Parallel Processing and Applied

Mathematics 11th International

Conference PPAM 2015 Krakow Poland
September 6 9 2015 Revised Selected
Papers Part II 1st Edition Roman
Wyrzykowski
Visit to download the full and correct content document:
https://textbookfull.com/product/parallel-processing-and-applied-mathematics-11th-int
ernational-conference-ppam-2015-krakow-poland-september-6-9-2015-revised-select
ed-papers-part-ii-1st-edition-roman-wyrzykowski/
More products digital (pdf, epub, mobi) instant
download maybe you interests ...

Languages and Compilers for Parallel Computing 28th

International Workshop LCPC 2015 Raleigh NC USA
September 9 11 2015 Revised Selected Papers 1st Edition
Xipeng Shen
https://textbookfull.com/product/languages-and-compilers-for-
parallel-computing-28th-international-workshop-lcpc-2015-raleigh-
nc-usa-september-9-11-2015-revised-selected-papers-1st-edition-
xipeng-shen/

Information and Communications Security 17th

International Conference ICICS 2015 Beijing China
December 9 11 2015 Revised Selected Papers 1st Edition
Sihan Qing
https://textbookfull.com/product/information-and-communications-
security-17th-international-conference-icics-2015-beijing-china-
december-9-11-2015-revised-selected-papers-1st-edition-sihan-
qing/

Web Information Systems and Technologies 11th

International Conference WEBIST 2015 Lisbon Portugal
May 20 22 2015 Revised Selected Papers 1st Edition
Valérie Monfort
https://textbookfull.com/product/web-information-systems-and-
technologies-11th-international-conference-webist-2015-lisbon-
portugal-may-20-22-2015-revised-selected-papers-1st-edition-
valerie-monfort/

Smart Card Research and Advanced Applications 14th

International Conference CARDIS 2015 Bochum Germany
November 4 6 2015 Revised Selected Papers 1st Edition
Naofumi Homma
https://textbookfull.com/product/smart-card-research-and-
advanced-applications-14th-international-conference-
cardis-2015-bochum-germany-november-4-6-2015-revised-selected-
E Learning E Education and Online Training Second
International Conference eLEOT 2015 Novedrate Italy
September 16 18 2015 Revised Selected Papers 1st
Edition Giovanni Vincenti
https://textbookfull.com/product/e-learning-e-education-and-
online-training-second-international-conference-
eleot-2015-novedrate-italy-september-16-18-2015-revised-selected-
papers-1st-edition-giovanni-vincenti/

Data Driven Process Discovery and Analysis 5th IFIP WG

2 6 International Symposium SIMPDA 2015 Vienna Austria
December 9 11 2015 Revised Selected Papers 1st Edition
Paolo Ceravolo
https://textbookfull.com/product/data-driven-process-discovery-
and-analysis-5th-ifip-wg-2-6-international-symposium-
simpda-2015-vienna-austria-december-9-11-2015-revised-selected-
papers-1st-edition-paolo-ceravolo/

Brain Inspired Computing Second International Workshop

BrainComp 2015 Cetraro Italy July 6 10 2015 Revised
Selected Papers 1st Edition Katrin Amunts

https://textbookfull.com/product/brain-inspired-computing-second-
international-workshop-braincomp-2015-cetraro-italy-
july-6-10-2015-revised-selected-papers-1st-edition-katrin-amunts/

Internet of Things IoT Infrastructures Second

International Summit IoT 360 2015 Rome Italy October 27
29 2015 Revised Selected Papers Part II 1st Edition
Benny Mandler
https://textbookfull.com/product/internet-of-things-iot-
infrastructures-second-international-summit-iot-360-2015-rome-
italy-october-27-29-2015-revised-selected-papers-part-ii-1st-
edition-benny-mandler/

Computer and Computing Technologies in Agriculture IX

9th IFIP WG 5 14 International Conference CCTA 2015
Beijing China September 27 30 2015 Revised Selected
Papers Part I 1st Edition Daoliang Li
https://textbookfull.com/product/computer-and-computing-
technologies-in-agriculture-ix-9th-ifip-wg-5-14-international-
conference-ccta-2015-beijing-china-september-27-30-2015-revised-
 Roman Wyrzykowski · Ewa Deelman
Jack Dongarra · Konrad Karczewski
Jacek Kitowski · Kazimierz Wiatr (Eds.)
LNCS 9574

Parallel Processing
and Applied Mathematics
11th International Conference, PPAM 2015
Krakow, Poland, September 6–9, 2015
Revised Selected Papers, Part II

123
Lecture Notes in Computer Science 9574
Commenced Publication in 1973
Founding and Former Series Editors:
Gerhard Goos, Juris Hartmanis, and Jan van Leeuwen

Editorial Board
David Hutchison
Lancaster University, Lancaster, UK
Takeo Kanade
Carnegie Mellon University, Pittsburgh, PA, USA
Josef Kittler
University of Surrey, Guildford, UK
Jon M. Kleinberg
Cornell University, Ithaca, NY, USA
Friedemann Mattern
ETH Zurich, Zürich, Switzerland
John C. Mitchell
Stanford University, Stanford, CA, USA
Moni Naor
Weizmann Institute of Science, Rehovot, Israel
C. Pandu Rangan
Indian Institute of Technology, Madras, India
Bernhard Steffen
TU Dortmund University, Dortmund, Germany
Demetri Terzopoulos
University of California, Los Angeles, CA, USA
Doug Tygar
University of California, Berkeley, CA, USA
Gerhard Weikum
Max Planck Institute for Informatics, Saarbrücken, Germany
More information about this series at http://www.springer.com/series/7407
Roman Wyrzykowski Ewa Deelman
•

Jack Dongarra Konrad Karczewski

•

Jacek Kitowski Kazimierz Wiatr (Eds.)

•

Parallel Processing
and Applied Mathematics
11th International Conference, PPAM 2015
Krakow, Poland, September 6–9, 2015
Revised Selected Papers, Part II

123
Editors
Roman Wyrzykowski Konrad Karczewski
Czestochowa University of Technology Institute of Computer and Information
Czestochowa Science
Poland Czestochowa University of Technology
Czestochowa
Ewa Deelman Poland
Department of Computer Science
University of Southern California Jacek Kitowski
Marina Del Rey, CA Department of Computer Science
USA AGH University of Science and Technology
Krakow
Jack Dongarra Poland
Electrical Engineering and Computer
Science Kazimierz Wiatr
University of Tennessee AGH University of Science and Technology
Knoxville, TN Krakow
USA Poland

ISSN 0302-9743 ISSN 1611-3349 (electronic)

Lecture Notes in Computer Science
ISBN 978-3-319-32151-6 ISBN 978-3-319-32152-3 (eBook)
DOI 10.1007/978-3-319-32152-3

Library of Congress Control Number: 2016934929

LNCS Sublibrary: SL1 – Theoretical Computer Science and General Issues

© Springer International Publishing Switzerland 2016

This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of the
material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation,
broadcasting, reproduction on microfilms or in any other physical way, and transmission or information
storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now
known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication
does not imply, even in the absence of a specific statement, that such names are exempt from the relevant
protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this book are
believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors
give a warranty, express or implied, with respect to the material contained herein or for any errors or
omissions that may have been made.

Printed on acid-free paper

This Springer imprint is published by Springer Nature

The registered company is Springer International Publishing AG Switzerland
 Preface

This volume comprises the proceedings of the 11th International Conference on Par-
allel Processing and Applied Mathematics – PPAM 2015, which was held in Krakow,
Poland, September 6–9, 2015. It was organized by the Department of Computer and
Information Science of the Częstochowa University of Technology together with with
the AGH University of Science and Technology, under the patronage of the Committee
of Informatics of the Polish Academy of Sciences, in cooperation with the ICT COST
Action IC1305 “Network for Sustainable Ultrascale Computing (NESUS).” The main
organizer was Roman Wyrzykowski.
PPAM is a biennial conference. Ten previous events have been held in different
places in Poland since 1994. The proceedings of the last six conferences have been
published by Springer in the Lecture Notes in Computer Science series (Nałęczów,
2001, vol. 2328; Częstochowa, 2003, vol. 3019; Poznań, 2005, vol. 3911; Gdańsk,
2007, vol. 4967; Wrocław, 2009, vols. 6067 and 6068; Toruń, 2011, vols. 7203 and
7204; Warsaw, 2013, vols. 8384 and 8385).
The PPAM conferences have become an international forum for exchanging ideas
between researchers involved in parallel and distributed computing, including theory
and applications, as well as applied and computational mathematics. The focus of
PPAM 2015 was on models, algorithms, and software tools that facilitate efficient and
convenient utilization of modern parallel and distributed computing architectures, as
well as on large-scale applications, including big data problems.
This meeting gathered more than 190 participants from 33 countries. A strict ref-
ereeing process resulted in the acceptance of 111 contributed presentations, while
approximately 43 % of the submissions were rejected. Regular tracks of the conference
covered important fields of parallel/distributed/cloud computing and applied mathe-
matics such as:
– Numerical algorithms and parallel scientific computing
– Parallel non-numerical algorithms
– Tools and environments for parallel/distributed/cloud computing
– Applications of parallel computing
– Applied mathematics, neural networks, evolutionary computing, and metaheuristics
The plenary and invited talks were presented by:
– David A. Bader from the Georgia Institute of Technology (USA)
– Costas Bekas from IBM Research — Zurich (Switzerland)
– Pete Beckman from the Argonne National Laboratory (USA)
– Christopher Carothers from the Rensselaer Polytechnic Institute (USA)
– Barbara Chapman from the University of Houston (USA)
– Willem Deconinck from the European Centre for Medium-Range Weather Forecast
(UK)
– Geoffrey C. Fox from Indiana University (USA)
VI Preface

– Dieter Kranzlmueller from the Ludwig-Maximilians-Universität München

(Germany)
– Vladik Kreinovich from the University of Texas at El Paso (USA)
– Alexey Lastovetsky from the University College Dublin (Ireland)
– Carlos Osuna from ETH Zurich (Switzerland)
– Srinivasan Parthasarathy from the Ohio State University (USA)
– Enrique S. Quintana-Orti from the Universidad Jaime I (Spain)
– Thomas Rauber from the University of Bayreuth (Germany)
– Daniel Reed from the University of Iowa (USA)
– Rizos Sakellariou from the University of Manchester (UK)
– Boleslaw K. Szymanski from the Rensselaer Polytechnic Institute (USA)
– Manuel Ujaldon from Nvidia
– Jeffrey Vetter from the Oak Ridge National Laboratory and Georgia Institute of
Technology (USA)
– Richard W. Vuduc from the Georgia Institute of Technology (USA)
– Torsten Wilde from the Leibnitz Supercomputing Centre (LRZ) (Germany)
Important and integral parts of the PPAM 2015 conference were the workshops:
– Minisympsium on GPU Computing organized by José R. Herrero from the
Universitat Politecnica de Catalunya (Spain), Enrique S. Quintana-Ortí from the
Universidad Jaime I (Spain), and Robert Strzodka from Heidelberg University
(Germany).
– The Third Workshop on Models, Algorithms and Methodologies for Hierarchical
Parallelism in New HPC Systems organized by Giulliano Laccetti and Marco
Lapegna from the University of Naples Federico II (Italy), and Raffaele Montella
from the University of Naples Parthenope (Italy).
– Workshop on Power and Energy Aspects of Computation organized by Jee Choi
from the IBM T.J. Watson Research Center (USA), Piotr Luszczek from the
University of Tennessee (USA), Leonel Sousa from the Technical University of
Lisbon (Portugal), and Richard W. Vuduc from the Georgia Institute of Technology
(USA).
– Workshop on Scheduling for Parallel Computing — SPC 2015 organized by Maciej
Drozdowski from the Poznań University of Technology (Poland).
– The 6th Workshop on Language-Based Parallel Programming Models — WLPP
2015 organized by Ami Marowka from the Bar-Ilan University (Israel).
– The 5th Workshop on Performance Evaluation of Parallel Applications on Large-
Scale Systems organized by Jan Kwiatkowski from the Wrocław University of
Technology (Poland).
– Workshop on Parallel Computational Biology — PBC 2015 organized by Bertil
Schmidt from the University of Mainz (Germany) and Jarosław Żola from the
University at Buffalo (USA).
– Workshop on Applications of Parallel Computations in Industry and Engineering
organized by Raimondas Čiegis from the Vilnius Gediminas Technical University
(Lithuania) and Julius Žilinskas from the Vilnius University (Lithuania).
 Preface VII

– Minisymposium on HPC Applications in Physical Sciences organized by Grzegorz

Kamieniarz and Wojciech Florek from the A. Mickiewicz University in Poznań
(Poland).
– The Second Workshop on Applied High-Performance Numerical Algorithms in
PDEs organized by Piotr Krzyżanowski and Leszek Marcinkowski from Warsaw
University (Poland) and Talal Rahman from Bergen University College (Norway).
– Minisymposium on High-Performance Computing Interval Methods organized by
Bartłomiej J. Kubica from the Warsaw University of Technology (Poland).
– Workshop on Complex Collective Systems organized by Paweł Topa and Jarosław
Wąs from the AGH University of Science and Technology (Poland).
– Special Session on Efficient Algorithms for Problems with Matrix and Tensor
Decompositions organized by Marian Vajtersic from the University of Salzburg
(Austria) and Gabriel Oksa from the Slovak Academy of Sciences.
– Special Session on Algorithms, Methodologies, and Frameworks for HPC in
Geosciences and Weather Prediction organized by Zbigniew Piotrowski from the
Institute of Meteorology and Water Management (Poland) and Krzysztof Rojek
from the Częstochowa University of Technology (Poland).
The PPAM 2015 meeting began with four tutorials:
– Scientific Computing with GPUs, by Dominik Göddeke from the University of
Stuttgart (Germany), Robert Strzodka from Heidelberg University (Germany), and
Manuel Ujaldon from the University of Malaga (Spain) and Nvidia.
– Advanced Scientific Visualization with VisNow, by Krzysztof Nowiński, Bartosz
Borucki, Kerstin Kantiem, and Szymon Jaranowski from the University of Warsaw
(Poland).
– Parallel Computing in Java, by Piotr Bała from the Warsaw University of Tech-
nology (Poland) and Marek Nowicki, Łukasz Górski, Magdalena Ryczkowska from
the Nicolaus Copernicus University (Poland).
– Introduction to Programming with Intel Xeon Phi, by Krzysztof Rojek and Łukasz
Szustak from the Częstochowa University of Technology (Poland).
An integral part of the GPU Tutorial was the CUDA quiz with participants chal-
lenged to maximize the performance on a common GPU model. The winner was Miłosz
Ciżnicki from the Poznan Supercomputing and Networking Center. The winner
received the prize of a Tesla K40 GPU generously donated by Nvidia for the conference
given its role of PPAM sponsor. The second and third prizes were granted, respectively,
to Michał Antkowiak and Łukasz Kucharski, both from the A. Mickiewicz University in
Poznań.
Nvidia also donated another prize, GeForce GTX480 GPU, for the authors of the
best paper presented at the Minisymposium on GPU Computing. This prize was
awarded to Jan Gmys, Mohand Mezmaz, Nouredine Melab, and Daniel Tuyttens from
the University of Mons, who presented the paper “IVM-Based Work Stealing for
Parallel Branch-and-Bound on GPU.”
Special Session on Algorithms, Methodologies, and Frameworks for HPC in Geo-
sciences and Weather Prediction: Contemporary and future applications of numerical
VIII Preface

weather prediction, climate research, and studies in geosciences demand multidisciplinary

advancements in computing methodologies, including the use of multi-/manycore
processors and accelerators, scalable and energy-efficient frameworks, and big data
strategies, as well as new or improved numerical algorithms. This includes, for example,
development of scalable, high-resolution methods for integration of fluid PDEs and efficient
iterative solvers, highly optimized ports to modern hardware (CPU, GPU, Xeon Phi), code
development and portability strategies, and libraries for handling geophysical datasets.
The special session served as a multidisciplinary forum for the discussion of state-
of-the-art research and development toward the next-generation geophysical fluid
solvers and weather/climate prediction applications.
The special session featured a number of invited and contributed talks, covering
recent advances in numerical algorithms, accelerator methodologies, energy-efficent
computing, and large dataset managements, including:
• Algorithms and tools for the extreme-scale numerical weather prediction (invited
plenary talk by Willem Deconinck et al.)
• Adaptation of COSMO Consortium weather and climate numerical models to
hybrid architecures (invited plenary talk by Carlos Osuna et al.)
• Highly efficient port of the GCR solver using high-level stencil framework on
multi- and many-core architectures (by M. Ciżnicki et al.)
• Autotuned scheduler for time/energy optimization for a fully three-dimensional
MPDATA advection scheme on the hybrid CPU-GPU clusters (by K. Rojek et al.)
• Parallel alternating direction implicit preconditioners for all-scale atmospheric
models (by Z. Piotrowski et al.)
The organizers are indebted to the PPAM 2015 sponsors, whose support was vital to
the success of the conference. The main sponsor was Intel Corporation and the other
sponsors were: Nvidia, Action S.A., and Gambit. We thank all the members of the
international Program Committee and additional reviewers for their diligent work in
reviewing the submitted papers. Finally, we thank all of the local organizers from the
Częstochowa University of Technology and the AGH University of Science and
Technology, who helped us run the event very smoothly. We are especially indebted to
Grażyna Kołakowska, Urszula Kroczewska, Łukasz Kuczyński, Adam Tomaś, and
Marcin Woźniak from the Częstochowa University of Technology; and to Krzysztof
Zieliński, Kazimierz Wiatr, and Jacek Kitowski from the AGH University of Science
and Technology.
We hope that this volume will be useful to you. We would like everyone who reads
it to feel invited to the next conference, PPAM 2017, which will be held during
September 10–13, 2017, in Lublin, the largest Polish city east of the Vistula River.

January 2016 Roman Wyrzykowski

Jack Dongarra
Ewa Deelman
Konrad Karczewski
Jacek Kitowski
Kazimierz Wiatr
 Organization

Program Committee
Jan Węglarz Poznań University of Technology, Poland,
(Honorary Chair)
Roman Wyrzykowski Częstochowa University of Technology, Poland,
(Program Chair)
Ewa Deelman University of Southern California, USA,
(Program Co-chair)
Francisco Almeida Universidad de La Laguna, Spain
Pedro Alonso Universidad Politecnica de Valencia, Spain
Peter Arbenz ETH, Zurich, Switzerland
Cevdet Aykanat Bilkent University, Ankara, Turkey
Piotr Bała Warsaw University, Poland
David A. Bader Georgia Institute of Technology, USA
Michael Bader TU München, Germany
Olivier Beaumont Inria Bordeaux, France
Włodzimierz Bielecki West Pomeranian University of Technology, Poland
Paolo Bientinesi RWTH Aachen, Germany
Radim Blaheta Institute of Geonics, Czech Academy of Sciences,
Czech Republic
Jacek Błażewicz Poznań University of Technology, Poland
Pascal Bouvry University of Luxembourg
Jerzy Brzeżiński Poznań University of Technology, Poland
Marian Bubak AGH Kraków, Poland, and University of Amsterdam,
The Netherlands
Tadeusz Burczyński Polish Academy of Sciences, Warsaw
Christopher Carothers Rensselaer Polytechnic Institute, USA
Jesus Carretero Universidad Carlos III de Madrid, Spain
Raimondas Čiegis Vilnius Gediminas Technical University, Lithuania
Andrea Clematis IMATI-CNR, Italy
Zbigniew Czech Silesia University of Technology, Poland
Jack Dongarra University of Tennessee and ORNL, USA,
and University of Manchester, UK
Maciej Drozdowski Poznań University of Technology, Poland
Mariusz Flasiński Jagiellonian University, Poland
Tomas Fryza Brno University of Technology, Czech Republic
Jose Daniel Garcia Universidad Carlos III de Madrid, Spain
Pawel Gepner Intel Corporation, USA
Domingo Gimenez University of Murcia, Spain
X Organization

Mathieu Giraud LIFL and Inria, France

Jacek Gondzio University of Edinburgh, Scotland, UK
Andrzej Gościński Deakin University, Australia
Laura Grigori Inria, France
Adam Grzech Wroclaw University of Technology, Poland
Inge Gutheil Forschungszentrum Juelich, Germany
Georg Hager University of Erlangen-Nuremberg, Germany
José R. Herrero Universitat Politecnica de Catalunya, Barcelona, Spain
Ladislav Hluchy Slovak Academy of Sciences, Bratislava, Slovakia
Sasha Hunold Vienna University of Technology, Austria
Florin Isaila Universidad Carlos III de Madrid, Spain
Ondrej Jakl Institute of Geonics, Czech Academy of Sciences,
Czech Republic
Emmanuel Jeannot Inria, France
Bo Kagstrom Umea University, Sweden
Christos Kartsaklis Oak Ridge National Laboratory, USA
Eleni Karatza Aristotle University of Thessaloniki, Greece
Ayse Kiper Middle East Technical University, Turkey
Jacek Kitowski Institute of Computer Science, AGH, Poland
Joanna Kołodziej Cracow University of Technology, Poland
Jozef Korbicz University of Zielona Góra, Poland
Stanislaw Kozielski Silesia University of Technology, Poland
Dieter Kranzlmueller Ludwig Maximillian University, Munich, and Leibniz
Supercomputing Centre, Germany
Henryk Krawczyk Gdańsk University of Technology, Poland
Piotr Krzyżanowski University of Warsaw, Poland
Krzysztof Kurowski PSNC, Poznań, Poland
Jan Kwiatkowski Wrocław University of Technology, Poland
Jakub Kurzak University of Tennessee, USA
Giulliano Laccetti University of Naples Federico II, Italy
Marco Lapegna University of Naples Federico II, Italy
Alexey Lastovetsky University College Dublin, Ireland
Joao Lourenco University Nova of Lisbon, Portugal
Tze Meng Low Carnegie Mellon University, USA
Hatem Ltaief KAUST, Saudi Arabia
Emilio Luque Universitat Autonoma de Barcelona, Spain
Vyacheslav I. Maksimov Ural Branch, Russian Academy of Sciences, Russia
Victor E. Malyshkin Siberian Branch, Russian Academy of Sciences, Russia
Pierre Manneback University of Mons, Belgium
Tomas Margalef Universitat Autonoma de Barcelona, Spain
Svetozar Margenov Bulgarian Academy of Sciences, Sofia
Ami Marowka Bar-Ilan University, Israel
Ricardo Morla INESC Porto, Portugal
Norbert Meyer PSNC, Poznań, Poland
Jarek Nabrzyski University of Notre Dame, USA
Raymond Namyst University of Bordeaux and Inria, France
 Organization XI

Maya G. Neytcheva Uppsala University, Sweden

Robert Nowicki Częstochowa University of Technology, Poland
Gabriel Oksa Slovak Academy of Sciences, Bratislava
Ozcan Ozturk Bilkent University, Turkey
Tomasz Olas Częstochowa University of Technology, Poland
Marcin Paprzycki IBS PAN and SWPS, Warsaw, Poland
Dana Petcu West University of Timisoara, Romania
Enrique Universidad Jaime I, Spain
S. Quintana-Ortí
Jean-Marc Pierson Paul Sabatier University, France
Thomas Rauber University of Bayreuth, Germany
Krzysztof Rojek Częstochowa University of Technology, Poland
Jacek Rokicki Warsaw University of Technology, Poland
Gudula Runger Chemnitz University of Technology, Germany
Leszek Rutkowski Częstochowa University of Technology, Poland
Robert Schaefer Institute of Computer Science, AGH, Poland
Olaf Schenk Università della Svizzera Italiana, Switzerland
Stanislav Sedukhin University of Aizu, Japan
Franciszek Seredyński Cardinal Stefan Wyszyński University in Warsaw, Poland
Happy Sithole Centre for High-Performance Computing, South Africa
Jurij Silc Jozef Stefan Institute, Slovenia
Karolj Skala Ruder Boskovic Institute, Croatia
Leonel Sousa Technical University of Lisbon, Portugal
Radek Stompor Université Paris Diderot and CNRS, France
Przemysław Stpiczyński Maria Curie Skłodowska University, Poland
Maciej Stroiński PSNC, Poznań, Poland
Lukasz Szustak Częstochowa University of Technology, Poland
Boleslaw Szymanski Rensselaer Polytechnic Institute, USA
Domenico Talia University of Calabria, Italy
Christian Terboven RWTH Aachen, Germany
Andrei Tchernykh CICESE Research Center, Ensenada, Mexico
Parimala Thulasiraman University of Manitoba, Canada
Roman Trobec Jozef Stefan Institute, Slovenia
Denis Trystram Grenoble Institute of Technology, France
Marek Tudruj Polish Academy of Sciences and Polish-Japanese
Academy of Information Technology, Warsaw, Poland
Pavel Tvrdik Czech Technical University, Prague, Czech Republic
Bora Ucar Ecole Normale Superieure de Lyon, France
Marian Vajtersic Salzburg University, Austria
Vladimir Voevodin Moscow State University, Russian Federation
Jerzy Waśniewski Technical University of Denmark
Kazimierz Wiatr Academic Computer Center CYFRONET AGH, Poland
Bogdan Wiszniewski Gdańsk University of Technology, Poland
Roel Wuyts IMEC Belgium
Andrzej Wyszogrodzki Institute of Meteorology and Water Management, Warsaw,
Poland
XII Organization

Ramin Yahyapour University of Göttingen/GWDG, Germany

Jianping Zhu Cleveland State University, USA
Krzysztof Zieliński Institute of Computer Science, AGH, Poland
Julius Žilinskas Vilnius University, Lithuania
Jarosław Żola University of Buffalo, USA
 Contents – Part II

The Third Workshop on Models, Algorithms, and Methodologies

for Hierarchical Parallelism in New HPC Systems

Virtualizing CUDA Enabled GPGPUs on ARM Clusters . . . . . . . . . . . . . . . 3

Raffaele Montella, Giulio Giunta, Giuliano Laccetti, Marco Lapegna,
Carlo Palmieri, Carmine Ferraro, and Valentina Pelliccia

A Distributed Hash Table for Shared Memory . . . . . . . . . . . . . . . . . . . . . . 15

Wytse Oortwijn, Tom van Dijk, and Jaco van de Pol

Mathematical Approach to the Performance Evaluation of Matrix Multiply

Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
Luisa D’Amore, Valeria Mele, Giuliano Laccetti, and Almerico Murli

How to Mitigate Node Failures in Hybrid Parallel Applications . . . . . . . . . . 35

Maciej Szpindler

A Scalable Numerical Algorithm for Solving Tikhonov Regularization

Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
Rosella Arcucci, Luisa D’Amore, Simone Celestino, Giuliano Laccetti,
and Almerico Murli

Workshop on Power and Energy Aspects of Computation

Energy Performance Modeling with TIA and EML . . . . . . . . . . . . . . . . . . . 57

Francisco Almeida, Javier Arteaga, Vicente Blanco,
and Alberto Cabrera

Considerations of Computational Efficiency in Volunteer and Cluster

Computing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
Paweł Czarnul and Mariusz Matuszek

Workshop on Scheduling for Parallel Computing (SPC 2015)

Parallel Programs Scheduling with Architecturally Supported Regions . . . . . . 77

Łukasz Maśko and Marek Tudruj

Adaptive Multi-level Workflow Scheduling with Uncertain Task Estimates. . . 90

Tomasz Dziok, Kamil Figiela, and Maciej Malawski

Accelerating the Min-Min Heuristic. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101

Martín Pedemonte, Pablo Ezzatti, and Álvaro Martín
XIV Contents – Part II

Divisible Loads Scheduling in Hierarchical Memory Systems with Time

and Energy Constraints . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
Maciej Drozdowski and Jędrzej M. Marszałkowski

The 6th Workshop on Language-Based Parallel Programming Models

(WLPP 2015)

Extending Gustafson-Barsis’s Law for Dual-Architecture Computing . . . . . . . 123

Ami Marowka

Free Scheduling of Tiles Based on the Transitive Closure of Dependence

Graphs. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133
Wlodzimierz Bielecki, Marek Palkowski, and Tomasz Klimek

Semiautomatic Acceleration of Sparse Matrix-Vector Product Using

OpenACC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143
Przemysław Stpiczyński

Multi-threaded Construction of Neighbour Lists for Particle Systems

in OpenMP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153
Rene Halver and Godehard Sutmann

NumCIL and Bohrium: High Productivity and High Performance . . . . . . . . . 166

Kenneth Skovhede and Simon Andreas Frimann Lund

Parallel Ant Brood Graph Partitioning in Julia . . . . . . . . . . . . . . . . . . . . . . 176

Jose Juan Mijares Chan, Yuyin Mao, Ying Ying Liu,
Parimala Thulasiraman, and Ruppa K. Thulasiram

The 5th Workshop on Performance Evaluation of Parallel Applications

on Large-Scale Systems

Scalability Model Based on the Concept of Granularity . . . . . . . . . . . . . . . . 189

Jan Kwiatkowski and Lukasz Olech

Performance and Power-Aware Modeling of MPI Applications for Cluster

Computing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 199
Jerzy Proficz and Paweł Czarnul

Running Time Prediction for Web Search Queries. . . . . . . . . . . . . . . . . . . . 210

Oscar Rojas, Veronica Gil-Costa, and Mauricio Marin

The Performance Evaluation of the Java Implementation of Graph500 . . . . . . 221

Magdalena Ryczkowska, Marek Nowicki, and Piotr Bala
 Contents – Part II XV

Workshop on Parallel Computational Biology (PBC 2015)

Performance Analysis of a Parallel, Multi-node Pipeline for DNA

Sequencing. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 233
Dries Decap, Joke Reumers, Charlotte Herzeel, Pascal Costanza,
and Jan Fostier

Parallelising the Computation of Minimal Absent Words . . . . . . . . . . . . . . . 243

Carl Barton, Alice Heliou, Laurent Mouchard, and Solon P. Pissis

Accelerating 3D Protein Structure Similarity Searching on Microsoft Azure

Cloud with Local Replicas of Macromolecular Data . . . . . . . . . . . . . . . . . . 254
Dariusz Mrozek, Tomasz Kutyła, and Bożena Małysiak-Mrozek

Workshop on Applications of Parallel Computation in Industry

and Engineering

Modeling and Simulations of Edge-Emitting Broad-Area Semiconductor

Lasers and Amplifiers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 269
Mindaugas Radziunas

Application of the Parallel INMOST Platform to Subsurface Flow

and Transport Modelling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 277
Igor Konshin, Ivan Kapyrin, Kirill Nikitin, and Kirill Terekhov

Parallel Procedure Based on the Swarm Intelligence for Solving

the Two-Dimensional Inverse Problem of Binary Alloy Solidification . . . . . . 287
Edyta Hetmaniok, Damian Słota, and Adam Zielonka

Minisymposium on HPC Applications in Physical Sciences

A Highly Parallelizable Bond Fluctuation Model on the Body-Centered

Cubic Lattice . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 301
Christoph Jentzsch, Ron Dockhorn, and Jens-Uwe Sommer

Genetic Algorithm and Exact Diagonalization Approach for Molecular

Nanomagnets Modelling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 312
Michał Antkowiak, Łukasz Kucharski, and Grzegorz Kamieniarz

Augmented Symmetry Approach to the DFT Simulations

of the Chromium-Based Rings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 321
Michał Wojciechowski, Bartosz Brzostowski, and Grzegorz Kamieniarz

Parallel Monte Carlo Simulations for Spin Models with Distributed Lattice . . . 332
Szymon Murawski, Grzegorz Musiał, and Grzegorz Pawłowski
XVI Contents – Part II

The Second Workshop on Applied High Performance Numerical

Algorithms in PDEs

Schwarz Preconditioner with Face Based Coarse Space for Multiscale

Elliptic Problems in 3D . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 345
Leszek Marcinkowski and Talal Rahman

A Compact Parallel Algorithm for Spherical Delaunay Triangulations . . . . . . 355

Florian Prill and Günther Zängl

On Conforming Local Post-refinement of Adjacent Tetrahedral

and Hexahedral Meshes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 365
Sergey Korotov and Talal Rahman

Fast Static Condensation for the Helmholtz Equation in a Spectral-Element

Discretization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 371
Immo Huismann, Jörg Stiller, and Jochen Fröhlich

An Iterative Regularization Algorithm for the TV-Stokes in Image

Processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 381
Leszek Marcinkowski and Talal Rahman

Discretization of the Drift-Diffusion Equations with the Composite

Discontinuous Galerkin Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 391
Konrad Sakowski, Leszek Marcinkowski, Pawel Strak, Pawel Kempisty,
and Stanislaw Krukowski

Additive Nonoverlapping Schwarz for h-p Composite Discontinuous

Galerkin. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 401
Piotr Krzyżanowski

Minisymposium on High Performance Computing Interval Methods

Up-to-date Interval Arithmetic: From Closed Intervals to Connected Sets

of Real Numbers. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 413
Ulrich Kulisch

Optimizing Cloud Use Under Interval Uncertainty. . . . . . . . . . . . . . . . . . . . 435

Vladik Kreinovich and Esthela Gallardo

The TOPSIS Method in the Interval Type-2 Fuzzy Setting . . . . . . . . . . . . . . 445

Ludmila Dymova, Pavel Sevastjanov, and Anna Tikhonenko

A Study on Vectorisation and Paralellisation of the Monotonicity Approach . . . 455

Iwona Skalna and Jerzy Duda

Preliminary Experiments with an Interval Model-Predictive-Control Solver. . . 464

Bartłomiej Jacek Kubica
 Contents – Part II XVII

Interval Nine-Point Finite Difference Method for Solving the Laplace

Equation with the Dirichlet Boundary Conditions . . . . . . . . . . . . . . . . . . . . 474
Malgorzata A. Jankowska

Workshop on Complex Collective Systems

How Do People Search: A Modelling Perspective . . . . . . . . . . . . . . . . . . . . 487

Isabella von Sivers, Michael J. Seitz, and Gerta Köster

A Sandpile Cellular Automata-Based Approach to Dynamic Job Scheduling

in Cloud Environment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 497
Jakub Gasior and Franciszek Seredynski

Conflict Solution According to “Aggressiveness” of Agents

in Floor-Field-Based Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 507
Pavel Hrabák and Marek Bukáček

Computer Simulation of Traffic Flow Based on Cellular Automata

and Multi-agent System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 517
Magda Chmielewska, Mateusz Kotlarz, and Jarosław Wąs

A Stochastic Optimal Velocity Model for Pedestrian Flow . . . . . . . . . . . . . . 528

Antoine Tordeux and Andreas Schadschneider

On the Evacuation Module SigmaEva Based on a Discrete-Continuous

Pedestrian Dynamics Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 539
Ekaterina Kirik, Andrey Malyshev, and Maria Senashova

Towards Effective GPU Implementation of Social Distances Model

for Mass Evacuation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 550
Adrian Kłusek, Paweł Topa, and Jarosław Wąs

GPU and FPGA Parallelization of Fuzzy Cellular Automata for the

Simulation of Wildfire Spreading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 560
Vasileios G. Ntinas, Byron E. Moutafis, Giuseppe A. Trunfio,
and Georgios Ch. Sirakoulis

eVolutus: A New Platform for Evolutionary Experiments . . . . . . . . . . . . . . . 570

Paweł Topa, Maciej Komosinski, Jarosław Tyszka, Agnieszka Mensfelt,
Sebastian Rokitta, Aleksander Byrski, and Maciej Bassara

Special Session on Algorithms, Methodologies and Frameworks

for HPC in Geosciences and Weather Prediction

Accelerating Extreme-Scale Numerical Weather Prediction . . . . . . . . . . . . . . 583

Willem Deconinck, Mats Hamrud, Christian Kühnlein,
George Mozdzynski, Piotr K. Smolarkiewicz, Joanna Szmelter,
and Nils P. Wedi
XVIII Contents – Part II

Scaling the GCR Solver Using a High-Level Stencil Framework

on Multi- and Many-Core Architectures . . . . . . . . . . . . . . . . . . . . . . . . . . . 594
Milosz Ciznicki, Michal Kulczewski, Piotr Kopta,
and Krzysztof Kurowski

Parallel ADI Preconditioners for All-Scale Atmospheric Models . . . . . . . . . . 607

Zbigniew P. Piotrowski, Bartlomiej Matejczyk, Leszek Marcinkowski,
and Piotr K. Smolarkiewicz

Author Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 619

 Contents – Part I

Parallel Architectures and Resilience

Exploring Memory Error Vulnerability for Parallel Programming Models. . . . 3

Isil Oz, Marisa Gil, Gladys Utrera, and Xavier Martorell

An Approach for Ensuring Reliable Functioning of a Supercomputer Based

on a Formal Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
Alexander Antonov, Dmitry Nikitenko, Pavel Shvets, Sergey Sobolev,
Konstantin Stefanov, Vadim Voevodin, Vladimir Voevodin,
and Sergey Zhumatiy

Sparse Matrix Multiplication on Dataflow Engines . . . . . . . . . . . . . . . . . . . 23

Vladimir Simic, Vladimir Ciric, Nikola Savic, and Ivan Milentijevic

Energy Efficient Calculations of Text Similarity Measure

on FPGA-Accelerated Computing Platforms . . . . . . . . . . . . . . . . . . . . . . . . 31
Michał Karwatowski, Paweł Russek, Maciej Wielgosz,
Sebastian Koryciak, and Kazimierz Wiatr

Numerical Algorithms and Parallel Scientific Computing

A Bucket Sort Algorithm for the Particle-In-Cell Method on Manycore

Architectures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
Andreas Jocksch, Farah Hariri, Trach-Minh Tran, Stephan Brunner,
Claudio Gheller, and Laurent Villard

Experience on Vectorizing Lattice Boltzmann Kernels for Multi- and

Many-Core Architectures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
Enrico Calore, Nicola Demo, Sebastiano Fabio Schifano,
and Raffaele Tripiccione

Performance Analysis of the Kahan-Enhanced Scalar Product on Current

Multicore Processors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
Johannes Hofmann, Dietmar Fey, Michael Riedmann, Jan Eitzinger,
Georg Hager, and Gerhard Wellein

Performance Analysis of the Chebyshev Basis Conjugate Gradient Method

on the K Computer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
Yosuke Kumagai, Akihiro Fujii, Teruo Tanaka, Yusuke Hirota,
Takeshi Fukaya, Toshiyuki Imamura, and Reiji Suda
XX Contents – Part I

Dense Symmetric Indefinite Factorization on GPU Accelerated

Architectures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
Marc Baboulin, Jack Dongarra, Adrien Rémy, Stanimire Tomov,
and Ichitaro Yamazaki

A Parallel Multi-threaded Solver for Symmetric Positive Definite

Bordered-Band Linear Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
Peter Benner, Pablo Ezzatti, Enrique S. Quintana-Ortí,
and Alfredo Remón

Parallel Algorithm for Quasi-Band Matrix-Matrix Multiplication . . . . . . . . . . 106

Dharma Teja Vooturi and Kishore Kothapalli

Comparative Performance Analysis of Coarse Solvers for Algebraic

Multigrid on Multicore and Manycore Architectures . . . . . . . . . . . . . . . . . . 116
Alex Druinsky, Pieter Ghysels, Xiaoye S. Li, Osni Marques,
Samuel Williams, Andrew Barker, Delyan Kalchev,
and Panayot Vassilevski

LU Preconditioning for Overdetermined Sparse Least Squares Problems . . . . 128

Gary W. Howell and Marc Baboulin

Experimental Optimization of Parallel 3D Overlapping Domain

Decomposition Schemes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 138
Sofia Guzzetti, Alessandro Veneziani, and Vaidy Sunderam

Parallel Implementation of the FETI DDM Constraint Matrix on Top of

PETSc for the PermonFLLOP Package . . . . . . . . . . . . . . . . . . . . . . . . . . . 150
Alena Vasatova, Martin Cermak, and Vaclav Hapla

Accelerating Sparse Arithmetic in the Context of Newton’s Method

for Small Molecules with Bond Constraints . . . . . . . . . . . . . . . . . . . . . . . . 160
Carl Christian Kjelgaard Mikkelsen, Jesús Alastruey-Benedé,
Pablo Ibáñez-Marín, and Pablo García Risueño

Massively Parallel Approach to Sensitivity Analysis on HPC Architectures

by Using Scalarm Platform. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172
Daniel Bachniak, Jakub Liput, Lukasz Rauch, Renata Słota,
and Jacek Kitowski

GPU Implementation of Krylov Solvers for Block-Tridiagonal Eigenvalue

Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 182
Alejandro Lamas Daviña and Jose E. Roman

Parallel Non-numerical Algorithms

Comparison of Large Graphs Using Distance Information . . . . . . . . . . . . . . 195

Wojciech Czech, Wojciech Mielczarek, and Witold Dzwinel
 Contents – Part I XXI

Fast Incremental Community Detection on Dynamic Graphs. . . . . . . . . . . . . 207

Anita Zakrzewska and David A. Bader

A Diffusion Process for Graph Partitioning: Its Solutions and Their

Improvement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 218
Andreas Jocksch

A Parallel Algorithm for LZW Decompression, with GPU Implementation. . . 228

Shunji Funasaka, Koji Nakano, and Yasuaki Ito

Parallel FDFM Approach for Computing GCDs Using the FPGA . . . . . . . . . 238
Xin Zhou, Koji Nakano, and Yasuaki Ito

Parallel Induction of Nondeterministic Finite Automata . . . . . . . . . . . . . . . . 248

Tomasz Jastrzab, Zbigniew J. Czech, and Wojciech Wieczorek

Parallel BSO Algorithm for Association Rules Mining Using

Master/Worker Paradigm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 258
Youcef Djenouri, Ahcene Bendjoudi, Djamel Djenouri,
and Zineb Habbas

Tools and Environments for Parallel/Distributed/Cloud Computing

Distributed Computing Instrastructure as a Tool for e-Science . . . . . . . . . . . 271

Jacek Kitowski, Kazimierz Wiatr, Łukasz Dutka, Maciej Twardy,
Tomasz Szepieniec, Mariusz Sterzel, Renata Słota, and Robert Pająk

A Lightweight Approach for Deployment of Scientific Workflows in Cloud

Infrastructures. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 281
Bartosz Balis, Kamil Figiela, Maciej Malawski, Maciej Pawlik,
and Marian Bubak

Distributed Execution of Dynamically Defined Tasks on Microsoft Azure . . . 291

Piotr Wiewiura, Maciej Malawski, and Monika Piwowar

Scalable Distributed Two-Layer Block Based Datastore . . . . . . . . . . . . . . . . 302

Adam Krechowicz, Stanisław Deniziak, Mariusz Bedla,
Arkadiusz Chrobot, and Grzegorz Łukawski

Metadata Organization and Management for Globalization of Data Access

with Onedata. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 312
Michał Wrzeszcz, Krzysztof Trzepla, Rafał Słota, Konrad Zemek,
Tomasz Lichoń, Łukasz Opioła, Darin Nikolow, Łukasz Dutka,
Renata Słota, and Jacek Kitowski

Hypergraph Based Abstraction for File-Less Data Management. . . . . . . . . . . 322

Bartosz Kryza and Jacek Kitowski
XXII Contents – Part I

Using Akka Actors for Managing Iterations in Multiscale Applications . . . . . 332

Katarzyna Rycerz and Marian Bubak

Application of Parallel Computing

Synthetic Signature Program for Performance Scalability . . . . . . . . . . . . . . . 345

Javier Panadero, Alvaro Wong, Dolores Rexachs, and Emilio Luque

FEniCS-HPC: Automated Predictive High-Performance Finite Element

Computing with Applications in Aerodynamics. . . . . . . . . . . . . . . . . . . . . . 356
Johan Hoffman, Johan Jansson, and Niclas Jansson

Accelerating NWChem Coupled Cluster Through Dataflow-Based

Execution. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 366
Heike Jagode, Anthony Danalis, George Bosilca, and Jack Dongarra

Parallelization and Optimization of a CAD Model Processing Tool from the

Automotive Industry to Distributed Memory Parallel Computers . . . . . . . . . . 377
Luis Ayuso, Juan J. Durillo, Bernhard Kornberger, Martin Schifko,
and Thomas Fahringer

GPU Accelerated Simulations of Magnetic Resonance Imaging of Vascular

Structures. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 389
Krzysztof Jurczuk, Dariusz Murawski, Marek Kretowski,
and Johanne Bezy-Wendling

Parallel Algorithms for Wireless LAN Planning . . . . . . . . . . . . . . . . . . . . . 399

Andrzej Gnatowski and Jan Kwiatkowski

Toward Parallel Modeling of Solidification Based on the Generalized Finite

Difference Method Using Intel Xeon Phi . . . . . . . . . . . . . . . . . . . . . . . . . . 411
Lukasz Szustak, Kamil Halbiniak, Adam Kulawik, Joanna Wrobel,
and Pawel Gepner

Optimized Parallel Model of Human Detection Based on the Multi-Scale

Covariance Descriptor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 423
Nesrine Abid, Tarek Ouni, Kais Loukil, A. Chiheb Ammari,
and Mohamed Abid

Neural Networks, Evolutionary Computing and Metaheuristics

Parallel Extremal Optimization with Guided Search and Crossover Applied

to Load Balancing. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 437
Eryk Laskowski, Marek Tudruj, Ivanoe De Falco, Umberto Scafuri,
Ernesto Tarantino, and Richard Olejnik
 Contents – Part I XXIII

Parallel Differential Evolution in the PGAS Programming Model

Implemented with PCJ Java Library . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 448
Łukasz Górski, Franciszek Rakowski, and Piotr Bała

Adaptation of Deep Belief Networks to Modern Multicore Architectures . . . . 459

Tomasz Olas, Wojciech K. Mleczko, Robert K. Nowicki,
and Roman Wyrzykowski

Implementing Deep Learning Algorithms on Graphics Processor Units . . . . . 473

Karol Grzegorczyk, Marcin Kurdziel, and Piotr Iwo Wójcik

Fuzzy Transducers as a Tool for Translating Noisy Data in Electrical

Load Forecast System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 483
Mariusz Flasiński, Janusz Jurek, and Tomasz Peszek

Towards a Scalable Distributed Fitness Evaluation Service . . . . . . . . . . . . . . 493

Włodzimierz Funika and Paweł Koperek

Minisymposium on GPU Computing

Revisiting the Gauss-Huard Algorithm for the Solution of Linear Systems

on Graphics Accelerators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 505
Peter Benner, Pablo Ezzatti, Enrique S. Quintana-Ortí,
and Alfredo Remón

Increasing Arithmetic Intensity in Multigrid Methods on GPUs

Using Block Smoothers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 515
Matthias Bolten and Oliver Letterer

Optimized CUDA-Based PDE Solver for Reaction Diffusion Systems on

Arbitrary Surfaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 526
Samira Michèle Descombes, Daljit Singh Dhillon, and Matthias Zwicker

Comparing Different Programming Approaches for SpMV-Operations on

GPUs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 537
Jan P. Ecker, Rudolf Berrendorf, Javed Razzaq, Simon E. Scholl,
and Florian Mannuss

IVM-Based Work Stealing for Parallel Branch-and-Bound on GPU . . . . . . . . 548

Jan Gmys, Mohand Mezmaz, Nouredine Melab, and Daniel Tuyttens

Massively Parallel Construction of the Cell Graph. . . . . . . . . . . . . . . . . . . . 559

Krzysztof Kaczmarski, Paweł Rzążewski, and Albert Wolant

Benchmarking the Cost of Thread Divergence in CUDA . . . . . . . . . . . . . . . 570

Piotr Bialas and Adam Strzelecki
XXIV Contents – Part I

Special Session on Efficient Algorithms for Problems with Matrix

and Tensor Decompositions

Fast Algorithm for the Fourth-Order Elliptic Problem Based on Orthogonal

Matrix Decomposition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 583
Paolo Di Stolfo and Marian Vajteršic

Performance of the Parallel One-Sided Block Jacobi SVD Algorithm

on a Modern Distributed-Memory Parallel Computer . . . . . . . . . . . . . . . . . . 594
Shuhei Kudo, Yusaku Yamamoto, Martin Bečka, and Marian Vajteršic

New Approach to Local Computations in the Parallel One–Sided Jacobi

SVD Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 605
Martin Bečka and Gabriel Okša

Author Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 619

 The Third Workshop on Models,
Algorithms, and Methodologies for
Hierarchical Parallelism in New HPC
Systems
 Virtualizing CUDA Enabled GPGPUs
on ARM Clusters

Raffaele Montella1(B) , Giulio Giunta1 , Giuliano Laccetti2 , Marco Lapegna2 ,

Carlo Palmieri1 , Carmine Ferraro1 , and Valentina Pelliccia1
1
Department of Science and Technologies, University of Napoli Parthenope,
Centro Direzionale di Napoli, Isola C4, 80143 Naples, Italy
{raffaele.montella,giulio.giunta,carlo.palmieri,carmine.ferraro,
valentina.pelliccia}@uniparthenope.it
2
Department of Mathematics and Applications, University of Naples Federico II,
Complesso Universitario Monte S. Angelo, Via Cintia, Naples, Italy
{giuliano.laccetti,marco.lapegna}@unina.it

Abstract. The acceleration of inexpensive ARM-based computing

nodes with high-end CUDA enabled GPGPUs hosted on x86 64 machines
using the GVirtuS general-purpose virtualization service is a novel app-
roach to hierarchical parallelism. In this paper we draw the vision
of a possible hierarchical remote workload distribution among differ-
ent devices. Preliminary, but promising, performance evaluation data
suggests that the developed technology is suitable for real world
applications.

Keywords: GPGPU · Virtualisation · Remoting · Cloud computing ·

ARM · CUDA · HPC

1 Introduction
The High Performance Cloud Computing is now offering an outstanding elas-
tic infrastructure providing the performances required by most e-science appli-
cations. Public, private, hybrid and campus clouds are production level reali-
ties managing virtual clusters instanced on cloud infrastructures in a relatively
straightforward fashion [10]. This issue impacts on science democratization in
a scenario where the scientific computing massively relies on general purpose
graphics processing units (GPGPUs) to accelerate data parallel computing tasks
especially using NVIDIA CUDA framework [7].
The virtualization currently provided by popular open source hypervisors
(XEN, KVM, Virtual Box) does not allow software based transparent use of
accelerators as CUDA based GPUs while VMWare and XEN support GPU
on the basis of hardware virtualization provided natively by NVIDIA GRID
devices [8].
In this paper we present the updated component GVirtuS (Generic Virtual-
ization Service) as results in GPGPUs software based transparent virtualization

c Springer International Publishing Switzerland 2016
R. Wyrzykowski et al. (Eds.): PPAM 2015, Part II, LNCS 9574, pp. 3–14, 2016.
DOI: 10.1007/978-3-319-32152-3 1
4 R. Montella et al.

and remoting with the use of NVIDIA CUDA as main aim [4]. In the latest GVir-
tuS incarnation we enforced the architecture independence making it working
with both CUDA and OpenCL on Intel and ARM architecture.
While GVirtuS is a transparent and VMM independent framework to allows
an instanced virtual machine to access GPUs implementing various communica-
tor components (TCP/IP, VMCI for VMware, VMSocket for KVM) to connect
the front- end in guestOS and back-end in hostOS, rCUDA [2], GViM [6] and
vCUDA [15] are three recent research projects on CUDA virtualization in GPU
clusters and virtual machines as GPGPU library. They all use an approach sim-
ilar to GVirtuS.
The rest of the paper is organized in the following way: the Sect. 2 is about
how GVirtuS works on different architectures; the Sect. 3 is dedicated to the
design and technical issues about the latest version of GVirtuS; the Sect. 4 deals
with the scenarios and prototypal applications; the Sect. 5 shows the preliminary
evaluation results; the Sect. 6 is about the conclusions and the future directions
of this promising research.

2 GVirtuS on Heterogeneous Architectures

Different application fields motivate an ARM port of GVirtuS such as, but
not limited to Sensor as a Service [1], High Performance Internet of Things
(HPIoT) [9] and High Performance Cloud Computing (HPCC) [3].
In order to fit the GPGPU/x86 64/ARM application into our generic virtual-
ization system we mapped the back-end on the x86 machine directly connected to
the GPU based accelerator device and the front-end on the ARM board(s) using
the GVirtuS tcp/ip based communicator. GVirtuS as NVIDIA CUDA remoting
and virtualization tool achieve good results in terms of performances and system
transparency [11].
The CUDA applications are executed on the ARM board through the GVir-
tuS front-end. Thanks to the GVirtuS architecture, the front-end is the only
component needed on the guest side. This component acts as a transparent vir-
tualization tool giving to a simple and inexpensive ARM board the illusion to
be directly connected to a high-end CUDA enabled GPGPU device or devices
(Fig. 1).
The computing nodes of a regular old-style cluster behave as input/output
nodes for ARM based inexpensive sub-clusters. In this way the amount of heat
produced decreases while the high computing power demanding applications
have to be refactored in order to fit this new heterogenic approach. Thanks to
GVirtuS, these devices are seen by each of the ARM based sub-cluster computing
nodes as directly connected to them in a transparent way. This vision permits to
gain more computing power reducing the expensive, power hungry and heat pro-
ducing x86 64 based computing nodes. In the same way, this approach increases
the parallelism at the sub-cluster level and, last but not the least, unchain the
high-end GPGPU power to ARM based computing nodes [12].
 Virtualizing CUDA Enabled GPGPUs on ARM Clusters 5

Fig. 1. The GVirtuS architecture independent from computing architecture (ARM,

x86 64) and acceleration model (CUDA, OpenCL).

3 Design and Technical Issues

In GVirtuS we used the classic split-driver approach based on front-end, com-
municator and back-end components.
The front-end is a module that uses the driver APIs supported by the plat-
form. The interposer library provides the familiar driver API abstraction to the
guest application. It collects the request parameters from the application and
passes them to the back-end driver, converting the driver API call into a cor-
responding frontend driver call. When a callback is received from the frontend
driver, it delivers the response messages to the application. In GVirtuS the front-
end is deployed on the virtual machine instance and it is implemented as a stub
library.

Fig. 2. The GVirtuS approach to the split driver model.

Another random document with
no related content on Scribd:
The Project Gutenberg eBook of Heedless Hetty
This ebook is for the use of anyone anywhere in the United States
and most other parts of the world at no cost and with almost no
restrictions whatsoever. You may copy it, give it away or re-use it
under the terms of the Project Gutenberg License included with this
ebook or online at www.gutenberg.org. If you are not located in the
United States, you will have to check the laws of the country where
you are located before using this eBook.

Title: Heedless Hetty

Author: Annette Lyster

Release date: August 25, 2023 [eBook #71484]

Language: English

Original publication: London: The Religious Tract Society, 1890

*** START OF THE PROJECT GUTENBERG EBOOK HEEDLESS

HETTY ***
Transcriber's note: Unusual and inconsistent spelling is as printed.
 HEEDLESS HETTY

BY

ANNETTE LYSTER

AUTHOR OF

"KARL KRAPP'S LITTLE MAIDENS," "WHAT SHE COULD,"

"RALPH TRULOCK'S CHRISTMAS ROSES," "THE

RUTHERFORD FROWN," ETC.

London

THE RELIGIOUS TRACT SOCIETY

56, PATERNOSTER ROW; 65, ST. PAUL'S CHURCHYARD

 AND 164, PICCADILLY

BUTLER & TANNER,

THE SELWOOD PRINTING WORKS,

FROME, AND LONDON.

CONTENTS.

CHAPTER

I. MRS. EYRE WANTS A GIRL

II. LITTLE FLO

III. FLO'S KITTEN

IV. "WHAT'S IN A NAME?"

V. UPS AND DOWNS

VI. MRS. GOODENOUGH'S ADVICE

VII. CHERRIES

VIII. AT THE SEA-SIDE

IX. THE BIG BLACK DOG

X. FORGIVEN
 HEEDLESS HETTY.

CHAPTER I.

MRS. EYRE WANTS A GIRL.

"GOOD-EVENING, Mrs. Hardy," said a pleasant voice, as the

speaker tapped with her hand upon the half-open door of Mrs.
Hardy's cottage.

Mrs. Hardy was a washerwoman, and her visitor knew that

sometimes there was but scant room in her kitchen for strangers;
indeed, she often wondered how the children managed on a wet day,
and how the little ones escaped scalds and burns. However, this
being Friday evening, the actual work was over, and the big deal
table was piled with heaps of snowy linen, which Mrs. Hardy and her
daughter Martha were sorting out and packing in nice large baskets,
ready to be carried home the next day.

"Oh, come in, Mrs. Eyre; you needn't be afraid of the wash-tubs or
the hot irons to-day. We've finished everything, ma'am."

"And such lots of things," said Mrs. Eyre, as she took the seat
offered her by Martha. "I am sure I don't know how you get through it
all, Mrs. Hardy."

"Well, ma'am, it takes a power of method. When I first took up this

business, often I had all the ironing to do on Saturday, or the most of
it; and then 'twas hurry-scurry in the evening to get the things home.
I used to get so worried that I fairly thought I'd die. And one Saturday
morning, who should come in but your good mother, ma'am, that's in
heaven now; and the pleasant way she had. There was I on that
chair in the corner, crying, and all the children crying round me. So
says she, 'My poor Hannah, are you fretting so badly yet?' I dried my
eyes and felt ashamed—for she thought I was crying for my poor
man that had died about a year before; and I had to confess that I
was crying because I didn't see how to get the ironing done. But
indeed I have too much talk—all this don't matter to you."

"Ah, but it does—anything about my dear mother matters to me. Go

on with your work, Mrs. Hardy, and tell me the rest of your story. I'm
very sure she helped you."

"That she did, ma'am. The place was in a mess, with half-done
collars and cuffs on the chairs, and the rector's shirts piled on the
table; some of the linen in the baskets, and more on the stool over
there. Well, not a word did she say about that, though I knew she
saw the untidy way the place was in well enough. Says she, 'The
first thing, Hannah, is to get the ironing finished, and then you and I
will have a talk. Suppose you send the children out, all but Annie and
Matty, who can bring us the hot irons. I am a good ironer, and I'll help
you all I can,' says she; and tucked up her sleeves and went to work
as if she'd done nothing else all her life. So pleasant with the two
girls too, with a word when they brought the irons, that they worked
as willing as possible. And of course I wasn't idle; so, before I
thought it could be done, the clothes were in the basket. Annie and
Matty carried them off; and your dear mother sat down and talked to
me."

"'It's all method, Hannah,' she says. 'People sometimes ask me how
I get through so much work, and am never in a hurry; now it is just
method,' says she. And before she left me she wrote out that paper
that you see on the wall there. See, ma'am. 'Monday, collect the
wash, put the things in soak, and boil such articles as must be
boiled. Tuesday—' You see, ma'am? it's all laid out. 'And make your
girls help you when they come home from school; it will be much
better for them than running about idle; be pleasant with them, and
they will like it well.' Ah, she was a great help to me that day, the
dear lady."

"I think she had a willing hearer, Mrs. Hardy."

"Yes, ma'am, because she had a pleasant, kindly, friendly way. It

wasn't, 'My good woman, your house is little better than a pigsty,' or,
'Hannah Hardy, why don't you manage a little better about your
work?'—not she. Ah, a real lady she was, and a real friend to me."

"But people may often mean very kindly who have not my dear
mother's pleasant ways. That kind of manner is a great gift, but
some people have not got it, and that they cannot help. They must
do the best they can."

"The best they could do, Mrs. Eyre, meaning no offence, would be to
stay at home. Folks are only human after all, if they are
washerwomen; and they have their feelings."

"Miss Posnett was very kind that time I had a bad whitlow," put in
Martha.

"Who's named Miss Posnett?" inquired her mother. "Mind your

manners, Matty, and name no names."

"All this time, Mrs. Hardy, I have not told you my errand here to-day.
You know the doctors say that my little Flora must not be allowed to
walk, or even to stand. She has never been strong since her bad fall.
Neither will they allow her to be drawn about in a little carriage,
because she gets so dreadfully cold. They say she must be carried.
The consequence of this is that I must have a girl to help me, for I
never could carry her—she is light enough, but I am not very strong.
Now I remember what a comfort your Annie was to me during the
short time I had her, and I want to know if you can spare me one of
your other girls. It may be only for a time, for Flora may get well and
strong again, but I would teach her as I taught Annie, and then when
she leaves me she could get a good place, as Annie has done."
"Lady Drysdale says that Annie is a right good servant, and that
even the grand nurse is pleased with her. Well, it would be the
making of Matty, but I can't spare her, and that's the plain truth.
Though I hate refusing you, ma'am."

"But is not Hetty fifteen? Older, I think, than Matty was when Annie
came to me."

"No doubt, ma'am. But Matty was Matty, and Hetty is Hetty. There's a
sight of difference in girls!"

"Mother," said Matty, "I know you could not spare me, and I shouldn't
like to leave you. But if Mrs. Eyre would try Hetty. She is very strong,
and very willing. Fond of children too, and used to them—very good-
tempered Hetty is. Don't give Mrs. Eyre a bad opinion of poor Hetty,
mother, for it's my belief she would do well."

Mrs. Hardy left off working and sat down, in a curiously divided frame
of mind. Hetty had been peculiarly heedless and troublesome that
whole week, and was just now crying in the bedroom behind the
kitchen, after what her mother called "a raking good scolding." It was
hard to keep silence, for she had been very angry, and yet she had a
notion that Hetty might do better away from home, and from all the
temptations to idleness that beset her there. Not that the girl was
exactly idle, for she could work well, and liked to work, but let any
one interrupt her, if it were only a kitten running into the kitchen, or a
noise in the street, and the work was forgotten. Only last night she
had been bringing a hot iron from the fire, when a fiddle struck up a
doleful air outside, and Hetty clapped down the iron on the ironing
blanket and ran out of the house. Mrs. Hardy had been apprised of
her carelessness by the horrible smell of the burning blanket, in
which there was, of course, a big hole. It was the last of many sins,
and no one could deny that the "raking good scolding" was well
deserved.

"Matty, are you in your right mind?" asked Mrs. Hardy.

"Yes, mother. If Hetty was in Mrs. Eyre's service, or carrying Miss Flo
while Mrs. Eyre drew the little carriage, she would be safe enough.
And she would do her best, and indeed, ma'am, Hetty is a good girl.
Mother will tell you, she never was known to tell a lie yet."

"It is true enough," Mrs. Hardy admitted.

"There's not a bit of harm in Hetty. I'll even allow that she means
well. But I couldn't find it in my conscience to recommend you to try
her, ma'am. There's Mrs. Simmons' Emma, she's sixteen, and a
steady girl."

"No, no; I will not have her. I heard Emma Simmons using such
coarse, violent language to her brother the other day. I would not like
my children to hear it."

"You will never hear a bad word from Hetty, ma'am," said Matty. "She
is heedless, she does forget things, I know. But she's a good girl,
that knows the Commandments, and wants to keep them; and
mother knows that too. Will you see her, ma'am? I know she'd do
well with you. Hetty, come here."

The door of the inner room opened—Hetty must have been pretty
close to it. Out she came—a tall, well-made girl, much taller than
neat little Matty. Mrs. Eyre knew her face very well, which was lucky,
for just now any one might have objected to her, as likely to frighten
the children. Her eyes were quite lost in her swollen eyelids and
cheeks, her poor lips were swelled, her whole face was crimson, and
her apron was soaking wet, having been freely cried into. Her stuff
skirt was torn in several places, her calico bodice displayed two
corking pins where buttons were wanting. Her thick, short, brown
hair hung over her forehead; altogether, as she sneaked into the
room and stood, ashamed to look up, she presented a most forlorn
appearance.

"Hetty, did you hear what we were saying?" asked Matty.

"Yes; I couldn't help hearing."

The girl had a very sweet voice, and spoke nicely, Mrs. Eyre
observed.

"You're a nice-looking article to be looking for a situation," remarked

Mrs. Hardy. "Now, how often would you clap the child on the ground
and run off, if you heard the squeak of Blind Davie's fiddle?"

"Mother, sure you know, when the children were little, 'twas always
me that kept them best. I love little children, and I would never hurt
one—and you know that, mother."

"Well, I don't think you would, to say true," answered the mother. "Try
her for a month, Mrs. Eyre, without wages. Washing is a scattery
trade, no doubt—takes a power of method. And Hetty has no
method."

"Oh, do, Mrs. Eyre—please do! If—if—I didn't see—or hear—Oh,

ma'am, do try me! I'll do my best to please you."

"Well, Hetty, I will try you. Come to me on Monday."

"To-morrow, ma'am, if you like. I could have her ready."

"Monday will do. Come early, Hetty. I will try you for a month, and
after that, if you stay with me, I will pay you at the rate of five pounds
a year, paid quarterly, and we will count this first month in your first
quarter. You will have plenty to do, but you look strong and healthy,
so you will not find it too much. But you must try to remember what I
tell you to do."

"I will try, indeed, ma'am. I am real tired of always being wrong."

"Then good-bye until Monday. And don't cry any more, Hetty; crying
never did any good yet. If you will remember that you are one of
Christ's servants as well as mine, and that to please Him should be
your first thought, I am sure you will get over your heedless ways.
Good-bye, Mrs. Hardy. I must go now."
But Mrs. Hardy followed her visitor out of the house and shut the
door.

"I wouldn't let her go to you, ma'am, only I do think she may do well
with you. She is fond of children, and children take to her at once.
My little Bob, that was a sickly baby, was never so good as when
Hetty had him. And I know things go on here that take her mind off
her work. People coming and going, and the door obliged to be kept
open, and all. She may be more correct-like when there's none of
that going on. But don't you be soft with her. She's a girl that takes a
deal of scolding, and I'm just afraid you are not one to give her
enough of it. And if you praise her, ma'am, her head's turned directly.
She's not a bit like Annie; so don't expect it."

"Ah, well, I will try her for a month, Mrs. Hardy. I can promise no
more than that."

"Nor would I ask more, ma'am. Good-bye, ma'am, and thank you. If
you tame our Hetty,—Heedless Hetty, as our boys call her,—I'll say
you could do anything."

"I shall try to make her tame herself, Mrs. Hardy."

"She'll never do that, ma'am."

"Ah, Mrs. Hardy, you don't remember that she will not have to do it in
her own strength. That would be too much for any of us. But think of
the words, 'If any lack wisdom, let him ask of God, who giveth
liberally, and upbraideth not.' My mother said to me once, 'The
difficulty does not lie so much in your faults as in the fact that you do
not see that they are sins; and even when you do see this, you do
not go the right way to be cured of them; for nothing but the love of
God shed abroad in our hearts by His Holy Spirit can cure the least
fault.' But I must really get home now; so good-bye, Mrs. Hardy."

Mrs. Hardy went indoors again. She found that the two girls had
finished putting the things into the baskets, and she did not observe
that Hetty, in the hurry of her mind, had put three heavy sheets on
the top of Miss Posnett's stiff collars and frilled nightcaps. But when
Miss Posnett sent those articles back on Monday, it was well for
Hetty that she was out of the way.

"Hetty," began Mrs. Hardy, "you are in luck for once, and I hope
you're aware of it. Mrs. Eyre ain't rich, but a lady down to her very
shoes, and she'll be kind to you. If you lose this chance, I think you'd
better emigrate to some savage place where folk won't mind your
wild ways; only mind they're no cannibals, for you're plump and
young, and if they found you of no use, they might think it better to
eat you."

"Mother! how can you?" cried Hetty.

"Take off that dress now, and give it a good patching. Matty, look up
all her things; we must mend and wash them. And then I'll go and
buy her some neat aprons. Oh, dear, look at her Sunday frock! Did
you sleep in it, Hetty? Here, Matty—your fingers are cleverer than
mine; mend this, like a good girl. Even if we get her back in a week,
let us send her out decent."

CHAPTER II.
LITTLE FLO.

IF Hetty had been allowed to follow her own way, she would have
gone to Adelaide Terrace at six o'clock in the morning, to show her
zeal, but her mother would not hear of it.
"You'd find her in bed, most likely, and some one would have to get
up to let you in. No; at nine Mr. Eyre goes off to his business, and
you be there soon after nine. Try and keep out of mischief till then—if
you can."

As the clock struck nine, Matty and Hetty set out together, carrying
between them the small wooden, paper-covered box which
contained Hetty's very modest outfit. She could easily have carried it
alone, but Matty thought it looked better between them, and perhaps
was not sorry to make sure that Heedless Hetty went at once to her
new home, and reached it in a presentable state. Hetty had cried, of
course, when saying good-bye to her mother and brothers, but for all
that she was in fine spirits, and full to the lips of the most excellent
resolutions.

"Matty," said she, "you tell Dan that he may leave off calling me
Heedless Hetty. I mean to learn to be a good servant, as Annie did;
and when I come home, it's Handy Hetty that Dan will be calling me."

"Look where you're going! There now! You've stepped into that
puddle—the only one in the road—and dirted your shoe, that Dan
blacked so lovely for you!"

"Oh, so I have! Wait! I must rub it off," cried Hetty, and setting down
her end of the box into the puddle which had already soiled her
shoe, she ran to the side of the road, where she had espied some
grass.

"Well, of all the girls!" said Matty to herself, as she tried to see if the
box was very wet. "Heedless Hetty will suit well enough yet a bit.
Come along; there'll be a scraper and a mat at Mrs. Eyre's, and if I
could see you safe there, I'd be glad."

Hetty came back, looking a little ashamed of herself. She did not
refer to her message to Dan, and in a few moments they reached
No. 1, Adelaide Terrace.
"Set the box down on the step. Give me a kiss, Hetty. Dear heart! Do
try to do well here. Mind, if you don't, even I must allow that it is your
own fault, and you'll never be worth anything if you don't take hold
now and mind what you're about. You've got all your senses like
other girls, and it is high time you began to use them."

"I do try, Matty. I never mean to do wrong. But somehow I do forget

things so easily."

"Because you don't try to keep your mind fixed on what you're doing,
and so you're at the mercy of every little thing that happens. Just
heedless—that's about it, Hetty dear. Do you ever pray to be made
heedful?"

"Oh, Matty! I'd never think of asking such a thing. I pray to be made
good, and holy, and kept from saying bad words, like Emma
Simmons, or stealing, like—"

"Now listen, Hetty. You've no temptations to do those things, thanks

to your good, careful mother. It's just as if a railway man in the
station down yonder should pray that he might not be drowned,
when there is not so much as a pond in the place big enough to hold
him, and never give a thought to the real dangers he lives among.
You pray for what you really want, Hetty. That kind of prayer is only
words. Promise me you will, dear—quick! For I must ring now."

"I'll try. Oh, Matty, whatever shall I do without you? I wish—"

But the door opened, and the figure of an ancient dame, who spent
her mornings in doing Mrs. Eyre's rough work, appeared before
them.

"So here's our new nursemaid," said she, laughing at Hetty's

dolorous face. "Which of you is coming here?"

"This is Hetty," said the elder sister.

"Ah, I wish it was you," was the reply.

Hetty would have felt less abashed had she known that the speaker
would have made the same remark if Matty had been the new maid.

"Good-bye, Hetty. I'll try to see you some evening; but you know we'll
be very busy, wanting your help."

Matty lifted the box into the hall, pushed her sister in very gently, and
went quickly away. Hetty felt and looked very forlorn; and, but for the
amused smile on Mrs. Goodenough's wrinkled face, she would have
begun to cry again. But now a door opened, and Mrs. Eyre, with her
baby in her arms, came into the narrow hall.

"Hetty, how nice and early you have come! Leave your box there for
the present, and come here to Miss Flo; she is very anxious to see
you."

She led Hetty into the parlour, where all her children were
assembled. There were four—two little girls, a boy of about three,
and the baby, who was a boy also.

The eldest girl, whom they called Lina, was a pretty, active, healthy-
looking little maiden, about six years old, very good-tempered, and
very fond of her own way—which, after all, is not a very uncommon
liking. Then came Flora, who was five, but such a tiny creature that it
was hard to believe that she was so old. Little Edgar, the eldest boy,
was quite as big and far heavier than this poor wee fairy. She lay on
a sofa near the window, and her small face, which was usually very
grave and pathetic in its sad patience, was all alive now with anxiety
and curiosity. She had lovely dark eyes and pretty brown curls, but
her face was too white and pinched to be called pretty, though she
had been a lovely baby. She fixed her eyes on Hetty's face, and a
little shy, timid smile crept over her own; then she said, in a soft,
clear little voice,—

"Is this Hetty? Oh, mamma, she looks kind. I shall not be afraid of
Hetty."
She spoke quite plainly and distinctly, much more so than did Lina,
who often gabbled so fast that it was hard to understand her.

"This is Hetty, who will carry my little Flo so safely that there will be
nothing to be afraid of. My little Flo—she likes Hetty, I think."

"I like Hetty. Her eyes look kind. Please, Hetty, stoop and kiss me.
Will you be kind, Hetty, and patient with me? I'm sometimes peevish,
I'm afraid."

"Kind? Oh, Miss Flora, that I will!" said Hetty earnestly.

"But don't cry, Hetty. Why should you cry?"

"Well, miss, you see I've just said good-bye to my sister. But I won't
cry," Hetty answered, with a choke in her voice. The sight of the child
had touched her soft heart.

"Now, Hetty, before you take off your hat, please take Miss Lina to
school. It is close by, and she knows the way. Make haste back, for
Miss Flo is longing to be out in the sunshine."

"So you see, Flo," cried Lina, "after all your saying that Hetty is to be
yours, I am to have her first." And Lina nodded her curly head at the
little one.

"She belongs to me," Flo calmly replied. "But I will not be selfish. You
can have her now."

Lina laughed, and ran off for her hat. All the way to school she
chattered unceasingly, but Hetty had no idea what it was all about.
She had left the child at her school, and was on her way back, when
she met her brother Ned, who was on his way to the shop where he
was errand boy.

"Hilloa, Hetty! Is this you?"

"I've been leaving Miss Lina at school. Oh, Ned, if you only saw Miss
Flo! she's such a little darling."