Open AccessJournal Article
Task Parallel Models Based on Dynamic Data Placement to Reduce NUMA Effects
Yan Wang,Brian Vinter +1 more
TL;DR: This work bases on the strategy of dynamic data placement improving system performance by reducing the frequently data access to remote memory and also by keeping load balance between each NUMA domain and demonstrates that the benchmarks using the task parallel models achieve system performance improvement by 50% at least.
read more
Abstract: NUMA (Non-Uniform Memory Access) multicore computers become popular in scientific and industrial fields due to its scalable memory performance. However, large-scale intensive data computing on NUMA architecture are facing up to the challenges in data locality problems called NUMA effects that are caused by the overhead accesses of cross-node data. Our task parallel model bases on the strategy of dynamic data placement improving system performance by reducing the frequently data access to remote memory and also by keeping load balance between each NUMA domain. The task parallel models involved OpenMP, numactl and libnuma. The evaluation demonstrates that the benchmarks using our task parallel models on a 32-core NUMA computer with various workloads achieve system performance improvement by 50% at least.
read more
Chat with Paper
AI Agents for this Paper
Find similar papers on Google Scholar, PubMed and Arxiv
Write a critical review of this paper
Analyze citations of this paper to find unaddressed research gaps
References
•Book
Computer Architecture: A Quantitative Approach
John L. Hennessy,David A. Patterson +1 more
- 01 Dec 1989
TL;DR: This best-selling title, considered for over a decade to be essential reading for every serious student and practitioner of computer design, has been updated throughout to address the most important trends facing computer designers today.
12.6K
Local and Remote Memory: Memory in a Linux/NUMA System
Christoph Lameter
- 01 Jan 2006
TL;DR: Memory seems to be so simple, if you need it then just get some and use it, however the distance of the memory to the executing process matters, and performance can sink dramatically if memory references are made too frequently to pages on remote nodes.
Data placement in widely distributed systems
Tevfik Kosar,Miron Livny +1 more
- 01 Jan 2005
TL;DR: This dissertation proposes a framework that de-couples computation and data placement, allows asynchronous execution of each, and treats data placement as a full-fledged job that can be queued, scheduled, monitored and check-pointed like computational jobs.
40
Experimental evaluation of NUMA effects on database management systems
Tim Kiefer,Benjamin Schlegel,Wolfgang Lehner +2 more
- 01 Jan 2013
TL;DR: It is shown that partitioning a database's memory with respect to the data’s access patterns can improve the query performance by as much as 75% and grouping database worker threads on CPUs, based on their data partitions, improves cache behavior, which in turn improves query performance.
33
Improving memory affinity of geophysics applications on NUMA platforms using minas
Christiane Pousa Ribeiro,Márcio Castro,Jean-François Méhaut,Alexandre Carissimi +3 more
- 22 Jun 2010
TL;DR: This paper introduces Minas, a framework which provides either explicit or automatic memory affinity management with architecture abstraction for ccNUMAs and evaluates the solution on two ccNUMA platforms using two geophysics parallel applications.
31
Related Papers (5)
Mathieu Faverge,Xavier Lacoste,Pierre Ramet +2 more
- 01 Jan 2008
Tanuj Kr Aasawat,Tahsin Reza,Matei Ripeanu +2 more
- 01 Nov 2018