Optimizing OpenCL Barrier Synchronization and Memory Efficiency on Multi-Core DSPs

Sharing data among asynchronous processes is considered to be a hard systems problem in multithreaded modern shared-memory multicore systems. Throughout the literature, multiple solutions have been proposed, like the so-called barrier synchronization. A Barrier is a synchronization primitive that provides guarantees that any thread will not continue execution from a given point until all threads have reached that point. This primitive is widely used in different parallel programming models, but it can easily become a hot-spot for performance critical applications due to its global nature as one preempted thread will stop execution of all other threads waiting at the barrier. This paper suggests a technique to change the global nature of barrier synchronization into a non-blocking synchronization model with lock-free thread progression guarantees. The main idea is to exploit algorithm-based memory access patterns to implement self-synchronizable threads to protect concurrent reads and writes in a shared data structure without explicit use of a barrier primitive. To the best of our knowledge, this is the first attempt to provide a different synchronization mechanism based on the algorithm intrinsic characteristics rather than an explicit use of a global barrier in shared-memory architectures. Our experimental results show factors of performance improvement against its global barrier-based algorithm counterpart.

Leveraging large data sets, deep Convolutional Neural Networks (CNNs) achieve state-of-the-art recognition accuracy. Due to the substantial compute and memory operations, however, they require significant execution time. The massive parallel computing capability of GPUs make them as one of the ideal platforms to accelerate CNNs and a number of GPU-based CNN libraries have been developed. While existing works mainly focus on the computational efficiency of CNNs, the memory efficiency of CNNs have been largely overlooked. Yet CNNs have intricate data structures and their memory behavior can have significant impact on the performance. In this work, we study the memory efficiency of various CNN layers and reveal the performance implication from both data layouts and memory access patterns. Experiments show the universal effect of our proposed optimizations on both single layers and various networks, with up to 27.9× for a single layer and up to 5.6× on the whole networks.

This chapter provides solutions for increasing the efficiency of data transfer in modern computer network applications and computing network environments based on the TCP/IP protocol suite. In this work, an imitation model and simulation was used as the basic method in the research. A simulation model was developed for designing and analyzing the computer networks based on TCP/IP protocols suite which fully allows the exact features in realizing the protocols and their impact on increasing the efficiency of data transfer in local and corporate networks. The method of increasing efficiency in the performance of computer networks was offered, based on the TCP/IP protocols by perfection of the modes of data transfer in them. This allows an increased efficient usage of computer networks and network applications without additional expenditure on infrastructure of the network. Practically, the results obtained from this research enable significant increase in the performance efficiency of data transfer in the computer networks environment. An example is the “Donetsk National Technical University” network.

The development of smallholder irrigation is a policy priority in Ethiopia, yet little consideration has been given to the role state interventions play in enabling or constraining effective self‐governance in farmer‐managed schemes. To address this gap in evidence, research was conducted in Tigray, northern Ethiopia, in two irrigation sites. Focus group discussions and semi‐structured interviews with key informants were the primary means of collecting data on local institutional arrangements, government interventions and management challenges. An opinion poll provided additional insights. Our case studies indicate that effective self‐governance cannot easily be crafted by external actors. The introduction of modern irrigation technologies and formalized management arrangements has fostered dependency on external support and increased state influence in key decision‐making processes, whilst traditional water user associations remain largely autonomous but have little access to financial resources. State interventions have also been poorly tailored to local context with negative consequences for performance, although there have been some successes where traditional arrangements for water management have been integrated into new institutions. The incorporation of farmer perspectives into performance assessments could facilitate more responsive interventions, and remains an important area of irrigation research and practice. Copyright © 2017 John Wiley & Sons, Ltd.

A Robust Video Compression Algorithm for Efficient Data Transfer of Surveillance Data - written by Kuldeep Kaur , Baldeep Kaur published on 2020/08/01 download full article with reference data and citations

In Huascarán National Park (HNP), Peru, grazing and anthropogenic burning have been interacting for decades with natural ignitions and climate variability to reconfigure the fire regimes of the vegetative communities. However, over the last few decades, human alterations to the region’s fire regime were perceived by resource managers to have led to an overall increase in fire occurrence and fire extent. Resource managers are now very concerned about the impacts of increasing anthropogenic fires in the National Park because these fires seem to disrupt ecological processes and tourism. To test these beliefs and examine the fire activity from 2002 to 2014, we used a time series of several different data products including the MODIS Terra and Aqua platforms (Active Fire and Burned Area) as well as local air temperature and precipitation data. The intra-annual distribution of fires in the HNP and its buffer zone (BZ) showed a consistent increase in the number of active fires and burned areas during the dry season (Peruvian winter). The active fire anomalies showed higher fire activity in the HNP over the period 2002 to 2014. However, burned area anomalies showed consistently higher fire activity in the BZ during the high fire years in 2003, 2005, 2010, and 2014. The spatial distribution of areas burned in the HNP and the BZ from 2002 to 2014 showed that most recurrent burning was inside the BZ, which is in compliance with the HNP’s policy but in contradiction with local resource managers’ perceptions. Although human activity may influence local fire dynamics, fire responses seemed to be mainly influenced by local climate oscillations. Our results suggest a clear fire response to local climate and, in particular, to precipitation variability. However, it appears that there was a low fire return interval for any given location and that the total area burned for a given year is relatively small. Therefore, small-scale fires created an appearance of larger fire impacts than actually occur within the HNP and its BZ. In this instance, fire perception and fire reality are not aligning and the new challenge for resource managers is how best to reconcile these two factors to more effectively manage the parklands.

The 2004 Indian Ocean Tsunami affected Indonesia with massive devastation and suffering. It ignited the largest international humanitarian response and recovery in history. Academicians, responders, civil servants, teachers, and all professions were awakened to the urgency of disaster risk reduction, mitigation, preparedness, and readiness. Indonesia then instituted disaster management law, established national, provincial, and local disaster management organizations, and started developing a disaster management workforce. The United States of America (USA) supported Indonesia’s 2004 tsunami response and recovery and continues to assist in capabilities enhancement. This research aims to explore how knowledge exchange through training, education, policy improvement, system development, and organizational management contributes to the whole society’s capacity. Our method includes a literature review, focus group discussions, and site visits to explore collaborations between governments, private sectors, civil societies, non-governmental organizations, and regular citizens to generate improved disaster preparedness and readiness. Recent US-Indonesia comparative study missions offer lessons and recommendations for urban planning and green infrastructure, nature-based solutions, and readiness progression. Major findings through comparative lessons gained include the importance of developing incident management assistance team, supplemental response team, community emergency response team, regional office, biochemical hazard training, national interagency fire management, hazard and social vulnerability risk mapping, and incident workforce academy. New knowledge enrichment brings excitement and expectations, requiring competent human resource, financial, and policy investments. Capacity building must increase equity and resilience planning for vulnerable communities across the islands as Indonesia increases its regional disaster response contribution. Future implications would include prioritized efforts for short, medium, and long-term transformation in improving disaster management in Indonesia and continuing to collaborate with the USA in nurturing sustainability and resilience despite complex challenges in disaster risk management.

Problem: Private approaches to urban service provision are becoming more popular. Some argue these approaches are more efficient and more democratic than government provision because they are voluntary. While these club approaches can shift the burden of infrastructure finance to direct groups of users, they can also fragment urban service delivery and justify unevenness in service quality across the city. Purpose: This article explores examples of club goods, that is, common interest developments (CIDS) for housing, business improvement districts (BIDs) for commercial areas, and economic development zones (EDZs) for commercial and industrial areas, and assesses their implications for local government. Emphasis is given to how clubs internalize benefits to members of the club but shed externalities onto the broader local government system. A critical governance concern is the impact on the long-term ability of local government to coordinate across disparate elements and interests in the community. Methods: I discuss three types of clubs ranging from totally private common interest developments (home owners associations), to partially private business improvement districts, to totally public economic development zones. These club types are analyzed in terms of economic benefits, externalities, governance structure, and broader concerns with equity and sustainability. Results and conclusions: Club good approaches to urban infrastructure delivery enhance private investment and reduce costs to cities, but they also shed externalities onto the broader city. Although these clubs are often private associations, this article shows how they are critically supported by government, beyond the property rights requirement assumed by most theorists. A further concern is that club approaches may undermine support for equity and redistribution at the broader city level. Takeaway for practice: Local governments are under pressure to provide public goods efficiently and engage private voluntary approaches whenever possible. The efficiency and popularity of club approaches derives in part from their ability to capture the benefits of increased investment for internal benefit. However, local government managers also must manage diversity and build public support for investment to ensure equity across the urban territory. Balancing the benefits of club goods with the need for broader urban integration is a key challenge for planners and urban managers. Research support: None.

Local knowledge is found in communities and consists of a set of accumulated experiences that provide information regarding a community's cultural and behavioural characteristics. This study therefore aims to Role and Challenges of Local Knowledge Management among the Minangkabau Community. The study used qualitative instruments with 12 leaders, namely: traditional leaders, religious leaders, and community leaders. Qualitative research data analysis is through NVIVO 14. As for the qualitative findings from structured interviews, the role of local knowledge management through maintaining cultural identity, sustainable use of resources, solves local problems, and Local economic development. Challenges of local knowledge management among the Minangkabau community are oral tradition, information storage media, modernization, acculturation, knowledge owner, and foreign culture. This research will contribute to communities' local knowledge management for future generations to understand and appreciate traditional knowledge and generate the Minangkabau community's socio-economy and knowledge development.

Is sustainable exploitation of coral reefs possible? A view from the standpoint of the marine aquarium trade

This chapter describes the growth of South East Queensland since 2000 before charting chronologically the various attempts since the beginning of this century to meet the challenges of growth management through processes of metropolitan and regional planning. It focuses on the political and institutional pressures and problems associated with developing collaborative arrangements between state and local governments and consider the consequences of a shifting balance of power and responsibility between these levels. The Queensland divisions of the Planning Institute of Australia and the Urban Development Institute of Australia and, importantly, the local media began to campaign for change, alongside the Brisbane Institute under the leadership of urban historian Peter Spearritt. Regional and indeed local growth management strategies such as the South East Queensland Regional Plan must create an environment in which growth is encouraged, but then regulated in ways that are widely accepted as reasonable and sensible.

Energy efficient multipath data transfer scheme to mitigate false data injection attack in wireless sensor networks

Capacity and Funding: Examining Local Disaster Risk Reduction and Management (DRRM) Gaps, Challenges, and Solutions for High-Risk, Low-Income (LGUs)

As a processing technology of data and information, building information modeling (BIM) is often used for producing digital engineering model and promoting interoperability among participants involved in the Architecture, Engineering, and Construction (AEC) industry. The real values of BIM largely depend on effective information integrating and sharing among different stages or disciplines through the whole building lifecycle. To this day, although BIM has made a substantial contribution to the Chinese AEC industry, there are still some debates on its roles. Moreover, due to the complexity of structural design process, the weak link between structural model and BIM model makes its deep application even harder. To address these issues, this paper discusses the roles of BIM from the perspective of structural engineers through comparing with the traditional CAD. Then, the BIM structural model is studied by analyzing the model conformation, design mode, and Industry Foundation Classes (IFC)-format structural model. By comparing the differences between BIM physical model and structural mechanical model, an indirect method for the data transformation from BIM model to structural analysis model isproposed. An interface is developed based on the analysis of data formats and mapping rules. Three typical cases are adopted to demonstrate the data transfer efficiency by using the interface. The case study shows that the proposed method achieves higher efficiency for the data transformation from BIM model to structural analysis model.

Local area ring networks operate without causing collision, and without station competition. It is quite opposite to currently the most popular technology, Ethernet network, where collisions may occur in a large amount. These two features of the ring networks provide significantly better performance and greater amount of data that can be effectively transmitted in the same time unit. This is especially true during high network load. This paper deals with the comparison of the efficiency of the two ring networks: Token Ring and FDDI (Fiber Distributed Data Interface), taking into account the access mechanisms to network stations. In relation to this issue general information about the efficiency of the Ethernet is also described.