Automatic Energy-Efficiency Monitoring of OpenMP Workloads

Abstract

Energy-efficiency has been a major challenge in compute systems over the last decade. Both embedded and highperformance computing domains are concerned. Many efforts have been currently spent to devise solutions that are capable of providing systems with the best compromises in terms of performance and power consumption. In this paper, we propose an approach for on-line energy-efficiency analysis when executing OpenMP workloads on multicore systems. The novelty of our approach lies in the ability to monitor energy efficiency at runtime without prior knowledge of the application profile or code annotation. The solution relies on two new metrics: the Chunks per Second (CpS) and Chunks per Joule (CpJ). The former captures the quantity of work achieved by threads per unit time (i.e. a performance indicator). The latter indicates the quantity of work achieved by threads per unit energy, also corresponding to the performance per watt (i.e. an energy efficiency indicator). As most programs are made of several phases performing different computations for which CpS and CpJ cannot be related, it is crucial to be capable of detecting phase changes such as to perform intra-phase energy efficiency optimizations. For that purpose we devise a specific neural network model derived from the popular auto-encoder largely explored in the machine learning community, that is capable of understanding application profile and track phase changes at run-time. We show that these new metrics allow to perform energy efficiency optimization, and illustrate our approach on the analysis of the SRAD application from the Rodinia benchmark. The energy-efficiency profile analysis of the application is conducted on both an Intel and ARM platforms, showing its flexibility.

Influence of loop transformations on performance and energy consumption of the multithreded WZ factorization

“…These challenges are analyzed on many levels. In particular, the works of [11], [13], [16] dealt with the topic of energy consumption in the context of using the OpenMP standard for multi-core computers with shared memory.…”

“…Reducing energy consumption in these architectures is one of the major challenges. The current research trends based on performance studies [12], [13], [16] and comparisons are to develop hardware and software to achieve the best performance and energy compromise. One of the aspects of creating energy-aware software is the optimization of implementation of complex numerical algorithms.…”

Online Learning for Dynamic Control of OpenMP Workloads

“…Intuitively, backpropagation algorithm (training) attempt to find the most prominent features in the input vector across the training dataset, such that loss (reflecting the difference between input and output) is minimized. Here, we use a specific autoencoder architecture proposed in [4], illustrated in Fig. 8.…”

“…This work assumes actions comprising DVFS and active core count. The reward function used in our RL engine is a relative energy efficiency metric that is extracted from the OpenMP runtime as described in [4]. This approach has the distinct advantage to require no prior program profiling nor code annotation: it relies on the fundamental concept of chunks that OpenMP uses for breaking down job into parallelizable job fragments.…”

“…an energy efficiency measure. The same setup as that of the original contribution [4] is here used as a frontend to the RL engine with additional phase detection logic described later on in this paper.…”