APMT: an automatic hardware counter-based performance modeling tool for HPC applications

A new performance and performability modeling tool is introduced in this paper. PENELOPE is the first tool which incorporates evaluation and optimization algorithms. It is the result of a combination between the performability modeling concept and Markov decision theory. Different algorithms are adopted and included in the tool under the unifying paradigm of reconfigurability as the basis for adaptation and optimization. In addition to transient and steady-state performability measures, also transient and stationary control functions can be computed and graphically presented. Model specification and specification of transient or stationary control functions can be separately performed and deliberately combined with each other. Besides providing a new modeling paradigm, the tool supports model creation, experimentation, storage and presentation of results by means of an easily usable interface and an integrated model data base system.KeywordsArrival RateService RateCongestion ControlReward RateSwitching CurveThese keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

PWLM[formula omitted]-based automatic performance model estimation method for HDFS write and read operations

With the astronomical growth in online presence vis-a-vis the IT industry across the globe, there is an urgent need to evolve cloud based DataCenter architectures that can rapidly accommodate web application integrations which can serve the energy industries, educational sector, finance sector, manufacturing sector, etc. Previous works in DataCenter domain have not investigated on cloud based DataCenter Servercentric characteristics for evaluation studies. This paper then proposed a Reengineered DataCenter (R-DCN) architecture for efficient web application integration. In this regard, attempt was made to address network scalability, efficient and distributed routing, packets traffic control, and network security using analytical formulations and behavioural algorithms having considered some selected architectures. In this work a simulation experiment was carried out to study six key performance metrics for all the architectures. It was observed that the network throughput, fault-tolerance/network capacity, utilization, latency, service availability, scalability and clustering effects of R-DCN responses with respect to above Key Performance (KPIs) metrics were satisfactory when compared with BCube and DCell architectures. Future work will show a detailed validation using a cloud testbed and CloudSim Simulator.

In 2017, Uber announced it had a “broken relationship” with its drivers. Yet what broke the relationship and what effect might it have on Uber’s organizational performance? This paper is a mixed-methods study of how conflict impacts the relationship between platforms and drivers. First, by drawing on 55 original interviews with rideshare drivers, this paper identifies conflict triggers that damage the relationship between platforms and drivers. Second, this paper uses new time diary data from 490 Uber, Lyft, Juno, and other Transportation Network Company (TNC) drivers from across the United States to empirically test if these conflict triggers are associated with key TNC performance metrics. Specifically, this paper finds that a “broken relationship” is associated with drivers withholding their working time from Uber and allocating it toward their competitor, Lyft. Additionally, this paper finds that drivers who have a broken relationship with Uber are more likely to recruit (‘steer’) passengers toward Lyft. These behaviors provide a link between organizational conflict and key performance metrics in the ‘gig economy.’

In this study, we conducted a comprehensive DC and AC analysis of Gate-All-Around (GAA) devices, focusing on the implications of increased gate capacitance both for wider and narrower device on key performance metrics. Our findings reveal that the elevated gate capacitance due to thin GaN cap layer significantly affects the transconductance (gm), Drain-Induced Barrier Lowering (DIBL), subthreshold swing (SS) and the on/off current ratio (Ion/Ioff). A new analytical linkage is formulated connecting gate capacitance ( C g ) with key DC performance metrics. The results indicated a significant enhancement in both ft and fmax, attributed to the high Ion/Ioff ratio of 1.72 × 10 13 , confirming the superior high-frequency performance of GAA structures. Our analysis underscores the critical role of gate capacitance in optimising the electrical characteristics of GAA devices, paving the way for their application in 5G wireless communication.

This work reduces the cost of automatic performance modeling and enables comparative analysis of workflow orchestration performance. We use parametric profiles to cut measurement costs, a method to identify performance-irrelevant options, and a serverless workflow benchmarking suite based on a Petri Net model. These contributions help developers and scientists better understand, model, and optimize application performance across diverse platforms and configurations.

A performance model for a parallel I/O system is essential for detailed performance analyses, automatic performance optimization of I/O request handling, and potential performance bottleneck identification. Yet how to build a portable performance model for parallel I/O system is an open problem. In this paper, we present a machine-learning approach to automatic performance modeling for parallel I/O systems. Our approach is based on the use of a platform-independent performance metamodel, which is a radial basis function neural network. Given training data, the metamodel generates a performance model automatically and efficiently for a parallel I/O system on a given platform. Experiments suggest that our goal of having the generated model provide accurate performance predictions is attainable, for the parallel I/O library that served as our experimental testbed on an IBM SP. This suggests that it is possible to model parallel I/O system performance automatically and portably, and perhaps to model a broader class of storage systems as well.

The primary purpose of the current paper is to design a fast and accurate performance model framework for exploring various thread-to-core mapping strategies (MS) and estimating steady state cycles per instruction (CPI). It is directed towards efficiently exploring these performance metrics for large parallel applications for shared memory multicores. This work establishes a hybrid Markov Chain Model (MCM) and Model Tree (MT) based system-level performance prediction model framework. The model is validated with an Electromagnetics application for 12 different mapping strategies. The average performance prediction error is 0.168% with standard deviation of 3.866%. The total run time of model is of the order of minutes, whereas the actual application execution time is in terms of several days.

The increasing demand for the coverage of high‐speed wireless local area networks (WLANs) is driving the installation of a very large number of access points (APs). Wireless mesh networks (WMNs) have emerged as a promising technology in the next generation networks to provide economical and scalable broadband access to wireless interconnection of multiple APs that manage individual WLANs in order to extend the coverage of the conventional single WLANs. In this paper, we develop a new analytical model to investigate the quality‐of‐service (QoS) performance metrics in WMNs interconnecting multiple WLANs. The model takes the phenomenon of communication locality into consideration. Extensive simulation experiments are conducted to validate the accuracy of the analytical model. The proposed model is then used to obtain the maximum achievable throughput, which is a key performance metrics in wireless networks. Moreover, we utilize the model to evaluate the impact of communication locality on network performance and investigate its effects on the optimization of integrated WLANs and multi‐hop mesh networks. Copyright © 2009 John Wiley & Sons, Ltd.

Objectives:The primary objective was to evaluate the impact of ulnar collateral ligament reconstruction (UCLR) on MLB pitcher performance, with a specific focus on advanced metrics such as fastball velocity, spin rate, FIP (fielding independent pitching), SIERA (skill interactive earned run average), and WHIP (walks and hits per inning pitched) when compared to both pre-injury performance and a non-injured cohort of pitchers. The secondary objective was to assess the extent to which velocity and spin rate are predictive of pitcher performance. The hypotheses for the primary objective was that there would be no difference in pre-injury and post-injury performance or in post-injury performance and healthy controls. The hypothesis for the secondary objective was that both velocity and spin rate would be predictive of performance regardless of injury status.Methods:Pitchers with confirmed UCL injuries between the 2017 and 2021 MLB seasons were identified using the Pro Sports Transactions Archive and baseball-reference.com. Inclusion criteria required pitchers to have thrown at least 8.0 innings in two consecutive seasons both pre- and post-injury. A control group of healthy pitchers was age-matched at a 1:2 ratio to injured pitchers (Fig 1). Key performance metrics, including FIP, SIERA, and WHIP, were extracted from fangraphs.com, and spin rate and velocity data were collected from Baseball Savant.Principal component analysis (PCA) was used to compress several pitching performance metrics (FIP, SIERA, WHIP) into a single, comprehensive performance measure, referred to as the first principal component (PC1), where lower values indicate better overall performance. The changes in performance before and after surgery were normalized to account for age-related decline, which was controlled for by using an age-matched group of healthy control pitchers. Pearson’s correlation coefficients were calculated to assess the relationship between spin rate, velocity, and performance. Comparisons between pre- and post-surgery performance and between injured and control pitchers were conducted using independent t-tests. A power analysis was conducted to ensure a sufficient sample size to detect meaningful differences in performance outcomes, and all analyses were performed using Python 3.7 and RStudio 2023. Statistical significance was set at α = 0.05.Results:The study included 34 MLB pitchers who had undergone UCLR, with an average age at the time of injury of 27.03 ± 3.05 years. Age-matched controls (n=68) were identified for comparison, allowing for the analysis of both performance differences and the potential influence of aging on pitching metrics. Performance was first analyzed in terms of pitching volume (number of pitches and innings pitched). While the injured group showed a slight decline in innings pitched post-surgery, this difference was not statistically significant when compared to the control group (p = 0.301). Furthermore, no significant differences were found in strikeouts (p= 0.992) or hits allowed (p= 0.207) at two seasons before versus after injury. The PC1 analysis of FIP, SIERA, WHIP revealed no significant difference between injured pitchers post-surgery compared to their control counterparts (p = 0.287) (Fig 2).Spin rate and velocity were further analyzed to determine their relationship with post-surgical performance. Fastball velocity showed no significant change post-surgery (p = 0.687), and spin rate also did not significantly differ between injured and control pitchers (p = 0.876). However, both spin rate and velocity were identified as a key predictors of performance for both groups for WHIP (pspin= 0.02, pvelo= 0.04), FIP(pFIPspin= 0.003, pFIPvelo< 0.001), SIERA (pSIERAspin< 0.001, pSIERAvelo< 0.001), and PC1 (pPC1spin< 0.001, pPC1velo< 0.001) (Fig 3). There were no significant relationship between age at the time of injury and changes in performance, spin rate, or velocity, indicating that age did not influence post-operative outcomes (Pearson’s r = -0.072, p = 0.685).Conclusions:Following UCLR, MLB pitchers maintain their pre-injury level of performance, with no significant decline in fastball velocity, spin rate, or advanced statistics post-surgery. Additionally, both spin rate and velocity emerged as significant predictors of pitching performance across both injured and healthy pitchers, highlighting their potential utility as a key metric in assessing pitching effectiveness. These findings support the conclusion that UCLR is an effective intervention for MLB pitchers, allowing them to return to competitive play without significant deterioration in their key performance metrics.

Climate modeling efforts often grapple with two overarching categories of forecast errors: systematic errors, arising from a model's inherent inability to fully capture complex physical processes, and accumulated errors, which emerge as inaccuracies compound over iterative predictions. In this paper, we correct both error types by introducing DecGAN, a novel multistep generative adversarial network designed for end‐to‐end bias correction and high‐resolution downscaling of numerical models. Leveraging the Community Earth System Model (CESM), we employ observational and reanalysis data from Hadley Centre Sea Ice and Sea Surface Temperature (HadiSST) ( https://www.metoffice.gov.uk ) and European Centre for Medium‐Range Weather Forecasts Reanalysis v5 (ERA5) ( https://cds.climate.copernicus.eu ) as reference labels to train our model and apply it to the Pacific sea surface temperature (SST) and Asian precipitation for evaluation. Comprehensive results demonstrate that DecGAN substantially reduces the average prediction errors over 12 consecutive months by 77.5% for Pacific SST and 18.8% for Asian precipitation, while significantly enhancing the structural fidelity of the forecasts as evidenced by Structural Similarity Index (SSIM) increases of 17.63% and 2.41%, respectively. Beyond correcting biases, DecGAN also downscales predictions from a coarse spatial resolution of 1° × 1° to 0.25° × 0.25° for SST, and 2.5° × 2.5° to 0.25° × 0.25° for precipitation, yielding high‐fidelity maps crucial for regional impact assessments. Our error analysis reveals that accumulated errors are strongly influenced by the model's ability to simulate internal variability, whereas systematic biases primarily stem from external forcings that the base model struggles to represent accurately. This distinction underscores the need to advance both the physics‐based core of numerical modeling and the data‐driven corrective strategies to improve SST forecasts. Furthermore, given El Niño–Southern Oscillation (ENSO)'s far‐reaching impacts on Asian precipitation via atmospheric circulation anomalies, incorporating rigorous bias correction and downscaling is pivotal for more reliable long‐term precipitation projections. By bridging numerical model outputs with deep‐learning innovations, DecGAN paves the way for a new generation of climate prediction systems that are both more accurate and spatially refined.

Call centre employees experience of mental game playing during work breaks were recorded in this study. The qualitative study involving two focus group discussions at different call centres, with 6 employees in one focus group and 9 in the second, stated employee's reasons for variations in the key performance metric by which they were being evaluated. In this case the key performance metric was the average speed of answer. In a previous study, where the mental games were introduced as an intervention during work breaks, the participants were asked to provide feedback on variations in productivity during a 4-week intervention study. Volunteer sampling was utilized, and the information was categorized into codes. Similar codes were grouped into themes. It was found that players are efficient in the middle of the week either because daily routine settles middle of the week or because employees try to average their metrics related to time. They also thought that productivity increased towards the end of the intervention because of games being accepted in the routine by that time, because the focus shifted from fun to work towards the end or because the employees knew they were being evaluated. According to participants, the non-players were more efficient than the players because games served as a bad distraction for the players and that players were considering games as something to do rather than an actual break. As a conclusion, it felt important to improve on the timings and frequency of the games played during work before it can be considered an effective intervention.

The emergence of micro frontend architectures has revolutionized the way organizations approach frontend application development, enabling distributed teams to work independently while maintaining system consistency. However, performance optimization in these distributed systems presents unique challenges that differ significantly from traditional monolithic approaches. This study examines performance strategies for micro frontend-based applications through a comprehensive analysis of 30 applications across six key performance metrics. The research reveals significant performance variations across micro frontend implementations, with bundle sizes ranging from 345KB to 550KB and API response times ranging from 155ms to 300ms. Our analysis demonstrates strong correlations between optimization strategies and application performance, particularly highlighting the critical role of lazy loading implementations. Applications achieving lazy loading rates above 50% consistently outperformed those below 40%, with performance score improvements of up to 37 points. The study uses XGBoost regression models to predict key performance metrics, identifying challenges in CPU usage prediction due to overfitting concerns, while achieving exceptional accuracy for bundle size prediction (R² = 0.9647).The performance patterns indicate that successful micro frontend applications require integrated optimization across multiple dimensions, including composition strategies, dependency management, and inter-service communication protocols. The research identifies threshold values for optimal performance, including maintaining bundle sizes below 400KB and implementing aggressive lazy loading strategies. These findings provide actionable insights for development teams working with micro frontend architectures, providing data-driven guidance for architectural decisions and performance strategies in distributed frontend systems.Objective: This study examines performance optimization in micro-frontend-based applications. It analyzes 30 settings across six key metrics, focusing on bundle size, lazy loading, CPU utilization, and response times. The research highlights optimization constraints and a predictive model to guide architectural decisions for scalable, efficient distributed frontends using XGBoost regression.

Breast cancer remains a major healthcare challenge, highlighting the need for early and accurate diagnosis. Shear-wave elastography (SWE), an ultrasound-based imaging technique that quantifies tissue elasticity, has emerged as a promising tool. Recent studies suggest that SWE may provide additional diagnostic value when used alongside conventional imaging methods. This study aimed to assess the diagnostic performance of SWE when combined with conventional ultrasound and magnetic resonance imaging (MRI) in the evaluation of breast lesions. This retrospective study included patients with breast lesions who underwent SWE, conventional ultrasound and MRI. The diagnostic performance of each modality was evaluated individually and in combination. Histopathological results served as the gold standard for diagnosis. Key performance metrics - sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and overall accuracy - were calculated for each imaging approach. A total of 99 patients were included in the study, comprising 64 with benign lesions and 35 with malignant lesions. Malignant lesions were generally larger and exhibited distinct imaging characteristics across ultrasound, SWE and MRI. When assessed individually, SWE, ultrasound and MRI showed comparable diagnostic accuracy (64.6%, 62.6% and 62.6%, respectively). However, combining all 3 modalities significantly improved diagnostic performance, yielding sensitivity, specificity, PPV, NPV, and overall accuracy of 94.3%, 89.1%, 82.5%, 96.6%, and 90.9%, respectively (p < 0.001). The area under the curve (AUC) for the combined approach was significantly higher than for any single modality (0.917 vs 0.642, 0.627 and 0.633; p < 0.001). While SWE alone offers diagnostic performance comparable to that of ultrasound and MRI individually, its greatest value lies in combination with these imaging modalities. Integrating ultrasound, SWE and MRI significantly enhances diagnostic accuracy, sensitivity and specificity, offering a promising multimodal approach for more reliable differentiation between benign and malignant breast lesions.

Redox electrodialysis is a promising platform for energy-efficient freshwater production, particularly within the brackish water salinity regime. However, operating at seawater-level salinities presents new challenges. Large osmotic pressure gradients develop between process streams, driving undesired transmembrane water transport, which reduces overall device efficiency by decreasing the water recovery and increasing the volumetric energy demand. This work investigates the impact of water transport on redox electrodialysis performance and explores strategies to minimize its effects. To decrease undesired water transport, we synthesized crosslinked polyphenyl alkylene anion exchange membranes with varied crosslink densities and crosslinker structures. Using ex-situ H-cell experiments, we characterized membrane properties such as water and salt permeabilities, benchmarking them against commercially available anion exchange membranes. These membranes were then used in a redox electrodialysis flow cell operating with seawater-relevant feed concentrations. We quantified and compared key performance metrics – water recovery and volumetric energy demand – between synthesized and commercial membranes, highlighting pathways to improve redox electrodialysis efficiency under high-salinity conditions.