Multi Objective Optimization of HPC Kernels for Performance, Power, and Energy
Multi Objective Optimization of HPC Kernels for Performance, Power, and Energy
复制标题
HPC 内核的性能、功耗和能源多目标优化
DOI：
--
复制
发表时间：
2013
期刊：
影响因子：
--
通讯作者：
Stefan M. Wild
中科院分区：
文献类型：
--
作者：
Prasanna Balaprakash;Ananta Tiwari;Stefan M. Wild
高性能计算领域中的代码优化传统上关注于减少执行时间。这个问题，在数学术语中，已被表示为一个单目标优化问题。然而，下一代系统的预期问题，要求执行时间和其他指标之间的相互作用进行更详细的分析。诸如功率、性能、能量和弹性之类的目标可能都是一起的，并且相互交易。我们提出了一个多目标制定的代码优化问题。我们提出的框架可以帮助人们探索多个目标之间的潜在权衡，并提供了一个显着丰富的分析比可以通过处理额外的指标作为硬约束。我们经验性地研究了各种指标，架构和代码优化决策，并提供证据表明，这种权衡在实践中存在。
Code optimization in the high-performance computing realm has traditionally focused on reducing execution time. The problem, in mathematical terms, has been expressed as a single objective optimization problem. The expected concerns of next-generation systems, however, demand a more detailed analysis of the interplay among execution time and other metrics. Metrics such as power, performance, energy, and resiliency may all be targeted together and traded against one another. We present a multi objective formulation of the code optimization problem. Our proposed framework helps one explore potential tradeoffs among multiple objectives and provides a significantly richer analysis than can be achieved by treating additional metrics as hard constraints. We empirically examine a variety of metrics, architectures, and code optimization decisions and provide evidence that such tradeoffs exist in practice.