SEARCH
ISTC-CC NEWSLETTER
RESEARCH HIGHLIGHTS
Ling Liu's SC13 paper "Large Graph Processing Without the Overhead" featured by HPCwire.
ISTC-CC provides a listing of useful benchmarks for cloud computing.
Another list highlighting Open Source Software Releases.
Second GraphLab workshop should be even bigger than the first! GraphLab is a new programming framework for graph-style data analytics.
ISTC-CC Abstract
inTune: Coordinating Multicore Islands to Achieve Global Policy Objectives
First ACM Conference on Timely Results in Operating Systems (TRIOS’13), with SOSP’13, November 2013.
Priyanka Tembey, Ada Gavriloska, and Karsten Schwan
Georgia Institute of Technology
Multicore platforms are moving from small numbers of homogeneous cores to 'scale out' designs with multiple tiles or 'islands' of cores residing on a single chip, each with different resources and potentially controlled by their own resource managers. Applications running on such machines, however, operate across multiple such resource islands, and this also holds for global properties like platform power caps. The inTune software architecture meets the consequent need to support platform-level application requirements and properties. It (i) provides the base coordination abstractions needed for realizing platform-global resource management and (ii) offers management overlays that make it easy to implement diverse per-application and platform-centric management policies. A Xen hypervisor-level implementation of inTune supports policies that can (i) pro-actively prepare for increased or decreased resource usage when the inter-island dependencies of applications are known, or (ii) re-actively respond to monitored overloads, threshold violations or similar. Experimental evaluations on a larger-scale multi-core platform demonstrate that its use leads to notable performance and resource utilization gains: such as a reduction in the variability across request response times for a three-tier web server by up to 40%, and completion time gains of 15% for parallel benchmarks.
FULL PAPER: pdf
