This paper optimizes the Convolutional Neural Network (CNN) algorithm using high-performance computing (HPC) technologies. It uses multi-core processors, GPUs, and parallel computing frameworks like OpenMPI and CUDA to speed up CNN model training. The approach improves performance and training time and is superior to alternative strategies. The study demonstrates how HPC technologies can refine the CNN method, resulting in faster and more accurate training of large-scale CNN models.
There have been several reductions from multivalued consensus to binary consensus over the past 20 years. To the best of our knowledge, none of them solved it for Byzantine asynchronous settings. In this paper, we close this gap. Moreover, we do so in subquadratic communication, using newly developed subquadratic binary Byzantine Agreement techniques.
Lachesis protocol~\cite{lachesis2021} leverages a DAG of events to allow nodes to reach fast consensus of events. This work introduces DAG progress metrics to drive the nodes to emit new events more effectively. With these metrics, nodes can select event timing and can choose previous events as parents for their own new events. Our results show that our event timing and parent selection methods can help reaching consensus quicker and thus can reduce lower time to finality significantly.
Authorization currently introduces partial centralization in otherwise distributed network architectures, such as ICN approaches. Analyzing existing work in (partially) distributed authentication and authorization, and rearranging proven methods, this paper introduces a generalized, capability based and fully distributed authorization scheme. It argues that such a scheme can fit neatly into ICN architectures in order to enhance the trust model and mitigate against certain classes of denial-of-service attacks. Keywords: authorization, distributed systems security, ICN
Neighbor search is of fundamental important to many engineering and science fields such as physics simulation and computer graphics. This paper proposes to formulate neighbor search as a ray tracing problem and leverage the dedicated ray tracing hardware in recent GPUs for acceleration. We show that a naive mapping under-exploits the ray tracing hardware. We propose two performance optimizations, query scheduling and query partitioning, to tame the inefficiencies. Experimental results show 2.2X -- 65.0X speedups over existing neighbor search libraries on GPUs. The code is available at https://github.com/horizon-research/rtnn.
Angelo Capossele, Sebastian Mueller, Andreas Penzkofer
This paper investigates leaderless binary majority consensus protocols with low computational complexity in noisy Byzantine infrastructures. Using computer simulations, we show that explicit randomization of the consensus protocol can significantly increase the robustness towards faulty and malicious nodes. We identify the optimal amount of randomness for various Byzantine attack strategies on different kinds of network topologies.
A data store allows application processes to put and get data from a shared memory. In general, a data store cannot be modelled as a strictly sequential process. Applications observe non-sequential behaviours, called anomalies. The set of pos- sible behaviours, and conversely of possible anomalies, constitutes the consistency model of the data store.
This paper presents a simple generalization of causal consistency suited to any object defined by a sequential specification. As causality is captured by a partial order on the set of operations issued by the processes on shared objects (concurrent operations are not ordered), it follows that causal consistency allows different processes to have different views of each object history.
The XRP Ledger Consensus Protocol is a previously developed consensus protocol powering the XRP Ledger. It is a low-latency Byzantine agreement protocol, capable of reaching consensus without full agreement on which nodes are members of the network. We present a detailed explanation of the algorithm and derive conditions for its safety and liveness.
In this paper we recommend the use of multi-channel for XML data in wireless broadcasting. First we divide XML data into information units as bucket, then extract path information (XPath) for any unit and build an index tree from the data path. Finally, make wireless data stream with merging parts of index tree and parts of XML data in multichannel XML. Then, create a protocol that allows mobile users access to the wireless XML stream generated with our method. We study 11 channels in server side and 3 orthogonal channels in client side.
Interprocessor communication often dominates the runtime of large matrix computations. We present a parallel algorithm for computing QR decompositions whose bandwidth cost (communication volume) can be decreased at the cost of increasing its latency cost (number of messages). By varying a parameter to navigate the bandwidth/latency tradeoff, we can tune this algorithm for machines with different communication costs.
This paper presents a fully distributed resource discovery and reservation system. Verification of such a system is important to ensure the execution of distributed applications on a set of resources in appropriate conditions. A semi-formal model for his system is presented using probabilistic timed automata. This model is timed, parametric and probabilistic, making it a challenge to the parameter synthesis community.
We posit that striving for distributed systems that provide "single system image" semantics is fundamentally flawed and at odds with how systems operate in the physical world. We realize the database as an optimization of this system: a required, essential optimization in practice that facilitates central data placement and ease of access to participants in a system. We motivate a new model of computation that is designed to address the problems of computation over "eventually consistent" information in a large-scale distributed system.
In this paper we introduce algorithms for the construction of scale-free networks and for clustering around the nerve centers, nodes with a high connectivity in a scale-free networks. We argue that such overlay networks could support self-organization in a complex system like a cloud computing infrastructure and allow the implementation of optimal resource management policies.
The main open question is how to calculate the effect of switching between frequencies in DVFS technique on the lifetime of the cluster components. As moving from one frequency to another in DVFS technique always gives a shock to the component and consequently decreases the component lifetime, therefore, it becomes interesting to answer the question of how fast a component can change its speed in order to decrease power without changing its lifetime.
Tasks scheduling is the most challenging problem in the parallel computing. Hence, the inappropriate scheduling will reduce or even abort the utilization of the true potential of the parallelization. Genetic algorithm (GA) has been successfully applied to solve the scheduling problem. The fitness evaluation is the most time consuming GA operation for the CPU time, which affect the GA performance. The proposed synchronous master-slave algorithm outperforms the sequential algorithm in case of complex and high number of generations problem.
This article deals with some stochastic population protocols, motivated by theoretical aspects of distributed computing. We modelize the problem by a large urn of black and white balls from which at every time unit a fixed number of balls are drawn and their colors are changed according to the number of black balls among them. When the time and the number of balls both tend to infinity the proportion of black balls converges to an algebraic number. We prove that, surprisingly enough, not every algebraic number can be "computed" this way.
This volume represents the proceedings of the 5th Workshop on Membrane Computing and Biologically Inspired Process Calculi (MeCBIC 2011), held together with the 12th International Conference on Membrane Computing on 23rd August 2011 in Fontainebleau, France.