On the Performance of Snake Partitioning: A Data Decomposition Technique that Reduces Communication and Exploits Locality
نویسنده
چکیده
This paper presents performance results of a new data partitioning technique: snake partitioning , a data decomposition technique than can be derived at compile-time. Snake partitioning is suitable for multidimensional arrays with restricted aane references. The technique derives the data partitioning of these arrays and an execution order that exploits locality in loops. Experiments that compare the performance of the snake data partitioning against traditional data partitioning techniques like row, column and blocks are presented.
منابع مشابه
Blind Separation of Jointly Stationary Correlated Sources
The separation of unobserved sources from mixed observed data is a fundamental signal processing problem. Most of the proposed techniques for solving this problem rely on independence or at least uncorrelation assumption for source signals. This paper introduces a technique for cases that source signals are correlated with each other. The method uses Wold decomposition principle for extracting ...
متن کاملScalability Analysis of Partitioning Strategies forFinite Element Graphs
Issues of partitioning Finite Element Graphs are central to parallel formulations of the Finite Element Method. Due to the nature of the problem, optimal partitioning schemes should conform to three basic criteria: equal load on all processors; locality of communication; and maximum computation to communication ratio associated with each partition. Many techniques have been presented in literat...
متن کاملEEG Artifact Removal System for Depression Using a Hybrid Denoising Approach
Introduction: Clinicians use several computer-aided diagnostic systems for depression to authorize their diagnosis. An electroencephalogram (EEG) may be used as an objective tool for early diagnosis of depression and controlling it from reaching a severe and permanent state. However, artifact contamination reduces the accuracy in EEG signal processing systems. Methods: This work proposes a no...
متن کاملDistributed RDF Query Answering with Dynamic Data Exchange
Evaluating joins over RDF data stored in a shared-nothing server cluster is key to processing truly large RDF datasets. To the best of our knowledge, the existing approaches use a variant of the data exchange operator that is inserted into the query plan statically (i.e., at query compile time) to shuffle data between servers. We argue that such approaches often miss opportunities for local com...
متن کاملMulti-Variable Agents Decomposition for DCOPs
Current DCOP algorithms suffer from a major limiting assumption—each agent can handle only a single variable of the problem—which limits their scalability. This paper proposes a novel Multi-Variable Agent (MVA) DCOP decomposition, which: (i) Exploits co-locality of an agent’s variables, allowing us to adopt efficient centralized techniques; (ii) Enables the use of hierarchical parallel models, ...
متن کامل