many power plants. In distributed generation generators are distributed asynchronous type and mostly affected by the reactive power. They basically cannot generate reactive power but draws away from the power system. It causes the voltage drop to the consumer side which may be not acceptable. Then reactive power management is essential for reducing those problems and provides continuity of electricity to the consumers. Distributed generation systems and FACTS devices are converting conventional power grid into smart grid. Reactive power compensation is now challenging issue to preserve adequate power quality and improve the performance of distribution system. There are many FACTS devices such as we can control power factor of the circuit and reactive power compensation. Many researchers have focused on behavior of FACTS device in an unbalanced condition and majority work is done only with DSTATCOM. This paper deals with different FACTS devices in grid systems with analysis and overview of D-FACTS devices. Hence in this paper, applications of various D-FACTS devices have been described to find location of compensation devices for reliable operation.
Can we ever develop an ideal RNA force field? Lessons learned from simulations of the UUCG RNA tetraloop and other systems
(American Chemical Society, 2025) Mlýnský, Vojtěch; Kührová, Petra; Pykal, Martin; Krepl, Miroslav; Stadlbauer, Petr; Otyepka, Michal; Banáš, Pavel; Šponer, Jiří
Molecular dynamics (MD) simulations are an important and well-established tool for investigating RNA structural dynamics, but their accuracy relies heavily on the quality of the employed force field (ff). In this work, we present a comprehensive evaluation of widely used pair-additive and polarizable RNA ffs using the challenging UUCG tetraloop (TL) benchmark system. Extensive standard MD simulations, initiated from the NMR structure of the 14-mer UUCG TL, revealed that most ffs did not maintain the native state, instead favoring alternative loop conformations. Notably, three very recent variants of pair-additive ffs, OL3CP–gHBfix21, DES-Amber, and OL3R2.7, successfully preserved the native structure over a 10 × 20 μs time scale. To further assess these ffs, we performed enhanced sampling folding simulations of the shorter 8-mer UUCG TL, starting from the single-stranded conformation. Estimated folding free energies (ΔG°fold) varied significantly among these three ffs, with values of 0.0 ± 0.6, 2.4 ± 0.8, and 7.4 ± 0.2 kcal/mol for OL3CP–gHBfix21, DES-Amber, and OL3R2.7, respectively. The ΔG°fold value predicted by the OL3CP–gHBfix21 ff was closest to experimental estimates, ranging from −1.6 to −0.7 kcal/mol. In contrast, the higher ΔG°fold values obtained using DES-Amber and OL3R2.7 were unexpected, suggesting that key interactions are inaccurately described in the folded, unfolded, or misfolded ensembles. These discrepancies led us to further test DES-Amber and OL3R2.7 ffs on additional RNA and DNA systems, where further performance issues were observed. Our results emphasize the complexity of accurately modeling RNA dynamics and suggest that creating an RNA ff capable of reliably performing across a wide range of RNA systems remains extremely challenging. In conclusion, our study provides valuable insights into the capabilities of current RNA ffs and highlights key areas for future ff development.
Striking impact of solvent polarity on the strength of hydrogen-bonded complexes: A nexus between theory and Experiment
(Wiley, 2025) Lo, Rabindranath; Manna, Debashree; Vacek, Jaroslav; Bour, Petr; Wu, Tao; Osifová, Zuzana; Socha, Ondřej; Dračínský, Martin; Hobza, Pavel
The binding free energy of hydrogen-bondedcomplexes is generally inversely proportional to thesolvent dielectric constant. This occurs because thesolvent-accessible surface area of the complex is alwayssmaller than that of the individual subsystems, leadingto a reduction in solvation energy. The present studyexplores the potential for stabilizing hydrogen-bondedcomplexes in a solvent with higher polarity. Contrary tothe established understanding, we have demonstratedthat the hydrogen-bonded complex(CH3CH2COOH···2,4,6-trimethylpyridine) can be betterstabilized in a solvent with higher polarity. In this case, asignificant charge transfer between the subsystemsresults in an increased dipole moment of the complex,leading to its stabilization in a more polar solvent. Theexpected inverse relationship between binding freeenergy and solvent dielectric constant is observed whenthe charge transfer between the subsystems is low. Thus,the magnitude of the charge transfer between subsys-tems is possibly the key factor in determining thestabilization or destabilization of H-bonded complexesin different solvents. Here, we present a comprehensivestudy that combines experimental and theoretical ap-proaches, including nuclear magnetic resonance (NMR),infrared (IR) spectroscopies and quantum chemicalcalculations to validate the findings.
Electrochemical stability of biodegradable Zn-Cu alloys through machine-learning accelerated high-throughput discovery
(Royal Society of Chemistry, 2024) Luo, Kun; Liu, Zhaorui; Yu, Rui; Xu, Tengfei; Legut, Dominik; Yin, Xing; Zhang, Ruifeng, Ruifeng
Zn-Cu alloys have attracted great attention as biodegradable alloys owing to their excellent mechanical properties and biocompatibility, with corrosion characteristics being crucial for their suitability for biomedical applications. However, the unresolved identification of intermetallic compounds in Zn-Cu alloys affecting corrosion and the complexity of the application environment hamper the understanding of their electrochemical behavior. Utilizing high-throughput first-principles calculations and machine-learning accelerated evolutionary algorithms for screening the most stable compounds in Zn-Cu systems, a dataset encompassing the formation energy of 2033 compounds is generated. It reveals that most of the experimentally reported Zn-Cu compounds can be replicated, especially the structure of R32 CuZn5 is first discovered which possesses the lowest formation energy of -0.050 eV per atom. Furthermore, the simulated X-ray diffraction pattern matches perfectly with the experimental ones. By formulating 342 potential electrochemical reactions based on the binary compounds, the Pourbaix diagrams for Zn-Cu alloys are constructed to clarify the fundamental competition between different phases and ions. The calculated equilibrium potential of CuZn5 is higher than that of Zn through the forward reaction Zn + CuZn5 reversible arrow CuZn5 + Zn2+ + 2e(-), resulting in microcell formation owing to the stronger charge density localization in Zn compared to CuZn5. The presence of chlorine accelerates the corrosion of Zn through the reaction Zn + CuZn5 + 6Cl(-) + 6H(2)O reversible arrow Cu + 6ZnOHCl + 6H(+) + 12e(-), where the formation of ZnOHCl disrupts the ZnO passive film and expands the corrosion pH range from 9.2 to 8.8. Our findings reveal an accurate quantitative corrosion mechanism for Zn-Cu alloys, providing an effective pathway to investigate the corrosion resistance of biodegradable alloys.
GPU acceleration of hybrid FETI solver for problems of transient nonlinear dynamics
(Elsevier, 2026) Homola, Jakub; Meca, Ondřej; Říha, Lubomír; Brzobohatý, Tomáš
FETI methods, which build on the Finite Element Method, are utilized for large-scale engineering simulations. They use domain decomposition techniques to divide a large domain into many smaller subdomains, which can be processed in parallel. Current trends in HPC focus on GPU-accelerated clusters. To utilize them efficiently, FETI solvers should be able to use these accelerators. Recent developments have demonstrated that the fundamental component of the FETI methods, the dual operator, can be successfully offloaded to the GPU.In this paper, we focus on GPU acceleration of the Hybrid FETI variant. It reduces the size of the projector by using a two-level decomposition, thus allowing for a significantly higher number of compute nodes to be efficiently utilized. In turn, it allows us to split the problem into a larger number of smaller subdomains, which improves single-process performance. We demonstrate the performance on a real-world problem of transient nonlinear dynamics that requires reassembling of the dual operator, preconditioner, and projector during each call of the solver. On the MareNostrum 5 supercomputer, using Nvidia H100 GPUs, we achieved a speedup of 2.9 for the whole Hybrid FETI solver compared to a CPU-only run.
Theoretical investigation on fractal-fractional nonlinear ordinary differential equations
(Elsevier, 2025) Atangana, Abdon; Araz, Seda Igret
In this study, we examine the existence and uniqueness conditions of the solutions of the nonlinear fractal-fractional differential equations. Particular emphasis is placed on four cases: exponential decay, power law, generalized Mittag-Leffler kernels and the Delta-Dirac function. Our first contribution is the formulation of some basic inequalities inspired from Gronwall inequality setting up a solid foundation for our analysis to follow. We subsequently carefully obtain the maximal and minimal solutions in each scenario, providing a complete picture of their structure. Finally we show convergence of four different successive approximation schemes, validating their applicability in the various contexts. This is an important finding that adds to the growing literature on the use of fractional calculus in complex dynamical systems.
Parametrized predictor-corrector method for initial value problems with classical and Caputo-Fabrizio derivatives
(World Scientific Publishing, 2024) Atangana, Abdon; Araz, Seda Igret
Ordinary nonlinear differential equations with classical and fractional derivatives are used to simulate several real-world problems. Nonetheless, numerical approaches are used to acquire their solutions. While various have been proposed, they are susceptible to both disadvantages and advantages. In this paper, we propose a more accurate numerical system for solving nonlinear differential equations with classical and Caputo-Fabrizio derivatives by combining two concepts: the parametrized method and the predictor-corrector method. We gave theoretical analyses to demonstrate the method's correctness, as well as several illustrated examples for both scenarios.
Fire hazards caused by equipment used in offshore oil and gas operations: prescriptive vs. goal-oriented legislation
(MDPI, 2025) Brkic, Dejan
This article offers a concise overview of the best practices for safety in offshore oil and gas operations, focusing on the risks associated with various types of equipment, particularly on the risk of fire. It identifies specific machinery and systems that could pose hazards, assesses their potential impact on safety, and explores conditions that may lead to accidents. Some of the largest accidents were analyzed for their associations with fire hazards and specific equipment. Two primary regulatory approaches to offshore safety are examined: the prescriptive approach in the United States (US) and the goal-oriented approach in Europe. The prescriptive approach mandates strict compliance with specific regulations, while in the goal-oriented approach a failure to adhere to recognized best practices can result in legal accountability for negligence, especially concerning human life and environmental protection. This article also reviews achievements in safety through the efforts of regulatory authorities, industry collaborations, technical standards, and risk assessments, with particular attention given to the status of Mobile Offshore Drilling Units (MODUs). Contrary to common belief, the most frequent types of accidents are not those involving a fire/explosion caused by the failure of the Blowout Preventer (BOP) after a well problem has already started. Following analysis, it can be concluded that the most frequent type of accident typically occurs without fire and is due to material fatigue. This can result in the collapse of the facility, capsizing of the platform, and loss of buoyancy of mobile units, particularly in bad weather or during towing operations. It cannot be concluded that accidents can be more efficiently prevented under a specific type of safety regime, whether prescriptive or goal-oriented.
Shared entanglement for three-party causal order guessing game
(IOP Publishing, 2025) Kukulski, Ryszard; Lewandowska, Paulin; Zyczkowski, Karol
In a variant of communication tasks, players cooperate in choosing their local strategies to compute a given task later, working separately. Utilizing quantum bits for communication and sharing entanglement between parties is a recognized method to enhance performance in these situations. In this work, we introduce the game for which three parties, Alice, Bob and Charlie, would like to discover the hidden order in which they make the moves. We show the advantage of quantum strategies that use shared entanglement and local operations over classical setups for discriminating operations’ composition order. The role of quantum resources improving the probability of successful discrimination is also investigated. Our research provides a basis for examining computational model featuring a specific gate set while examining the diverse operations achievable through permutations of its elements.
Assessing the global dynamics of Nipah infection under vaccination and treatment: A novel computational modeling approach
(PLOS, 2025) Yu, Fang; Khan, Muhammad Younas; Riaz, Muhammad Bilal; Ullah, Saif
In biology and life sciences, fractal theory and fractional calculus have significant applications in simulating and understanding complex problems. In this paper, a compartmental model employing Caputo-type fractional and fractal-fractional operators is presented to analyze Nipah virus (NiV) dynamics and transmission. Initially, the model includes nine nonlinear ordinary differential equations that consider viral concentration, flying fox, and human populations simultaneously. The model is reconstructed using fractional calculus and fractal theory to better understand NiV transmission dynamics. We analyze the model’s existence and uniqueness in both contexts and instigate the equilibrium points. The clinical epidemiology of Bangladesh is used to estimate model parameters. The fractional model’s stability is examined using Ulam-Hyers and Ulam-Hyers-Rassias stabilities. Moreover, interpolation methods are used to construct computational techniques to simulate the NiV model in fractional and fractal-fractional cases. Simulations are performed to validate the stable behavior of the model for different fractal and fractional orders. The present findings will be beneficial in employing advanced computational approaches in modeling and control of NiV outbreaks.
An in-depth study of GPU frequency-scaling latency and its optimization on modern architectures
(Elsevier, 2026) Velička, Daniel; Vysocký, Ondřej; Yasal, Osman; Říha, Lubomír
The move towards the exascale systems in High-Performance Computing and the demand for Artificial Intelligence brought together thousands of CPUs and even more GPU accelerators. This massive hardware consolidation has made energy optimization a critical challenge. The immense amount of energy consumption creates a cascade of secondary issues: it increases the carbon footprint, generates significant heat that demands advanced cooling, and causes dramatic power fluctuations that threaten the stability of the electrical grid. Although energy-saving techniques based on Dynamic Voltage and Frequency Scaling are well understood for CPUs, a critical knowledge gap exists for GPU accelerators, limiting the ability to apply similar optimizations. This paper presents a method for measuring how long it takes the CPU to adjust the operating frequency of the GPU (switching latency), and how long the frequency change itself takes to complete (transition latency). The approach employs a minimal iterative workload that allows statistically distinguishing runtime differences between frequency pairs. It first measures execution times for each frequency and then determines the switching and transition latency of the change from an initial to a target frequency by tracking runtime changes and repeating measurements to ensure statistical robustness. Finally, the methodology filters out outliers from external factors such as driver management or system interruptions. The methodology is implemented in the open-source LATEST [1] tool with support for NVIDIA GPU accelerators. It is evaluated on three GPUs based on different generations of architecture, GH200, A100-SXM4, and RTX Quadro 6000. These results show that the transition latency takes from hundreds of microseconds up to hundreds of milliseconds, while the absolute majority of the time is spent in the GPU applying the frequency change. From the analysed GPUs, the GH200 exhibited the widest range, with switching latencies spanning from 5.6 ms to 477 ms and transition latencies from 0.2 ms to 471 ms. Additionally, the transition latency measurement can be used to identify manufacturing variability of accelerators, showing differences in frequency scaling reactivity. Our analysis identifies specific frequency pairs with high switching latencies, creating a challenge that the slow transitions discourage their use, yet the target frequencies themselves may be highly efficient in terms of energy consumption. To address this, we introduce an indirect switching method that leverages an intermediate frequency. This technique effectively circumvents overhead, allowing the system to access these efficient frequency states without the high latency penalty of a direct transition. The use of the indirect frequency switching technique produced a latency reduction between 250 and 431 ms, for a single frequency change on GH200.
Environmental stability and ageing of ScN thin films from XPS Ar+ depth profiling
(Elsevier, 2024) Cichoň, Stanislav; More-Chevalier, Joris; Wdowik, Urszula D.; de Prado, Esther; Bulíř, Jiří; Novotný, Michal; Fekete, Ladislav; Duchoň, Jan; Legut, Dominik; Lančok, Ján
A basic knowledge on chemical stability and reactivity of a wide band gap ScN semiconductor in the presence of air atmosphere, where O2 and H2O represent the main degradation or corrosive agents, is of vital importance for a long-term performance of ScN-based devices for thermoelectric applications. Here, we present a systematic XPS Ar+ depth profiling analysis and optical and TEM characterizations of naturally room temperature air-aged thin ScN films prepared by a high-temperature DC sputtering on MgO(0 0 1) and SiO2 substrates. We find that superior crystalline quality ScN/MgO films degrade weakly after their quick initial surface oxidation and/or hydrolysis. Their oxidation is rather local and associated with the film-penetrating void-like interfaces. Significant initial surface oxidation and subsequent bulk oxidation after ageing is, however, observed for polycrystalline ScN/SiO2 films. Ab initio calculations of pure ScN and ScN with diluted nitrogen vacancies and/or substitutional oxygen impurities, which assist our experimental research, reveal pronounced impact of these defects on the ScN electronic structure. The modeled compositions reflect homogeneously and weakly oxidized films, while the real films correspond to relatively pure ScN crystallites with interfaces rich in oxygen.

Browse

Recent Submissions