MAi: Memory Affinity interface
نویسندگان
چکیده
In this document, we describe an interface called MAI. This interface allows developers to manage memory affinity in NUMA architectures. The affinity unit in MAI is an array of the parallel application. A set of memory policies implemented in MAI can be applied to these arrays in a simple way. High-level functions implemented in MAI minimize developers work when managing memory affinity in NUMA machines. MAI’s performance was evaluated on two different NUMA machines using three different parallel applications. Results obtained with MAi present important gains when compared with the standard memory affinity solutions. Key-words: multi-core architecture, memory affinity, static data, performance evaluation, programming in ria -0 03 44 18 9, v er si on 6 14 J un 2 01 0 MAi: Interface pour la Affinité de Mémoire Résumé : Ce document décrit MAI, une interface pour contrôler l’affinité de mémoire sur des architecture NUMA. L’unité d’affinité utilisé par MAI est un tableau de l’application. MAI dispose d’un ensemble de politiques de mémoire qui peuvent être appliquées pour le tableau du une application d’une façon simple. Fonctions de haut-niveau mis en œuvre en minimiser les travaux des programmeurs. Les performance de MAI sont évaluée sur deux différents NUMA machines utilisant trois applications parallèles. Les résultats obtenus avec cette interface présente d’importants gains par rapport aux solutions standard. Mots-clés : architectures NUMA, affinité mémoire, données statiques, étude de performances in ria -0 03 44 18 9, v er si on 6 14 J un 2 01 0
منابع مشابه
Evaluation of Shared Memory System with OpenMP Applications in λ Computing Environment
Our research group has proposed a new high performance computing architecture, which we call the λ computing environment. In the λ computing environment, every computing node has a shared memory connected with the optical wavelength path. In this paper, we execute an application of OpenMP, Parallel Programming API, and evaluate the performance of the λ computing environment. The results show an...
متن کاملDesign and Implementation of OpenMP Library for λ Computing Environment
Our research group has proposed a new high performance computing architecture, which we call the λ computing environment. In the λ computing environment, every computing node has a shared memory connected with the optical wavelength path. In this paper, we establish the λ computing environment using the AWG–STAR system which is based on WDM technology, and we present the design and implementati...
متن کاملProceedings of the 3 rd USENIX Windows NT Symposium
We have built a 16-way, ccNUMA multiprocessor prototype to study the feasibility of building large scale servers out of Standard High Volume (SHV) components. Using a cache-coherent interconnect, our prototype combines four 4-processor SMPs built using 350MHz Intel Xeon processors, yielding a 16-way system with a total of 4 GBytes of physical memory distributed over the nodes. Such an environme...
متن کاملLIKWID: Lightweight Performance Tools
Exploiting the performance of today’s microprocessors requires intimate knowledge of the microarchitecture as well as an awareness of the ever-growing complexity in thread and cache topology. LIKWID is a set of command line utilities that addresses four key problems: Probing the thread and cache topology of a shared-memory node, enforcing thread-core affinity on a program, measuring performance...
متن کاملPerformance of Multistage Detectors in Synchronous CDMA Mobile Communication System
Multiuser detection (MUD) is central to the fulfillment of the capabilities of code-division-multiple access (CDMA), which is becoming the ubiquitous air-interface in future generation communication systems. The problem of multiple access interference (MAI) is vital for a CDMA system. A variety of MUD has been proposed to mitigate the MAI. The simplest one is the single-user matched filter appr...
متن کامل