Online auto-tuning for the time-step-based parallel solution of ODEs on shared-memory systems

Kalinnik, Natalia; Korch, Matthias; Rauber, Thomas

doi:10.1016/j.jpdc.2014.03.006

Cited by 8 publications

(8 citation statements)

References 23 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…), as in the case shown for ordinary differential equations in [55]. Online autotuning can especially be used successfully for time-stepping methods.…”

Section: Autotuning Approaches Towards Energy Efficiencymentioning

confidence: 96%

“…A model-based pre-selection phase can be used to reduce the number of implementation variants that need to be tested at runtime. For ordinary differential equations, this approach has been applied successfully [55], and it has been shown that the autotuning overhead at runtime is not too large. An automated online performance tuning approach for general applications is provided by the Active Harmony automated runtime system [29], which allows runtime switching of algorithms and tuning of libraries and application parameters to improve the resulting performance on a given hardware platform.…”

Section: Autotuning Approaches Towards Energy Efficiencymentioning

confidence: 99%

See 1 more Smart Citation

Energy-efficient Algorithms for Ultrascale Systems

Carretero

Distefano

Petcu

et al. 2015

JSFI

View full text Add to dashboard Cite

The chances to reach Exascale or Ultrascale Computing are strongly connected with the problem of the energy consumption for processing applications. For physical and economical reasons, the energy consumption has to be reduced significantly to make Ultrascale Computing possible. The research efforts towards energy-saving mechanisms of the hardware have already made energy-aware hardware systems available. However, to achieve a strong energy reduction, hardware mechanisms must be complemented with new energy-efficient software that can exploit them so that the foreseen energy savings actually result. In the software area, there also exist a multitude of research approaches towards energy saving, often concentrating either on the system software level or the application organization level, reflecting the expertise of the corresponding research group. The challenge of reducing the energy consumption dramatically to make Ultrascale Computing possible is so ambitious that a concerted action combining research efforts through all the software levels seems reasonable. In this article, we discuss the current research efforts and results related to energy efficiency in the diverse areas of software. We conclude with open problems and questions concerning energy-related techniques with an emphasis on the application or algorithmic side.

show abstract

“…), as in the case shown for ordinary differential equations in [55]. Online autotuning can especially be used successfully for time-stepping methods.…”

Section: Autotuning Approaches Towards Energy Efficiencymentioning

confidence: 96%

Section: Autotuning Approaches Towards Energy Efficiencymentioning

confidence: 99%

Energy-efficient Algorithms for Ultrascale Systems

Carretero

Distefano

Petcu

et al. 2015

JSFI

View full text Add to dashboard Cite

show abstract

“…The two application fields, particle simulation methods and solving differential equations, are well-known fields, which are also examined individually. In the approach of [18], an online tuning is presented to optimize the performance of parallel solution methods of solving differential equations. This approach does not consider the energy consumption and does not optimize the number of threads.…”

Section: Related Workmentioning

confidence: 99%

A performance- and energy-oriented extended tuning process for time-step-based scientific applications

et al. 2020

Self Cite

View full text Add to dashboard Cite

Scientific application codes are often long-running time-and energy-consuming parallel codes, and the tuning of these methods towards the characteristics of a specific hardware is essential for a good performance. However, since scientific software is often developed over many years, the application software usually survives several hardware generations, which might make a re-tuning of the existing codes necessary. To simplify the tuning process, it would be beneficial to have software with inherent tuning possibilities. In this article, we explore the possibilities of tuning methods for time-step-based applications. Two different time-step-based application classes are considered, which are solution methods for ordinary differential equations and particle simulation methods. The investigation comprises a broad range of tuning possibilities, starting from the choice of algorithms, the parallel programming model, static implementation variants, input characteristics as well as hardware parameters for parallel execution. An experimental investigation shows the different characteristics of the application classes on different multicore systems. The results show that a combination of offline and online tuning leads to good tuning results. However, due to the different input characteristics of the two application classes, regular versus irregular, different tuning aspects are most essential.

show abstract

“…Autotuning tries to determine the best configuration from a search space of possible code variants and parameters (e.g., block sizes for loop tiling, loop unroll factors, or number of threads). In [21], an online autotuning approach for system-parallel PIRK methods has been investigated.…”

Section: B Performance Optimizationmentioning

confidence: 99%