Evaluation of the HPC Applications Dynamic Behavior in Terms of Energy Consumption

Lecture Notes in Computer Science

et al. 2018

This paper introduces two tools for manual energy evaluation and runtime tuning developed at IT4Innovations in the READEX project. The MERIC library can be used for manual instrumentation and analysis of any application from the energy and time consumption point of view. Besides tracing, MERIC can also change environment and hardware parameters during the application runtime, which leads to energy savings. MERIC stores large amounts of data, which are difficult to read by a human. The RADAR generator analyses the MERIC output files to find the best settings of evaluated parameters for each instrumented region. It generates a L A T E X report and a MERIC configuration file for application production runs.

Section: Resultsmentioning

confidence: 95%

MERIC and RADAR Generator: Tools for Energy Evaluation and Runtime Tuning of HPC Applications

Vysocky

Beseda

Lecture Notes in Computer Science

et al. 2018

“…During the analysis, the application is executed multiple times with all combinations of the tuning parameters to find the optimal settings for each significant region. After manual annotation, the analysis is performed automatically using our in-house tools MERIC (performs parameter tuning and energy measurements) and RADAR (analyzes the measurements and creates the tuning model), see Vysocky et al (2017). The optimal settings are then saved into a file called tuning model , which is used in step 3.…”

Section: Improving Energy Efficiency Using Dynamic Tuningmentioning

confidence: 99%

A massively parallel and memory-efficient FEM toolbox with a hybrid total FETI solver with accelerator support

The International Journal of High Performance Computing Applica

Merta

Vavrik

et al. 2018

In this paper we present the ESPRESO FEM library, which includes a FEM toolbox with interfaces to professional and open-source simulation tools, and a massively parallel Hybrid Total FETI (HTFETI) solver which can fully utilize the OLCF Titan supercomputer, and achieves super-linear scaling. This paper presents several new techniques for FETI solvers designed for efficient utilization of supercomputers with a focus on: (i) performance-we present a fivefold reduction of solver runtime for the Laplace equation by redesigning the FETI solver, and offloading the key workload to the accelerator. We compare Intel Xeon Phi 7120p and Tesla K80 and P100 accelerators to Intel Xeon E5-2680v3 and Xeon Phi 7210 CPUs; and (ii) memory efficiency-we present two techniques which increase the efficiency of the HTFETI solver 1.8 times, and pushes the limits of the largest possible problem ESPRESO can solve from 124 to 223 billion unknowns for problems with unstructured meshes. Finally we show that by dynamicly tuning hardware parameters we can reduce energy consumption by up to 33 %.

“…After manual annotation, the analysis is performed automatically using two tools: MERIC (performs parameter tuning and energy measurements) and RADAR (analyzes the measurements and creates the tuning model), see [14]. The optimal settings are then saved into a file called tuning model, which is used in step 3.…”

Section: Dynamic Tuning Approach For Energy Optimizationmentioning

confidence: 99%

“…We implement C/C++ open source MERIC library [14] providing application profiling and tuning of selected application parts according the READEX approach. Main goal of the library is to simplify the process of application analysis and provide maximum possible energy savings with minimum overhead caused by MERIC.…”

Section: Introductionmentioning

confidence: 99%

Fine-grained application tuning on OpenPOWER HPC systems

The Sixth International Conference on Parallel, Distributed, GPU and Cloud Computing for Engineering

Bartolini²,

Vysocky

Self Cite

Energy consumption of HPC centers becomes major limitation in building a new peta or exascale system. In this paper we evaluate the approach of dynamic application tuning used to reduce energy consumption of HPC systems on the IBM's Power8+ system. A D.A.V.I.D.E. HPC system installed in CINECA has been selected for these experiments due to its advanced power consumption monitoring system which provides power samples at high sampling rate of 1 kHz, that are stored in a stand-alone scalable database and can be read on request.We have tuned two available hardware parameters: (1) Dynamic Voltage and Frequency Scaling and (2) concurrency throttling with focus on hyper-threading, which plays a significant role with respect to performance on Power8+ systems.As a test application we have used the ESPRESO library, which is fully fledged application. Each part of such application may have its optimal configuration of the tuned parameters to utilize the system however at the same time without wasting the resources. This way we were able to reduce the application runtime about 23.7 % when tuning to reduce runtime or 27.3 % of energy with 9.9 % runtime savings when tuning for minimal energy consumption.