PROPHET: Goal-Oriented Provisioning for Highly Tunable Multicore Processors in Cloud Computing

Free registration required

Executive Summary

As multicore and future many-core processors emerge with more complex optimization features, how to achieve optimal efficiency in terms of energy, power, and performance when running parallel workloads poses a new challenge to both large-scale system administrators and networked machines used by individuals. Unlike traditional high performance computing systems which were often designed with a slew of homogeneous processing elements, a modern computing environment such as data centers or distributed nodes in a computing cloud will consist of a variety of multicore or many-core processors with different underlying microarchitectures. For example, Intel's Core i7 processor (Code-named Nehalem) is significantly different from previous x86 multicore processors.

  • Format: PDF
  • Size: 196.03 KB