You are on page 1of 54

Architecture of Computing Systems

ARCS 2016 29th International


Conference Nuremberg Germany April
4 7 2016 Proceedings 1st Edition Frank
Hannig
Visit to download the full and correct content document:
https://textbookfull.com/product/architecture-of-computing-systems-arcs-2016-29th-int
ernational-conference-nuremberg-germany-april-4-7-2016-proceedings-1st-edition-fra
nk-hannig/
More products digital (pdf, epub, mobi) instant
download maybe you interests ...

Architecture of Computing Systems ARCS 2020 33rd


International Conference Aachen Germany May 25 28 2020
Proceedings André Brinkmann

https://textbookfull.com/product/architecture-of-computing-
systems-arcs-2020-33rd-international-conference-aachen-germany-
may-25-28-2020-proceedings-andre-brinkmann/

Architecture of Computing Systems ARCS 2014 27th


International Conference Lübeck Germany February 25 28
2014 Proceedings 1st Edition Erik Maehle

https://textbookfull.com/product/architecture-of-computing-
systems-arcs-2014-27th-international-conference-lubeck-germany-
february-25-28-2014-proceedings-1st-edition-erik-maehle/

Business Information Systems 19th International


Conference BIS 2016 Leipzig Germany July 6 8 2016
Proceedings 1st Edition Witold Abramowicz

https://textbookfull.com/product/business-information-
systems-19th-international-conference-bis-2016-leipzig-germany-
july-6-8-2016-proceedings-1st-edition-witold-abramowicz/

Communication and computing systems : proceedings of


the International Conference on Communication and
Computing Systems (ICCCS 2016), Gurgaon, India, 9-11
September, 2016 1st Edition B.M.K. Prasad
https://textbookfull.com/product/communication-and-computing-
systems-proceedings-of-the-international-conference-on-
communication-and-computing-systems-icccs-2016-gurgaon-
Ignition Systems for Gasoline Engines: 3rd
International Conference, November 3-4, 2016, Berlin,
Germany 1st Edition Michael Günther

https://textbookfull.com/product/ignition-systems-for-gasoline-
engines-3rd-international-conference-november-3-4-2016-berlin-
germany-1st-edition-michael-gunther/

Microactuators and Micromechanisms: Proceedings of


MAMM-2016, Ilmenau, Germany, October 5-7, 2016 1st
Edition Lena Zentner

https://textbookfull.com/product/microactuators-and-
micromechanisms-proceedings-of-mamm-2016-ilmenau-germany-
october-5-7-2016-1st-edition-lena-zentner/

Computational Science and Its Applications ICCSA 2016


16th International Conference Beijing China July 4 7
2016 Proceedings Part IV 1st Edition Osvaldo Gervasi

https://textbookfull.com/product/computational-science-and-its-
applications-iccsa-2016-16th-international-conference-beijing-
china-july-4-7-2016-proceedings-part-iv-1st-edition-osvaldo-
gervasi/

Internet and Distributed Computing Systems 9th


International Conference IDCS 2016 Wuhan China
September 28 30 2016 Proceedings 1st Edition Wenfeng Li

https://textbookfull.com/product/internet-and-distributed-
computing-systems-9th-international-conference-idcs-2016-wuhan-
china-september-28-30-2016-proceedings-1st-edition-wenfeng-li/

Artificial Intelligence Methodology Systems and


Applications 17th International Conference AIMSA 2016
Varna Bulgaria September 7 10 2016 Proceedings 1st
Edition Christo Dichev
https://textbookfull.com/product/artificial-intelligence-
methodology-systems-and-applications-17th-international-
conference-aimsa-2016-varna-bulgaria-
Frank Hannig · João M.P. Cardoso
Thilo Pionteck · Dietmar Fey
Wolfgang Schröder-Preikschat
Jürgen Teich (Eds.)

Architecture of
LNCS 9637

Computing Systems –
ARCS 2016
29th International Conference
Nuremberg, Germany, April 4–7, 2016
Proceedings

123
Lecture Notes in Computer Science 9637
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
Frank Hannig João M.P. Cardoso

Thilo Pionteck Dietmar Fey


Wolfgang Schröder-Preikschat Jürgen Teich (Eds.)


Architecture of
Computing Systems –
ARCS 2016
29th International Conference
Nuremberg, Germany, April 4–7, 2016
Proceedings

123
Editors
Frank Hannig Dietmar Fey
Friedrich-Alexander University Friedrich-Alexander University
Erlangen-Nürnberg Erlangen-Nürnberg
Erlangen Erlangen
Germany Germany
João M.P. Cardoso Wolfgang Schröder-Preikschat
Faculty of Engineering (FEUP) Friedrich-Alexander University
University of Porto Erlangen-Nürnberg
Porto Erlangen
Portugal Germany
Thilo Pionteck Jürgen Teich
Universität zu Lübeck Friedrich-Alexander University
Lübeck Erlangen-Nürnberg
Germany Erlangen
Germany

ISSN 0302-9743 ISSN 1611-3349 (electronic)


Lecture Notes in Computer Science
ISBN 978-3-319-30694-0 ISBN 978-3-319-30695-7 (eBook)
DOI 10.1007/978-3-319-30695-7

Library of Congress Control Number: 2016932341

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 SpringerNature


The registered company is Springer International Publishing AG Switzerland
Preface

The 29th International Conference on Architecture of Computing Systems (ARCS


2016) was hosted by the Department of Computer Science at Friedrich-Alexander
University Erlangen-Nürnberg (FAU), Germany, during April 4–7, 2016. ARCS took
place in Nuremberg, a beautiful city where medieval vistas meet modernism and
technology. The conference continued the long-standing ARCS tradition of reporting
top-notch results in computer architecture and other related areas. ARCS was founded
in 1970 by the German computer pioneer Prof. Wolfgang Händler, who also founded
the Computer Science Department at FAU in 1966. It was a privilege to have brought
ARCS back to its roots in honor of the CS Department’s 50th anniversary. The con-
ference was organized by the Special Interest Group on Architecture of Computing
Systems of the GI (Gesellschaft für Informatik e. V.) and ITG (Informationstechnische
Gesellschaft im VDE), which held the financial responsibility for this ARCS edition.
The conference was also supported by IFIP (International Federation of Information
Processing).
“Heterogeneity in Architectures and Systems – From Embedded to HPC” was the
specific focus of ARCS 2016. This leitmotif reflected the ongoing progress in semi-
conductor technology that allows for building fascinating, complex computing systems,
including multiple (heterogeneous) microprocessors, large on-chip memory hierar-
chies, advanced interconnection networks, and peripherals. The downside to this
technological progress is that computing has already hit a power and complexity wall.
Thus, energy efficiency has become the key driver behind performance scaling across
all areas, from portable devices, such as smartphones and tablet PCs, to high-perfor-
mance computing (HPC) systems. This is why computing systems have begun to
include more and more heterogeneous hardware with various, specialized resources,
such as accelerators (e.g., GPUs or FPGAs) dedicated to one application domain.
However, designing and testing as well as the parallel programming of such hetero-
geneous computing systems are challenging tasks. Aside from energy efficiency, pre-
dictability, fault tolerance, accuracy, and security are often at least equally important
aspects when designing hardware and software. Thus, novel concepts as well as long-
reaching research in the areas of computer architecture design, computation models,
parallelization methods, software stacks, and programming and debugging tools are
required.
The ARCS 2016 program included an exciting collection of contributions resulting
from a successful call for papers. In response to the call for papers, 87 submissions
were received with affiliations to 31 countries. Each submission was subjected to
rigorous review from four Program Committee members. After having intensely
scrutinized the reviews, we were pleased to present a high-quality technical program
that included a total of 29 papers at the conference. The selected papers were divided
into thematic areas (nine sessions), which highlighted the, at the time, current focus of
research endeavors within computing systems. The sessions covered topics on
VI Preface

architecture design (accelerators, NoCs, caches, security components), approximate


and energy-efficient computing, organic computing systems, parallelization and map-
ping approaches, including various aspects on timing, performance modeling, and
reliability. The strong technical program was complemented by three keynote talks on:
“Knights Landing Intel Xeon Phi CPU: Path to Parallelism with General Purpose
Programming” by Avinash Sodani, Intel, USA; “Massive Parallelism – C++ and
OpenMP Parallel Programming Models of Today and Tomorrow” by Michael Wong,
IBM, Canada; and “Heterogeneous Systems Era” by John Glossner, Optimum Semi-
conductor Technologies, USA. The five workshops focusing on specific sub-topics
within ARCS were organized in conjunction with the main conference, one on parallel
systems and algorithms, one on dependability and fault tolerance, one on self-opti-
mization in autonomic and organic computing systems, one on multi-objective many-
core design, and one on improvements to safe, multi-core, and software-intensive
systems.
We would like to thank the many individuals who contributed to the success of the
conference, in particular the authors who responded to our call for papers, the members
of the Program Committee and the additional external reviewers who, with their
opinion and expertise, ensured a program of the highest quality. The workshops and
tutorials were coordinated by Ana Lucia Varbanescu, while Thilo Pionteck assisted in
organizing the proceedings. Many thanks to the team at FAU, which substantially
helped in the local arrangements. In particular, Ina Derr’s and Ina Hümmer’s organi-
zational talent was greatly appreciated and Dominik Schönwetter ensured that the Web
interactivity remained engaging and responsive. Thank you all.

April 2016 Frank Hannig


João M.P. Cardoso
Dietmar Fey
Wolfgang Schröder-Preikschat
Jürgen Teich
Organization

General Co-chairs
Dietmar Fey Friedrich-Alexander University Erlangen-Nürnberg,
Germany
Wolfgang Friedrich-Alexander University Erlangen-Nürnberg,
Schröder-Preikschat Germany
Jürgen Teich Friedrich-Alexander University Erlangen-Nürnberg,
Germany

Program Co-chairs
Frank Hannig Friedrich-Alexander University Erlangen-Nürnberg,
Germany
João M.P. Cardoso University of Porto, Portugal

Workshop and Tutorial Chair


Ana Lucia Varbanescu University of Amsterdam, The Netherlands

Publication Chair
Thilo Pionteck Universität zu Lübeck, Germany

Program Committee
Michael Beigl Karlsruhe Institute of Technology, Germany
Mladen Berekovic Technische Universität Braunschweig, Germany
Simon Bliudze École Polytechnique Fédérale de Lausanne,
Switzerland
Florian Brandner École Nationale Supérieure de Techniques Avancées,
France
Jürgen Brehm Leibniz Universität Hannover, Germany
Uwe Brinkschulte Universität Frankfurt, Germany
João M.P. Cardoso University of Porto, Portugal
Luigi Carro Universidade Federal do Rio Grande do Sul, Brasil
Albert Cohen Inria, France
Nikitas Dimopoulos University of Victoria, Canada
Ahmed El-Mahdy Egypt-Japan University for Science and Technology,
Egypt
Fabrizio Ferrandi Politecnico di Milano, Italy
VIII Organization

Dietmar Fey Friedrich-Alexander University Erlangen-Nürnberg,


Germany
William Fornaciari Politecnico di Milano, Italy
Björn Franke University of Edinburgh, UK
Roberto Giorgi Università di Siena, Italy
Daniel Gracia Perez Thales Research and Technology, France
Jan Haase University of the Federal Armed Forces Hamburg,
Germany
Jörg Hähner Universität Augsburg, Germany
Frank Hannig Friedrich-Alexander University Erlangen-Nürnberg,
Germany
Jörg Henkel Karlsruhe Institute of Technology, Germany
Andreas Herkersdorf Technische Universität München, Germany
Christian Hochberger Technische Universität Darmstadt, Germany
Michael Hübner Ruhr-University of Bochum, Germany
Gert Jervan Tallinn University of Technology, Estonia
Ben Juurlink Technische Universität Berlin, Germany
Wolfgang Karl Karlsruhe Institute of Technology, Germany
Christos Kartsaklis Oak Ridge National Laboratory, USA
Jörg Keller Fernuniversität in Hagen, Germany
Raimund Kirner University of Hertfordshire, UK
Andreas Koch Technische Universität Darmstadt, Germany
Hana Kubátová Czech Technical University in Prague, Czech Republic
Olaf Landsiedel Chalmers University of Technology, Sweden
Paul Lukowicz Universität Passau, Germany
Erik Maehle Universität zu Lübeck, Germany
Christian Müller-Schloer Leibniz Universität Hannover, Germany
Alex Orailoglu University of California, San Diego, USA
Carlos Eduardo Pereira Universidade Federal do Rio Grande do Sul, Brazil
Luis Pinho CISTER, ISEP, Portugal
Thilo Pionteck Universität zu Lübeck, Germany
Pascal Sainrat Université Toulouse III, France
Toshinori Sato Fukuoka University, Japan
Wolfgang Friedrich-Alexander University Erlangen-Nürnberg,
Schröder-Preikschat Germany
Martin Schulz Lawrence Livermore National Laboratory, USA
Leonel Sousa Universidade de Lisboa, Portugal
Rainer Spallek Technische Universität Dresden, Germany
Olaf Spinczyk Technische Universität Dortmund, Germany
Benno Stabernack Fraunhofer Institut für Nachrichtentechnik, Germany
Walter Stechele Technische Universität München, Germany
Djamshid Tavangarian Universität Rostock, Germany
Jürgen Teich Friedrich-Alexander University Erlangen-Nürnberg,
Germany
Martin Törngren KTH Royal Institute of Technology, Sweden
Eduardo Tovar CISTER/INESC TEC, ISEP, Portugal
Organization IX

Pedro Trancoso University of Cyprus, Cyprus


Carsten Trinitis Technische Universität München, Germany
Sascha Uhrig Airbus, Germany
Theo Ungerer Universität Augsburg, Germany
Hans Vandierendonck Queen’s University Belfast, UK
Stephane Vialle CentraleSupelec and UMI GT-CNRS 2958, France
Lucian Vintan Lucian Blaga University of Sibiu, Romania
Klaus Waldschmidt Universität Frankfurt am Main, Germany

Additional Reviewers

Akesson, Benny Hamza, Hamza


Al Nahas, Beshr Hassan, Ahmad
Becker, Thomas Hedrich, Lars
Behere, Sagar Heid, Kris
Benjamin, Betting Horsinka, Sven
Berezovskyi, Kostiantyn Hönig, Timo
Betting, Benjamin Iacovelli, Saverio
Bletsas, Konstantinos Izosimov, Viacheslav
Blochwitz, Christopher Jean, Xavier
Bromberger, Michael Jooya, Ali
Buschhoff, Markus Joseph, Moritz
Cebrián Márquez, Gabriel Kantert, Jan
Chalios, Charalampos Karavadara, Nilesh
Chen, Dejiu Karray, Hassen
Chi, Chi Ching Keshavarz, Babak
Cosenza, Biagio Khoshbakht, Saman
Damschen, Marvin Kicherer, Mario
de Souza, Diego F. Klaus, Tobias
Eckert, Marcel Kuznetsov, Max
Eibel, Christopher Lal, Sohan
Falk, Joachim Le Rhun, Jimmy
Feng, Lei Lucas, Jan
Franzmann, Florian Lund, Andreas
Freitag, Johannes Martins, Paulo
Frieb, Martin Maurer, Simon
Gabriel, Dirk Meumeu Yomsi, Patrick
Gante, João Meyer, Dominik
Georgakoudis, Giorgis Mische, Jörg
Girbal, Sylvain Mohanty, Jaganath
Goebel, Matthias Mukhanov, Lev
Guerreiro, João F.D. Naghmouchi, Jamin
Gurdur, Didem Naji, Amine
Gómez Luna, Juan Nelis, Vincent
Haas, Florian Nelissen, Geoffrey
X Organization

Neves, Nuno Shuka, Romeo


Niazmand, Behrad Siegl, Patrick
Niemann, Sebastian Sreekar Shenoy, Govind
Nikolic, Borislav Stegmeier, Alexander
Pathania, Anuj Ulbrich, Peter
Payandeh Azad, Siavoosh Versick, Daniel
Penschuck, Manuel Wagner, Jan
Preußer, Thomas Weidendorfer, Josef
Rao, Srinivas Weis, Sebastian
Rashid, Syed Aftab Wenzel, Volker
Reiche, Oliver Westman, Jonas
Rochange, Christine Wildermann, Stefan
Ruschke, Tajas Wolf, Dennis
Sabel, David Zabel, Martin
Samie, Farzad Zaichenkov, Pavel
Schirmeier, Horst Zhang, Hongyan
Invited Talks
Avinash Sodani, Chief Architect ‘Knights
Landing’ Xeon-Phi Processor
at Intel Corporation

Knights Landing Intel Xeon Phi CPU: Path


to Parallelism with General Purpose Programming

Abstract of Talk: The demand for high performance will continue to


skyrocket in the future fueled by the drive to solve the challenging
problems in science and to provide the horsepower needed to support
the compute-hungry use cases that are emerging in consumer and
commercial space. Exploiting parallelism will be crucial in achieving
the massive amounts of performance required to solve these problems.
This talk will present the new Xeon Phi Processor, called Knights
landing, which is architected to provide huge amounts of parallelism in
a manner that is accessible with general purpose programming. The talk
will provide insight into the important architecture features of the pro-
cessor and explain their implications on software. It will provide the
inside story on the various architecture decisions made on Knights
Landing - why we architected the processor the way we did. It will show
measured performance numbers from the Knights Landing silicon on
range of workloads. The talk will conclude with showing the historical
trends in architecture and what they mean for software as we extend the
trends into the future.

Avinash Sodani is a Senior Principal Engineer at Intel Corporation and


the chief architect of the Xeon-Phi Processor called Knights Landing.
He specializes in the field of High Performance Computing (HPC).
Previously, he was one of the architects of the 1st generation Core
processor, called Nehalem, which has served as a foundation for today’s
line of Intel Core processors. Avinash is a recognized expert in com-
puter architecture and has been invited to deliver several keynotes and
public talks on topics related to HPC and future of computing. Avinash
holds over 20 US Patents and is known for seminal work on the concept
of “Dynamic Instruction Reuse”. He has a PhD and MS in Computer
Science from University of Wisconsin-Madison and a B.Tech (Hon’s)
in Computer Science from Indian Institute of Technology, Kharagpur in
India.
Michael Wong, IBM Corporation, XL C++
Compiler Kernel Development

Massive Parallelism - C++ and OpenMP Parallel


Programming Models of Today and Tomorrow

Abstract of Talk: Why is the world rushing to add Parallelism to base


languages when consortiums and companies have been trying to fill that
space for years? How is the landscape of Parallelism changing in the
various standards, and specifications? I will give an overview as well as
a deep dive into what C/C++ is doing to add parallelism such as the
proposaed Vector/SIMD model that employs implicit Wavefront, but
also how consortiums like OpenMP is pushing forward into the world’s
first High-level Language support for GPGPU/Accelerators and SIMD
programming. Both are necessary to express the Massive Parallelism of
tomorrow. GPU/Accelerator computing looks to be here to stay. Whe-
ther it is a flash in the pan or not, data parallelism is never going to be
stale, especially for high-performance computing. The good news is that
Clang 3.7 has OpenMP 3.1 support through a collaborative effort
between many institutions, and Clang 3.8 or later will have some form
of support for OpenMP 4 accelerators that targets many different
devices, including Intel Xeon Phi and NVIDIA GPUs. Other compilers
with high-level accelerator support will be GCC 6. The OpenMP model
was designed to fit all possible forms of accelerators. However, in a
changing world where discrete GPUs are transitioning into accelerated
processing units (APUs), and being combined with various forms of
high-bandwidth memory (HBM), is the OpenMP model, or even the
OpenACC model, the best model? Should we begin programming
future devices with a new programming model that ignores the “legacy”
discrete systems and aims for the future? I’ll contrast the OpenMP
design with the emerging C++ Standard design and how we can merge
the design experience from both HPC and consumer devices. As Chair
of C++ Standard’s SG14: Games Dev and Low Latency, I will show
where we might be going in five years with an accelerator model that is
appropriate for the future with description of Agency, SYCL, HCC, and
HPX based on an outline of future goals.

Michael Wong is the CEO of OpenMP. He is the IBM and Canadian


representative to the C++ Standard and OpenMP Committee. He is also
a Director of ISOCPP.org and a VP, Vice-Chair of Programming
Michael Wong, IBM Corporation, XL C++ Compiler Kernel Development XV

Languages for Canada’s Standard Council. He has so many titles, it’s a


wonder he can get anything done. He chairs the WG21 SG5 Transac-
tional Memory and SG14 Games Development/Low Latency, and is the
co-author of a number C++11/ OpenMP/ Transactional Memory fea-
tures including generalized attributes, user-defined literals, inheriting
constructors, weakly ordered memory models, and explicit conversion
operators. Having been the past C++ team lead to IBM’s XL C++
compiler means he has been messing around with designing C++
compilers for twenty years. His current research interest, i.e. what he
would like to do if he had time is in the area of parallel programming,
transactional memory, C++ benchmark performance, object model,
generic programming and template metaprogramming. He holds a B.Sc
from University of Toronto, and a Masters in Mathematics from
University of Waterloo. He has been asked to speak at ACCU,
C++Now, Meeting C++, ADC++, CASCON, Bloomberg, CERN, and
many Universities, research centers and companies.
John Glossner, President of HSAF and CEO
of Optimum Semiconductor Technologies

Heterogeneous Systems Era

Abstract of Talk: Heterogeneous processing represents the future of


computing, promising to unlock the performance and power efficiency
of parallel computing engines found in most modern electronic
devices. This talk will detail the HSA Foundation (HSAF) computing
platform infrastructure including features/advantages across comput-
ing platforms from mobile and tablets to desktops to HPC and servers.
The talk will focus on technical issues solved by HSAF technologies.
The presentation will also discuss important new developments that
are bringing the industry closer to broad adoption of heterogeneous
computing.

Dr. John Glossner is President of the Heterogeneous System


Architecture Foundation (HSAF) and CEO of Optimum Semicon-
ductor Technologies. OST and its processor division General Pro-
cessor Technologies (GPT-US). He is also a professor of Computer
Science at Daniel Webster College. Previously he served as Chair
of the Board of the Wireless Innovation Forum. In 2010 he joined
Wuxi DSP (a licensee of Sandbridge technology and parent company
of OST) and was named a China 1000 Talents. He previously
co-founded Sandbridge Technologies and received a World Economic
Forum award. Prior to Sandbridge, John managed both technical and
business activities in IBM and Lucent/Starcore. John received a Ph.D.
in Electrical Engineering from TU Delft in the Netherlands, M.S
degrees in E.E. and Eng. Mgt from NTU, and a B.S.E.E. degree from
Penn State. He has more than 40 patents and 120 publications.
Contents

Configurable and In-Memory Accelerators

Towards Multicore Performance with Configurable Computing Units . . . . . . 3


Anita Tino and Kaamran Raahemifar

Design and Evaluation of a Processing-in-Memory Architecture


for the Smart Memory Cube. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
Erfan Azarkhish, Davide Rossi, Igor Loi, and Luca Benini

Network-on-Chip and Secure Computing Architectures

CASCADE: Congestion Aware Switchable Cycle Adaptive


Deflection Router . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
Gnaneswara Rao Jonna, Vamana Murthi Thuniki, and Madhu Mutyam

An Alternating Transmission Scheme for Deflection Routing


Based Network-on-Chips . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
Armin Runge and Reiner Kolla

Exzess: Hardware-Based RAM Encryption Against Physical


Memory Disclosure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
Alexander Würstlein, Michael Gernoth, Johannes Götzfried,
and Tilo Müller

Hardware-Assisted Context Management for Accelerator Virtualization:


A Case Study with RSA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
Ying Gao and Timothy Sherwood

Cache Architectures and Protocols

Adaptive Cache Structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87


Carsten Tradowsky, Enrique Cordero, Christoph Orsinger,
Malte Vesper, and Jürgen Becker

Optimization of a Linked Cache Coherence Protocol for Scalable Manycore


Coherence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
Ricardo Fernández-Pascual, Alberto Ros, and Manuel E. Acacio
XVIII Contents

Mapping of Applications on Heterogeneous Architectures


and Real-Time Tasks on Multiprocessors

Generic Algorithmic Scheme for 2D Stencil Applications


on Hybrid Machines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
Stephane Vialle, Sylvain Contassot-Vivier, and Patrick Mercier

GPU-Accelerated BWA-MEM Genomic Mapping Algorithm


Using Adaptive Load Balancing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
Ernst Joachim Houtgast, Vlad-Mihai Sima, Koen Bertels,
and Zaid Al-Ars

Task Variants with Different Scratchpad Memory Consumption


in Multi-Task Environments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143
Martin Böhnert and Christoph Scholl

Feedback-Based Admission Control for Hard Real-Time Task Allocation


Under Dynamic Workload on Many-Core Systems . . . . . . . . . . . . . . . . . . . 157
Piotr Dziurzanski, Amit Kumar Singh, and Leandro Soares Indrusiak

All About Time: Timing, Tracing, and Performance Modeling

Data Age Diminution in the Logical Execution Time Model. . . . . . . . . . . . . 173


Christian Bradatsch, Florian Kluge, and Theo Ungerer

Accurate Sample Time Reconstruction for Sensor Data Synchronization . . . . 185


Sebastian Stieber, Rainer Dorsch, and Christian Haubelt

DiaSys: On-Chip Trace Analysis for Multi-processor System-on-Chip . . . . . . 197


Philipp Wagner, Thomas Wild, and Andreas Herkersdorf

Analysis of Intel’s Haswell Microarchitecture Using the ECM Model


and Microbenchmarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 210
Johannes Hofmann, Dietmar Fey, Jan Eitzinger, Georg Hager,
and Gerhard Wellein

Measurement-Based Probabilistic Timing Analysis for Graphics


Processor Units. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223
Kostiantyn Berezovskyi, Fabrice Guet, Luca Santinelli,
Konstantinos Bletsas, and Eduardo Tovar

Approximate and Energy-Efficient Computing

Reducing Energy Consumption of Data Transfers Using Runtime Data


Type Conversion. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 239
Michael Bromberger, Vincent Heuveline, and Wolfgang Karl
Contents XIX

Balancing High-Performance Parallelization and Accuracy in Canny


Edge Detector. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 251
Valery Kritchallo, Billy Braithwaite, Erik Vermij, Koen Bertels,
and Zaid Al-Ars

Analysis and Exploitation of CTU-Level Parallelism in the HEVC Mode


Decision Process Using Actor-Based Modeling . . . . . . . . . . . . . . . . . . . . . . 263
Rafael Rosales, Christian Herglotz, Michael Glaß, André Kaup,
and Jürgen Teich

Low-Cost Hardware Infrastructure for Runtime Thread Level


Energy Accounting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 277
Marius Marcu, Oana Boncalo, Madalin Ghenea, Alexandru Amaricai,
Jan Weinstock, Rainer Leupers, Zheng Wang, Giorgis Georgakoudis,
Dimitrios S. Nikolopoulos, Cosmin Cernazanu-Glavan, Lucian Bara,
and Marian Ionascu

Allocation: From Memories to FPGA Hardware Modules

Reducing NoC and Memory Contention for Manycores . . . . . . . . . . . . . . . . 293


Vishwanathan Chandru and Frank Mueller

An Efficient Data Structure for Dynamic Two-Dimensional


Reconfiguration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 306
Sándor P. Fekete, Jan-Marc Reinhardt, and Christian Scheffer

Organic Computing Systems

Runtime Clustering of Similarly Behaving Agents in Open Organic


Computing Systems. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 321
Jan Kantert, Richard Scharrer, Sven Tomforde, Sarah Edenhofer,
and Christian Müller-Schloer

Comparison of Dependency Measures for the Detection of Mutual


Influences in Organic Computing Systems . . . . . . . . . . . . . . . . . . . . . . . . . 334
Stefan Rudolph, Rainer Hihn, Sven Tomforde, and Jörg Hähner

Augmenting the Algorithmic Structure of XCS by Means of Interpolation . . . 348


Anthony Stein, Dominik Rauh, Sven Tomforde, and Jörg Hähner

Reliability Aspects in NoCs, Caches, and GPUs

Estimation of End-to-End Packet Error Rates for NoC Multicasts . . . . . . . . . 363


Michael Vonbun, Thomas Wild, and Andreas Herkersdorf
XX Contents

Protecting Code Regions on Asymmetrically Reliable Caches . . . . . . . . . . . . 375


Sanem Arslan, Haluk Rahmi Topcuoglu, Mahmut Taylan Kandemir,
and Oguz Tosun

A New Simulation-Based Fault Injection Approach for the Evaluation


of Transient Errors in GPGPUs. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 388
Sarah Azimi, Boyang Du, and Luca Sterpone

Author Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 401


Configurable and In-Memory
Accelerators
Towards Multicore Performance
with Configurable Computing Units

Anita Tino(&) and Kaamran Raahemifar

Department of Electrical and Computer Engineering,


Ryerson University, 350 Victoria St, Toronto, ON, Canada
a2tino@ee.ryerson.ca

Abstract. Energy efficiency and the need for high performance has steered
computing platforms toward customization. General purpose computing how-
ever remains a challenge as on-chip resources continue to increase with a limited
performance improvement. In order to truly improve processor performance, a
major reconsideration at the microarchitectural level must be sought with
regards to the compiler, ISA, and general architecture without an explicit
dependence on transistor scaling and increased cache levels. In attempts to
assign the processor transistor budget towards engineering ingenuity, this paper
presents the concept of Configurable Computing Units (CCUs). CCUs are
designed to make reconfigurability in general purpose computing a reality by
introducing the concept of logical and physical compilation. This concept allows
for both the application and underlying architecture to be considered during the
compilation process. Experimental results demonstrate that a single CCU core
(consisting of double engines) achieves dual core performance, with half the
area and power consumption required of a conventional monolithic CPU.

1 Introduction

Traditional microprocessors have long benefited from the transistor density gains of
Moore’s law. Diminishing transistor speeds and practical energy limits however have
created new challenges in technology, where the exponential performance improve-
ments we have been accustomed to from previous computing generations continue to
slowly cease. A common response to addressing various issues in computing has
revolved around increasing core counts in multiprocessors and employing hetero-
geneity (i.e. including accelerated units such as GPUs, FPGAs etc.) for offloading and
executing dataflow-like phases of an application. This type of computing has been
made possible with parallel programming models and languages such as OpenCL,
OpenMP, CUDA, and OmpSs [2, 7]. Much research in Chip Multiprocessors
(CMP) has also revolved around energy efficiency and specialization, noting that
processor performance improvement is attributed to increased cache levels and tran-
sistor scaling [1, 3]. Although these concepts have increased performance to an extent,
there is still the fundamental problem of how a single core’s organization and design
may be improved and applied to the multiprocessor domain [4, 5, 8–10].
Conventional CPU architectures possess a boundary between the compiler and
underlying hardware due to a processor’s Instruction Set Architecture (ISA). Current

© Springer International Publishing Switzerland 2016


F. Hannig et al. (Eds.): ARCS 2016, LNCS 9637, pp. 3–18, 2016.
DOI: 10.1007/978-3-319-30695-7_1
4 A. Tino and K. Raahemifar

ISAs are limited by program counters, control-flow, and fine-grained instruction exe-
cution, lacking the richness needed to express a programmer’s intent. Specifically, the
programmer and software may provide a considerable amount of information about an
application prior to compilation, however ISAs are unable to express this information
during code generation and consequently the compiler is not able to exploit this
knowledge to the underlying CPU [10]. Majority of the processor’s hardware units
must then try to rediscover the program’s characteristics at the cost of power and area
overhead. Hence, there is much ingenuity to be discovered based on these limitations if
transistor budgets are put towards improving and sophisticating processing element
performance, versus adopting the mantra of integrating more simple cores and memory
on a single die [5].
Given these factors, this work presents a nuanced approach to general purpose
computing using the concepts of Configurable Computing Units (CCUs), and logical
and physical compilation. CCUs are configurable processors that execute tasks which
are formed using the OmpSs programming model (Sect. 3.1). Each CCU processor
consists of multiple variable sized engines, where each engine comprises of unique
functional units (FUs) that are connected through a registerSwitch (rS) interconnect as
shown in Fig. 1. The rS interconnect provides distributed storage and single-cycle
multi-hop data communication amongst its FUs. Thus the rS interconnect avoids the
need to constantly access centralized register files, bypass networks, and tile-based
hotspots by revising the datapath layout. The engines are able to configure to a general
purpose application’s communication patterns using the programmable rS interconnect
and information of the underlying architecture, transferring data amongst the functional
units only when necessary. Therefore the configuration data (generated by the physical
compiler) allows the engine to temporally configure itself on every clock cycle to
support various data transfers and storages as required. To maintain memory ordering
imposed by the logical compiler, a small external register file and load/store unit(s) are
also included in the CCU back-end. A banked memory setup for storing configuration
data is also used to reduce configuration times and allow for reconfigurability in general
purpose processors.
The focus of this work pertains to the CCU back-end which will be prototyped in
hardware, with the full CCU simulated in software and compared to a conventional
monolithic CPU and multiprocessor.

2 Overview

2.1 Design Flow and CCU Microarchitecture


Figure 2 provides the CCU design flow implemented in this work. In the first stage of
the design flow, a benchmark is input to the logical compiler which generates a binary
and sends it to the physical compiler (and software simulator for testing). General
profiling information of the tasks is then obtained in the physical compiler using the
provided binary and embedded OmpSs pragmas. Exploiting the underlying processor
information, the physical compiler then (1) generates tasks, (2) analyzes the instruc-
tions within the tasks to obtain statistical data and eliminate Write-After-Write
Towards Multicore Performance with Configurable Computing Units 5

Fig. 1. Six FU (Single) engine CCU Fig. 2. Proposed CCU processor design flow
architecture

(WAW) hazards, which is then used to (3) extract producer-consumer dependencies,


(4) assign tasks to the most suitable engines, and finally (5) generate configuration and
task data to be used by the CCU engines for configuration and execution of task
instructions. Thus once successfully compiled, the generated configuration data is
stored to its respective engine’s (banked) configuration memory. Similarly, the
addresses used to store the configuration data are saved to a lookup table based on the
task’s ID (Fig. 2).
As execution in a CCU is task based and the “instructions” to be executed are
essentially the configuration bits generated by the physical compiler, the scheduler
must then simply keep track of task related dependencies for dispatch/execution. As
seen in the microarchitectural pipeline provided in Fig. 3, when a task is ready, the
scheduler outputs the corresponding task’s ID to the engine designated by the physical
compiler. Note that the scheduler may keep track of two engines concurrently. For this
work, the Extra instrumentation tool [6] is used by the software simulator’s scheduler
to obtain traces and determine ready tasks and dependencies to reduce design
complexity.

Fig. 3. CCU microarchitectural pipeline


6 A. Tino and K. Raahemifar

Assuming the engine is also ready, the task ID is then input to the engine’s lookup
table to determine the configuration memory addresses to be read (where the necessary
configuration bits are stored contiguously from that address). This address is then read
from memory which initiates the engine’s configuration and external read process
(where external data is sent to the Read Register Buffer (RRB)). Once configuration is
complete, task execution commences. Subsequent to execution, final values pending in
the Write Register Buffer (WRB) are written/stored back to the external register
file/cache, respectively.

2.2 Backend Execution Example


An example datapath execution flow for a six FU engine is presented in Fig. 4. The task
instructions to be executed are provided to the right of Fig. 4, where the circled
numbers signify an instruction’s execution time. Instructions are displayed as: <oper-
ation, result, operand1, operand2>. The engine’s “path” of execution for the listed
instructions is provided on the engine’s datapath. Note that this path is temporally and
spatially configured with the configuration data provided by the physical compiler. The
various dashed lines used in the figure represent execution in different cycles.

Fig. 4. CCU execution example (torus connection not shown)

In the first cycle of the example, the XOR and XNOR instructions are executed (as
buffered in their respective FU queue), each obtaining their source data from the RRB
(i.e. the external register reads and/or immediate values buffered to the RRB during
configuration). The result generated by the XNOR instruction is thereafter needed by
the three consumers ADD, SUB, and OR, all executing in the second cycle. The XOR
instruction however is needed in the next clock cycle by the ADD and SUB instruc-
tions, and in the third cycle by the AND instruction. The operand result (‘4’) therefore
requires both propagation and temporary storage for its consumers in the 2nd and 3rd
Towards Multicore Performance with Configurable Computing Units 7

clock cycle respectively. Therefore during the 3rd clock cycle when AND executes, the
operand is selected from the rS unit storage (marked ‘S’ in the 2nd row, 2nd column),
and propagated to its consumer as programmed during configuration.
This process continues for all instructions until completion, where each FU con-
tains a queue to buffer multiple task instructions. In terms of distributed storage, the rS
units may temporally store up to 3 values each (i.e. 3 input ports) where the torus
connection is used to transfer results from opposite edges of the rS interconnect (right
to left, and bottom to top) for a continuing stream of dataflow execution using the xy
protocol. Note that stores are also sent to the WRB during execution, which are then
forwarded to the store unit to handle writebacks. Once all instructions have finished
executing, the final register values present in the WRB are written to the external
register file as required.

3 Compilers
3.1 OmpSs Programming Model
OmpSs is a task-based programming model which exploits task-level parallelism on a
shared-memory architecture. OmpSs enables programmers to annotate standard
sequential-based applications with pragmas to help the compiler and runtime identify
dependent and independent tasks for parallel execution while maintaining program-
ming familiarity. Therefore the programmer does not need to synchronize or manage
task execution, but rather expose task side-effects by simply annotating each kernel’s
operands with input, output, or inout (bidirectional) [7]. Using the clauses specified by
the programmer, communication patterns amongst the tasks are obtained and analyzed
by the physical compiler to determine inter-dependencies.

3.2 Logical Compiler


A logical compiler is a standard compiler that may be used by any modern computing
system. Although any compiler may be chosen for a CCU, the logical (and physical)
compiler will likely benefit from a more RISC-like target architecture as more
Instruction Level Parallelism (ILP) may be extracted based on the structure of the ISA’s
layout, i.e. it is generally easier to eliminate false data dependencies, analyze register
usage etc. in RISC. Conversely, x86 (a more CISC-like ISA) relies on an
accumulator-based register architecture making it slightly more difficult for the physical
compiler to extract parallelism from a fairly sequential instruction stream.
For this work, the ARMv7 ISA was selected for both the physical and logical
compiler target architecture, with other ISA plugins remaining future work. To
maintain compatibility with the OmpSs model, the Mecurium compiler (gcc-based)
was also embedded into the logical compiler. Therefore the logical compiler may
compile any C, C++, and Fortran language with either a sequential, OpenMP, OmpSs
and/or MPI based programming model.
8 A. Tino and K. Raahemifar

3.3 Physical Compiler


A physical compiler is the joint effort between compiler designers and computer
architects, used for a smarter compilation process. The overall goal of the CCU’s
physical compiler is to form and map tasks according to specifications of the under-
lying architecture for higher performance and energy efficiency. Therefore the physical
compiler is responsible for forming tasks by decoding instructions, eliminating false
dependencies, and considering task dependencies and their effect on the underlying
architecture.
As standard with majority of task-based compilers, the first stage of physical
compilation requires tasks to be formed (and respective instructions gathered) based on
the binary and task information gathered with the OmpSs model. The physical compiler
then packs task instructions until a conditional branch is reached, eliminating all
unconditional branches while maintaining program correctness (i.e. following the
unconditional flow of control). Architecturally, this allows each engine’s branch unit to
possess low hardware complexity, and for a substantial amount of instructions to be
present per task. The physical compiler then assesses dependencies within the intra-task
instructions, eliminating false dependencies, extracting ILP, and creating a
producer-consumer dependency matrix for every task.
To appropriately assign tasks to engines, the task’s respective instructions and
inter-dependencies are analyzed and compared to the available (and ideal) CCU
engines for execution. The physical compiler invokes a demand-type algorithm to map
tasks to an engine - i.e. if a task contains a high instruction count and ILP, the algorithm
will give preference to assigning the task to a larger engine in the CCU that can exploit
the ILP and not overload the rS and FU buffers. Similarly, if an instruction requires a
specialized FU, the task will be mapped to the engine which possesses the FU. Thus
tasks are mapped in this way while following inter-dependencies. Note that large tasks
which do not fit into the FU buffers are divided into smaller tasks and assigned
accordingly.
Once the tasks have been mapped to an appropriate engine, the internal instructions
are assessed and mapped to FUs according to the instruction dependency matrix
generated and/or any functional unit dependencies. The technique employed by the
physical compiler is to place independent instructions side-by-side in the x direction
(spatially adjacent FUs), and dependent instructions in the y direction (FUs in the
following engine row(s)) so that the results may traverse with minimal hops. If an
adjacent FU is unavailable, the compiler will then search for the closest available FU to
its dependency.
Once the instructions have been successfully mapped, the physical compiler then
evaluates the mappings to determine how each rS unit should be temporally and
spatially configured for data propagation and/or storage of dependent instructions.
Once the rS have been successfully mapped, the configuration data needed by the CCU
engine is generated for each rS, FU (instruction opcodes), RRB, and WRB. Each task’s
configuration data is then stored in its respective memory bank located in the processor
back-end, with each task’s configuration address stored to the engine’s lookup table.
It is important to note that load (and store) instructions within tasks are extracted
prior to the rS mapping stage as these instructions must execute on an externally
Towards Multicore Performance with Configurable Computing Units 9

dedicated unit (See Fig. 1). Loads may be divided into two categories: External (a load
which is needed to start task execution) or internal (load requested by an inter-task
instruction). External loads possess a separate configuration memory bank which is
used to read data values during configuration, occurring concurrently while the other
units are configured. These loaded values are then written to the RRB so that they may
propagate to their consumers as required during the execution stage. Similarly, internal
loads are configured in the WRB which must wait for their producer data to generate in
the engine. Once ready, the address is then sent to the load unit, where the data is
brought back to the RRB when successfully read. Store instructions are also buffered
by the WRB during task execution and sent to the store unit when ready. Likewise,
external ‘‘register’’ values to be read prior and/or written after task execution are
configured in the RRB and WRB memory respectively (in the same manner as loads
and stores). Note that the multiplier/divider unit does not need to be configured as the
RRB and WRB units are configured in the same manner as loads.

4 Hardware Design

4.1 rS and Interconnect


The two types of units present in an engine’s interconnect are referred to as external
and internal rS. As seen in Fig. 1, external rS are found at the very top of the
interconnect and consist of four ports - two input and two output. These rS act as an
interface between the RRB and functional unit inputs, and connect the bottom rS units
to the top of the torus topology. Therefore external switches do not require storage
properties since they are directly connected to the RRB (which buffer data) and do not
receive any FU output. Horizontal propagation is also not supported by the external
switches due to the xy routing protocol. All these factors allow external switches to
possess low hardware area and a reduced number of I/O ports.
Internal switches are present in the main communication network and consist of six
ports- three inputs and three outputs. An example of an internal rS architecture is
presented in Fig. 5. All internal rS input ports consist of a 2:1 multiplexer that may
(1) propagate input data, (2) output a stored value, or (3) simultaneously store and
propagate a value. Control bits for all multiplexers and registers are dictated by the
configuration data generated by the physical compiler. As shown in Fig. 5, an output
port multiplexes all input ports, and therefore all rS output ports may propagate data
simultaneously to neighbouring rS. Conflicts involved in data traversal of results are
minimal. However if conflicts do exist, they are resolved by the physical compiler
which either reroutes the result, uses alternative storage locations, or in the worst case,
delays execution by one clock cycle.
The rS interconnect was designed to support single cycle multi-hop traversal,
signifying that any result may be sent to any functional unit source input (including its
own) within one clock cycle. To support this feature, the frequency of operation is
limited by an engine’s critical path. As each engine within a CCU varies in topological
size, its respective maximum frequency will also vary. As will be seen during exper-
imental testing however, engine size accounts for slight variation in the CCU’s
10 A. Tino and K. Raahemifar

Fig. 5. Internal registerswitch (rs) architecture

maximum achievable frequency due to the simplistic design of the rS (Sect. 5.2.5,
Frequency of Operation).

4.2 Functional Units


FUs present within an engine may be of any type including (but not limited to) ALUs,
Multiplier/dividers, Multiplier-Accumulator Units (MAC), ALU/barrel shifter combi-
nations etc. Each functional unit in an engine contains a buffer to hold its assigned task
instruction opcodes. A low complexity conditional branch unit is also included in every
engine. To support conditional branch execution, an engine contains a flag status
register which is updated as instructions execute. External to each engine is a dedicated
load unit(s) (see Fig. 1). A load takes approximately four clock cycles to execute, as
standard in majority of conventional processors. Since multiply/divide instructions
occur infrequently, one multiplier/divide unit is typically shared by the engines. Similar
to the load unit, the operand data for mult/div is sent by the WRB, and the
product/quotient is received by the RRB.

4.3 Read Register Buffer (RRB)


The RRB is responsible for managing (inputted) external data and communicating the
data to the engine’s components as needed during execution. Specifically, RRBs
manage:
• External register file reads and loads during configuration and communicate these
values to the required rS during execution.
• Internal loads (requested by the WRB) and communicate these loaded values to the
required rS.
• The communication of immediate values to the required rS.
• Incoming external (i.e. multiply/divide) results and communicate the values to the
correct rS.
Another random document with
no related content on Scribd:
on merkinnyt: »Entwurf einer reinen Philosophie», mutta
valitettavasti tuo vihko on jäänyt — aivan tyhjäksi. Irtonaisista
paperipalasista, joihin hän on merkinnyt ajatuksen juoksuaan ja
joista — niin paljon kuin niitä on säilynytkin — suuri osa lienee
joutunut hukkaan, ei saa mitään kokonaiskäsitystä, ne ovat
aforismeja ja niiden abstraktisuudesta jää lukijalle vahva
»qvasifilosofian»[26] sivumaku. On kuitenkin myönnettävä, että
näiden ajatussirpaleiden pohjana on persoonallisia elämyksiä.
Puhtaus tuntuu olevan hänen filosofiansa ydinkäsitteitä; se ja koko
tuon filosofeeraasimisen abstraktinen luonne löytänevät selvityksen
muualtakin kuin kirjaviisaudesta: tuntuu luonnolliselta, että
aristokraattinen äiti oli lapsuudesta asti varjellut poikaansa
joutumasta liian läheiseen kosketukseen pahan maailman kanssa.
Mutta vielä useammin kuin sana puhtaus esiintyy hänen
kirjoituksissaan sana vapaus. Avioliiton alkuvuosina se sana on
ytimenä kaikissa niissä purkauksissa, joissa hän valittaa
käytännöllisen elämän asettamia kahleita ja huokaa rahallisen
raadannan alla: »Minä en anna pakottaa itseäni, tahdon päästä
käytännöllisen elämän pakkovallasta, tähden elää mielipiteilleni, tuo
käytännöllinen elämä» (Praxis) iljettää minua puolinaisuudellaan»,
hän tavan takaa ja useissa eri muodoissa toistaa.

Menemällä avioliittoon Ida Aalbergin kanssa vapaaherra Uexküll-


Gyllenband oli, niinkuin jo etukäteen oli aavistanut, joutunut
käytännöllisen elämän palvelukseen. Että hän tunsi tehneensä
suuren uhrauksen, selviää esim. seuraavista hänen kirjoittamistaan
riveistä:

»Olen siis näyttänyt, mitä paljaalla ajattelemisella voi saavuttaa.


Sillä minulle ei anneta keinoja filosofian harrastamiseen. Ennen
kaikkea ei anneta joutoaikaa. Kun isoäitini lahjoitti minulle 10,000
markkaa, käytin siitä puolet opintoihin. Sitten olen mennyt naimisiin
ja olen käyttänyt toisen puolen tuosta summasta avioliittoni hyväksi;
siitä lähtien ovat avioliittoni edut, joskaan eivät ole olleet hallitsevina,
olleet ainakin yhtä tärkeitä kuin omat filosofiset harrastukseni.» —

Ida Aalberg ei osoittanut miehensä filosofisille harrastuksille


ymmärtämystä, sen on vapaaherra Uexküll-Gyllenband elämänsä
lopulla suurella haikeudella todennut. Päiväkirjassa vuodelta 1897
on pari alakuloista merkintää, jotka todistavat, että tässä suhteessa
aviopuolisoiden välit eivät vastanneet vapaaherran toivomuksia.
Vihon kanteen on Alexander Uexküll-Gyllenband kirjoittanut:

»Tästä on Ida ilman minun lupaani ja puhumatta siitä minulle


mitään repinyt pois sivun.»

Itse vihon sisällä on m.m. seuraava kuvaus, joka osoittaa, että


vapaaherra avioliittonsa vuoksi oli jo alkanut epäillä omaa
kutsumustaan:

»Kun olen levoton, kuljeskelen ympäri — löytämättä rauhaa


istumiseen, syömiseen, nukkumiseen. Saan rauhaa vain, kun
ajattelen. — En saa rauhaa edes lukemiseen; en edes
sanomalehden lukemiseen. Ja kun luen filosofista tekstiä, rupean
heti ajattelemaan enkä pääse alkua pitemmälle. — Ja kaiken, mitä
ajattelen, tahtoisin saada kootuksi ja kirjoitan sen muistiin; olen
kirjoittanut koreja täyteen — ilman, että niillä on mitään arvoa.»

Niin herkkä ja helposti ärtyvä hermosto kuin vapaaherra Uexküll-


Gyllenbandilla olikin, ei hänen hermostumisensa ollut tavallisten
jokapäiväisten ihmisten hermostumista. Voidaan pitää
kysymyksenalaisena, oliko hän nero — sen kysymyksen
perinpohjainen ratkaiseminen vie tutkijalta vuosikymmeniä, siksi
tuottelias oli vapaaherra Uexküll-Gyllenband — mutta varmaa on,
että hänellä oli nerojen yleinen tauti: hän oli nuoruudestaan asti
maanikko. — Hänen tarkoitusperänsä kyllä muuttuivat, mutta kun
hän sai jonkin aatteen, joka häntä innostutti, muuttui se hänelle
todelliseksi elämänkysymykseksi, hän ei nähnyt eikä tahtonut nähdä
mitään muuta, hän ei puhunut eikä tahtonut puhuttavan mistään
muusta, hän toimi hellittämättömältä tuntuvalla tarmolla sen hyväksi.

Vapaaherra Alexander Uexküll-Gyllenband oli gentleman. »Hän ei


valehtele koskaan, ei edes leikillään», kertoi Ida Aalberg elämänsä
loppuaikoina miehestään. Ellei vapaaherra Uexküll-Gyllenband olisi
ollut niin kiihkeä luonne, ellei hän olisi ollut siinä määrin
hermoileminen kuin oli ja ellei hänellä olisi ollut avioliitosta niin
korkeata käsitystä sielujen, ajatusten ja tunteitten yhdistäjänä kuin
hänellä oli, Ida Aalberg tuskin olisi saanut myöskään tietää, että
heidän yhdyselämällään oli omat traagilliset puolensa. Ida Aalberg
sai sen tietää, ja esim. seuraavasta vapaaherran puolisolleen
osoittamasta kirjallisesta lausunnosta selviää, ettei ristiriita ollut kovin
helppo:

»30/VIII 95.

1. Sinun taiteesi ulkonaiset edellytykset, t.s. teatteri, ovat elämän,


varsinkin avioelämän (sisäisin, syvin elämänyhteys) kanssa
ristiriidassa. Teatteri, semmoisenaan, on leikkiväinen ja alhainen
(spielerisch und gemein). Tämä teatteri on ristiriidassa sinun
luonteesi kanssa. Tämä ristiriita on tehnyt sinut tarpeeksi
onnettomaksi. Et ole säilynyt puhtaana työsi vuoksi teatterissa,
vaan siitä huolimatta. Se oli taistelua. Mutta juuri tuossa taistelussa
olet kuluttanut ruumiisi ja sielusi voimat. Nyt sinä tarvitset tukea.
(Aina siihen asti, kunnes olet saanut varman kannan elämässä).
Jos nyt heität luotasi tuon turvan, niin sinä pian, ottaen huomioon
nykyiset heikot ruumiilliset ja henkiset voimasi, sorrut teatterin alle.

2. Taiteesi on ristiriidassa elämäsi kanssa. Ristiriita johtuu taiteesi


edellytyksistä (teatterista). Sinun pitää uudistaa taiteesi ottamalla
lähtökohdaksesi elämä. Ei päinvastoin. Jos tahdot menetellä
päinvastoin, muodostuu elämäsi onnettomaksi, taiteesi epätodeksi
(elämää vailla olevaksi).

3. ‒ ‒ ‒ Tuleeko taiteestasi rahanhankkimisväline? Jos niin on,


tuot ristiriidan sekä elämääsi että taiteeseesi — (sisäisin ja syvin ei
kuulu joukolle eikä sen rahalle). Samalla teet myöskin yhteisen
taloutemme mahdottomaksi.

Nämä kolme asiaa sanon sinulle ennakolta.

30 p:nä elok. 1895.

Alexander.»

Vapaaherra Uexküll-Gyllenband on selittänyt, että hän Ida


Aalbergin vaatimuksesta ja Ida Aalbergin taiteen hyväksi luopui
omasta filosofisesta kutsumuksestaan. Ei ole mitään syytä epäillä
tämän vakuutuksen vilpittömyyttä. Ida Aalberg kaipasi taiteessaan
aina älyllistä tukea, ja on varsin todennäköistä, että hän, niinkuin
vapaaherra Alexander Uexküll-Gyllenband on kertonut, joskus kysyi
miehensäkin mielipidettä (vaikka tämä oli oppinut halveksimaan
teatteria), ja, havaittuaan saamansa neuvot hyviksi, alkoi niitä sitten
pyytää yhä enemmän.
Elämänkutsumuksen menettäminen on suuripiirteistä tragiikkaa,
niin suuripiirteistä, ettei se enää tahdo saada vastakaikua tavallisen,
jokapäiväisen ihmisen sielussa. Tavallinen ihminen pikemminkin
käsittäisi niin, että tuon opettaja-oppilas-suhteen muodostumisessa
Ida Aalbergilla täytyi olla omat kipunsa ja kärsimyksensä.

Ida Aalberg oli useita vuosia miestään vanhempi. Hänellä oli


ainakin Suomessa suuren taiteilijan maine. Hän oli tunnustetusti
kansallisen näyttämötaiteen etevin edustaja kotimaassaan. Hänen
työnsä arvoa ja merkitystä oli liian usein juhlittu ja vakuutettu, jotta
hän itse olisi voinut sitä epäillä.

Lähtien omasta filosofisesta kannastaan vapaaherra Uexküll-


Gyllenband sanoi, ettei Ida Aalbergin taide vakavassa mielessä ollut
mitään. Ida Aalberg tarvitsi filosofin apua ja kouluutusta, ennenkuin
hänen »hyvä, sisäinen olemuksensa» voi tulla tehoisaksi ja
ennenkuin hän pystyi luomaan todellista taidetta.

Vain suuret uudistajat ovat ensiluokan neroja. Vapaaherra Uexküll-


Gyllenband vaati Ida Aalbergin taiteessa ja näyttämötaiteessa
yleensäkin alusta alkaen suurta uudistusta.

Mennessään avioliittoon Ida Aalbergin kanssa ja avioliittonsa ensi


vuosina vapaaherra Alexander Uexküll-Gyllenband tietenkin oli
näyttämötaiteen alalla täysi maallikko. Hän oli nähnyt teatteria vain
katsomosta. Hänen näyttämöfilosofiansa oli silloin mitä puhtainta
kamarifilosofiaa. Hän väitti, että näyttämötaide oli valheellista ja
kaipasi uudistusta, mutta miten tuon uudistuksen piti tapahtua, siitä
on mahdotonta saada mitään varmaa kuvaa, ellei tyydy sellaisiin
sanoihin kuin: »taiteen on lähdettävä elämästä», »taiteen on oltava
sisäisesti puhdasta» j.n.e.
Työ opettaa tekijänsä. Ajan oloon, jouduttuaan Ida Aalbergin
elämänkumppanina käytännölliseen näyttämötyöhön, vapaaherra
Uexküll-Gyllenbandista kehittyi mitä tyypillisin ohjaaja-taiteen
edustaja. Ida Aalberg oli mitä tyypillisin näyttelijä-taiteen. edustaja.
Varsinaisessa ammattityössä heidän etunsa ja tarkoitusperänsä
tulivat olemaan jyrkästi ristiriitaiset. Tämä ristiriita oli todellinen
ristiriita, jolle vapaaherra Uexküll-Gyllenband kuitenkin ummisti
silmänsä loppuun asti. Voimakas yksilöllinen näyttelijä ja voimakas
yksilöllinen ohjaaja eivät tahdo menestyä saman katon alla, ja suuria
näyttelijöitä ei helposti synny teattereissa, jotka toimivat
ohjaajataiteen merkeissä — se tosiasia, että joku Josef Kainz on
nuoruutensa päivinä kuulunut Meiningenin herttuan seurueeseen tai
että Moskovan taideteatterissa on ollut eräitä verraten eteviä
näyttelijöitä, ei riitä kumoamaan väitettä. Ida Aalbergin muistoksi
perustetussa teatterissa, varsinkin avajaisnäytännössä, nähtiin
hyvää, suorastaan erinomaista taidetta, mutta se oli vapaaherra
Uexküll-Gyllenbandin ohjaajataidetta, joka paljon enemmän perustui
Moskovan Taide-teatterin antamiin vaikutelmiin kuin Ida Aalbergin
omaan taiteeseen. Kuka ajatteli yksityisiä näyttelijöitä tuossa
avajaisnäytännössä? Kuka ajatteli ohjaajaa, kun Ida Aalberg itse
seisoi näyttämöllä?

Yksilöllisen, voimakkaasti yksilöllisen ohjaajan käsissä näyttelijä


helposti muuttuu marionetiksi. Vapaaherra Uexküll-Gyllenband oli
voimakkaasti yksilöllinen, jos kukaan. Ida Aalbergilla oli liian suuri
menneisyys, jotta uusi koulu olisi voinut hänet täydellisesti muuttaa.
Teoriassa hän kyllä kannatti miehensä oppeja ja toisteli kiltisti
kaikkea, mitä vapaaherra Uexküll-Gyllenband teoreettisella älyllään
oli keksinyt, mutta käytännössä hän pysyi samana diivana, joka hän
oli ollut alusta alkaen.
Jos yksilöllinen näyttelijä ja yksilöllinen ohjaaja tahtovat toimia
yhdessä, syntyy kummankin puolelta loputon sarja kompromisseja.
Niin täytyi käydä myöskin Ida Aalbergin ja vapaaherra Uexküll-
Gyllenbandin yhteistoiminnassa, joskin nämä kompromissit lienee
tehty hiljaisuudessa, jopa useissa tapauksissa itsetiedottomastikin.

*****

Kun Ida Aalberg oli mennyt naimisiin saksankielisen miehen


kanssa ja joutui yksityisessä elämässään kaiket päivät käyttämään
saksaa, arvelivat Bergbomit, että hänen vanha halunsa päästä
esiintymään Keski-Euroopassa saisi siitä uutta virikettä. Niin kävikin.
Jo 1896 Ida Aalberg oli neuvotteluissa Meiningenin herttuan
seurueen kanssa, ja vähältä piti, ettei hänestä tullut silloin tuon
teatterin jäsen. Asia oli jo sovittu, kun vapaaherra Uexküll-
Gyllenband sai koko aikeen raukeamaan pitämällä itsepintaisesti
kiinni eräästä turhasta muodollisuudesta. Bertha Forsman on
kertonut, ettei Ida Aalberg koskaan voinut antaa miehelleen anteeksi
tämän silloin osoittamaa huonoa diplomatiaa, ja että vapaaherra
Uexküll-Gyllenbandkin myöhemmin katui erehdystään.

Ollessaan kesällä 1897 vaikeasti sairaana Hangossa Ida Aalberg


kirjoitti Bertha Forsmanille m.m.:

»Etten päässyt Meiningeniin, oli minulle raskas isku, oli kuin


jotain olisi murtunut minussa, — olinhan rakentanut koko tulevan
taiteellisen kehitykseni Meiningenin varaan. Se oli minulle kova isku
siksi, että minä aina panen kaiken voimani yhteen ideaan.
Harrastus puhtaaseen ja jaloon taiteeseen, harrastus, jonka tiedän
olevan vallalla Meiningenin näyttämöllä, antoi minulle uskon, että
minä panemalla taiteelliset voimani tuon teatterin käytettäväksi
voisin raivata itselleni tietä tarvitsematta nähdä sitä likaa, jota
nykyisin on niin monilla näyttämöillä ja joka on minusta niin
vastenmielistä. — Ja kuitenkin uskon nyttemmin, että oli parasta,
kun kävi niinkuin kävi. Olen lukenut muutamia tri Lindaun teoksia,
ja minä luulen, että meillä olisi ollut aivan erilaiset mielipiteet sekä
kirjallisissa että taiteellisissa kysymyksissä, ja senvuoksi olisi
yhteistyömme varmaan käynyt vaikeaksi. — Ette saa olla minulle
vihainen, että minä vasta nyt, kun kaikki on lopussa, yleensä
mainitsen Teille mitään koko Meiningen-jutusta, mutta Tehän
tiedätte, että minä vastenmielisesti puhun suunnitelmistani ja etten
milloinkaan puhu siitä, mikä on lähinnä sydäntäni. — Uskotteko,
että Meiningeniin nähden enää mitään olisi tehtävissä? — Millähän
tavoin voisi solmia siteet uudestaan.» —

Ida Aalberg oli valmistellut liittymistään Meiningenin herttuan


seurueeseen huolellisesti ja innokkaasti. Kesällä 1896 hän oli
miehensä kanssa ollut Sveitsissä ja ottanut opetusta saksan kielen
ääntämisessä kuuluisalta foneetikolta professori Edward Sieversiltä.
Hänen kirjeidensä sävy tältä ajalta on iloinen ja toivorikas, hän
kertoo tekevänsä alppivaelluksia ja olevansa tyytyväinen ja
onnellinen. Ida Aalberg ei, kuten professori Sievers myöhemmin on
kertonut, pystynyt voittamaan kielen vaikeuksia, mutta tämä ei
suinkaan ole pääsyy, jonka vuoksi on otaksuttava, ettei hän olisi
menestynyt Meiningenin näyttämöllä, vaikka olisikin tullut sen
jäseneksi. Paul Lindaun, silloisen meiningeniläisten johtajan, kirjoista
Ida Aalbergissa itsessäänkin heräsi epäilys, että hänen taiteelliset
pyrkimyksensä tähtäsivät toisaalle kuin meiningeniläisten, ja varmaa
on, että hän yhtymällä tuohon mainehikkaaseen seurueeseen olisi
joutunut aivan vieraaseen ja itselleen mahdottomaan ympäristöön.

Millaista olikaan meiningeniläisten näyttämötaide?


Voidaan tosin väittää, ja on usein väitettykin, että se
näyttämötaide, joka 1800-luvun jälkipuoliskolla suurten yksityisten
näyttelijäin saavutuksissa vietti voittojaan Euroopassa, ei ole oikean
näyttämötaiteen ihanne. Tähtisysteemi oli syntynyt Saksan
porvarillisilla näyttämöillä köyhyyden ja taloudellisen ahdingon
vuoksi, Italiassa vakinaisten teatterien puuttumisen vuoksi ja
Ranskassa ehkä lähinnä yksityisten näyttelijäin kurittomuuden ja
oikuttelujen vuoksi. Nuo kiertelevät suuruudet merkitsevät kuitenkin
— huolimatta siitä, missä olosuhteissa olivat syntyneet ja huolimatta
siitä, että heidän työnsä ei perustunut mihinkään suureen, koko
näyttämötaidetta käsittävään ajatukseen — _näyttelijä_taiteen
huippusaavutusta.

_Ohjaus_taidetta on kernaasti taituttu pitämään suuremmassa


arvossa kuin sitä taidetta, joka ilmenee yksityisten näyttelijäin
saavutuksissa. Meiningeniläisten tavaton merkitys uudemman
näyttämötaiteen historiassa perustuu yksinomaan regian, ohjauksen
kehittämiseen. Mutta meiningeniläisten historiasta voi nähdä, ettei
tuonkaan suunnan takana ole ollut mitään syvällistä ajatusta eikä
suurta taidekäsitystä, vaan että se pikemminkin johtui erään
ruhtinaan oikusta.

Nerokkaan näyttelijän Edmund Keanin »keskinkertaisen lahjakas


poika» Charles Kean oli 1850 Princess Theatressa Lontoossa
esittänyt Shakespearen »Venetian kauppiaan» suunnattoman
loisteliaassa ja komeilevassa näyttämöasussa ja saavuttanut näillä
puhtaasti ulkonaisilla keinoillaan menestyksen, joka ei ihastuttanut
vain sen ajan englantilaisia, vaan vaikutti koko englantilaisen
näyttämötaiteen myöhäisempään suuntaan, mikä on ilmennyt
sarjana toinen toistaan loistavampia näyttämöasetuksia. Tieto tästä
suuresta, mutta todellisuudessa sangen kysymyksenalaisesta, jopa
suorastaan sangen pienestä taiteellisesta voitosta levisi
mannermaalle, ja Kean, joka »Venetian kauppiaan» jälkeen tarjosi
Lontoolle Shakespearen kuningasdraamoja kulisseina ja pukuina,
sai pian matkijoita Saksassa. Kun Coburgin hoviteatterissa 1867
esitettiin englantilaisena jäljennöksenä »Venetian kauppiasta», sattui
katsomossa istumaan nuori Meiningenin herttua Georg, joka tuosta
näytännöstä sai jumalaisen kipinän. Meiningenin herttua ei ollut
mielipuoli eikä tullutkaan siksi niinkuin ovat tulleet muutamat muut
näyttämötaidetta rajattomasti suosivat ruhtinaat, mutta mikään ei
todista, että hänen taidekäsityksensä olisi ollut erikoisen syvällinen,
kun hän muutamia vuosia myöhemmin lähetti hyvin harjoitetun ja
varustetun teatterinsa valloittamaan Eurooppaa. Meiningeniläisille
merkitsivät kulissit, ihmeen ihanat matot, tyylinmukaiset sotilaspuvut
j.n.e. enemmän kuin yksityiset näyttelijät. Heidän suurella vaivalla ja
huolella valmistettuihin joukkokohtauksiinsa, joita on paljon kiitetty,
tahtoivat yksityisen näyttelijän vuorosanat hukkua, »Venetian
kauppiaassa» ei kuultu Shylockin merkillistä puhetta, kun hän eroaa
tyttärestään, koska samaan aikaan näyttämöllä monilukuinen
avustajakunta gondoloineen ja kirjavine pukuineen veti
näyttelemisellään suurimman huomion puoleensa. Koko
meiningeniläisyys kaikkine epäilemättä suurine ansioineen ja
ankaroine töineen ei merkinnyt runoluoman syventämistä
näyttämöesityksessä, vaan sen levittämistä.

Meiningenilaisilla oli ankara kuri. Näyttelijätär, joka tänään esitti


»Orleansin neitsyttä», saattoi huomenna kuulua avustajakuntaan.
Tuntuu tuiki mahdottomalta uskoa, että Ida Aalberg olisi voinut
alistua tuohon kuriin ja koko ohjaustaiteeseen yleensäkään, jos
ohjaajana oli vieras henkilö, jota hänen ei ollut äärimmäinen pakko
totella.
Meiningenilaisten tyyliä on, ja tuskin ilman syytä, mainittu
naturalistiseksi. Näyttämöasetuksessa tuli ovien, ikkunain,
huoneiden sisustuksen, pukujen, puiden, yleensä kaiken olla
mahdollisimman luonnollista. Heidän naturalisminsa ilmeni sentään
muussakin kuin tässä ulkonaisessa puolessa, Schillerin vuorosanoja
lausuttiin tuolta näyttämöltä juhlallisuutta ja pateettisuutta välttäen,
luonnollisuutta tavoitellen.

Meiningenilaisten vierailumatkasta Moskovaan sanotaan suureksi


osaksi johtuneen, että vuonna 1897 amatöörinäyttelijä ja
amatöörinäyttämön johtaja Konstantin Aleksejev, s.o. Stanislavski, ja
draamankirjoittaja Vladimir Nemirovits-Dantsenko päättivät perustaa
uuden teatterin, josta sittemmin kehittyi maailman maineen
saavuttanut Moskovan Taide-teatteri. Se kulki aluksi mitä
täydellisimmän naturalismin merkeissä, mutta myöhemmin on kyllä
sekä Stanislavskissa että eräissä hänen ohjaustaiteensa oppilaissa,
ennenkaikkea originellissa Meyerholdissa, tapahtunut täydellinen
luopuminen alkuperäisestä taiteellisesta kannasta. Stanislavski ei
anna yksityiselle näyttelijälle suurta merkitystä, mutta juuri hänen
teatteristaan Alexander Uexküll-Gyllenband löysi sen
näyttämötaiteen, jota hän kaikkein lämpimimmin kannatti. Sieltä hän
löysi totuuden ja sen hienouden ja yleensä ne ominaisuudet, joita
hän voi kunnioittaa taiteessa ja joiden nojalla hän voi taidetta
kunnioittaa. Moskovan Taide-teatterilla on vapaaherra Uexküll-
Gyllenbandin ohjaajataiteen kehittymiselle ainakin mitä tärkein, ellei
suorastaan keskeinen merkitys,[27] mutta Ida Aalbergille, ainakaan
hänen näyttelijätaiteelleen, ei moskovalaisuudella voinut olla paljon
annettavana. Hän, Ida Aalberg, kyllä tutustui henkilökohtaisestikin
Stanislavskiin, piti hänen ohjaustaiteestaan, arveli häntä neroksikin,
mutta tuo kaikki tuskin merkitsi hänelle sen enempää, varsinkin kun
ottaa huomioon, että Ida Aalberg perehtyi Moskovan taiteeseen
vasta elämänsä viimmeisinä aikoina.

Ohjaajan tultua vaikuttavaksi tekijäksi näyttämötaiteessa on tuon


taiteen luomisessa syntynyt ristiriita, joka koskee hänen ja näyttelijän
keskinäistä suhdetta. Ohjaaja voi yllyttää ja kiihoittaa näyttelijää yhä
suurempiin saavutuksiin, se on tietysti luvallista, ja se oli esim.
Kaarlo Bergbomin harvasanaisen ohjaustaiteen vahvin puoli. Mutta
ohjaaja voi myöskin pakottaa näyttelijän luopumaan itsenäisestä
käsityksestä, vaikka tämä siihen pystyisikin, tai rajoittaa hänen
luovaa mielikuvitustaan vetoamalla kokonaisuuden etuihin tai
muuten hillitsemällä häntä oman makunsa mukaiseksi. Ristiriita on
olennainen ja syvä ja niin vaikea, ettei sitä voida yleisillä laeilla ja
määräyksillä ratkaista, jää lopultakin riippumaan olosuhteista ja
psykologisesta vaistosta, minkä verran ohjaaja puuttuu yksityisen
näyttelijän työhön, ellei, niinkuin Meiningenin teatterissa tai
Stanislavskin taiteessa, näyttelijää periaatteellisesti aseteta
toisarvoiseen asemaan ja olla sitä mieltä, ettei hänen
persoonallisuutensa merkitse sen enempää draaman
tulkitsemisessa. Äärimmilleen kehitetty ohjaajataide ei edellytä vain
suurta ja yksilöllistä ohjaajapersoonallisuutta, vaan myöskin runsaita
taloudellisia mahdollisuuksia, joita voi tarjoutua vain suurissa maissa
ja suurissa, väkirikkaissa kaupungeissa, tai aineellista tukea
vallassaolijain tai rikkaiden henkilöiden kädestä.

Jos V. Meyerhold, joka nykyisin kuuluu entistä enemmän tulleen


vakuutetuksi yksityisen näyttelijän arvosta ja merkityksestä, mutta
joka jo vuosisadan vaihteessa, vaikka olikin ollut Stanislavskin
opissa, saattoi olla yhteistoiminnassa semmoisen diivan kuin
aikakauden suurimman venäläisen näyttelijättären Vera
Kommissarzevskin kanssa, olisi ollut vanhempi mies ja keksinyt
ohjaaja-ajatuksensa aikaisemmin ja jos vapaaherra Alexander
Uexküll-Gyllenband olisi tunnustautunut Meyerholdin oppien
kannattajaksi, olisi ehkä hänen ja Ida Aalbergin yhteistoiminnasta
voinut syntyä jotakin kerrassaan ihmeellistä. Mutta siihen aikaan,
kun vapaaherra Uexküll-Gyllenband luopui filosofiastaan ja alkoi
puuttua vaimonsa taiteeseen ja ohjata tätä erinomaisella tarmollaan,
ei vielä tunnettu edes Moskovan Taide-teatteria, vielä vähemmän
Meyerholdia. Aluksi vapaaherra Uexküll-Gyllenbandin ohjaus siis oli
vailla kaikkia näyttämötaiteesta saatuja vaikutelmia — ellei ota
lukuun niitä vaikutelmia, joiden mukaan hän hyljeksi kaikkea
näkemäänsä näyttämötaidetta, — ja perustui hänen omaan
filosofiaansa ja puhtaasti kirjallisiin tietoihinsa.

*****

Vapaaherra Uexküll-Gyllenband on epäilemättä vaikuttanut Ida


Aalbergin näyttelijätaiteeseenkin. Vapaaherra Uexküll-Gyllenband oli
mies, jolla oli oma tahto, jos kellään, ja joka kaikkialla halusi nähdä
oman tahtonsa toteutuvan. Henkilön, joka oli yhteistoiminnassa
hänen kanssaan, oli mahdoton välttää hänen vaikutusvoimaansa.
Vapaaherra Uexküll-Gyllenband tahtoi ulottaa määräämisvaltansa
toisen pienimpiinkin tekoihin ja ajatuksiin.

Tuommoisella voimakkaalla aggressiivisella persoonallisuudella on


omat heikkoutensa. Sen heikkous on siinä, että se helposti synnyttää
vastustusta ja vie konflikteihin. Ida Aalbergillakin oli oma voimakas
tahtonsa, sen hän oli ehkä liiankin usein osoittanut. He olivat lisäksi
kumpainenkin erinomaisen herkkähermoisia. Voisi luulla, että
yhteistyö tuommoisin edellytyksin olisi käynyt mahdottomaksi tai
ainakin kovin vaikeaksi, mutta kumpaisenkin kunniaksi on
mainittava, ettei niin suinkaan ollut laita. Kirjeistä näkee, että
vapaaherra Uexküll-Gyllenband päivä päivältä innostuu yhä
enemmän uuteen alaan ja alkaa tuntea sen omakseen ja että hän
antaa ohjeensa vaimolleen mitä rakastettavimmalla ja hienoimmalla
tavalla. Vuonna 1898 hän tunnustaa siihen asti pitäneensä jokaista
draamaa vain lukudraamana, mutta nyt alkavansa ajatella niitä
myöskin näyttämöluomina. Kuten luonnollista on, pysyvät hänen
ohjeensa edelleenkin vuosikausia kirjanoppineen selvityksinä eivätkä
näyttämöllä toimineen ammattimiehen.

Ibsen on niitä harvoja runoilijoita, jotka jo elinaikanaan saavat


opeilleen ja runoudelleen selvittäjiä. Hänen problemaattiset
draamansa tarjosivat mitä kiitollisimman kentän vapaaherra Uexküll-
Gyllenbandin filosofisesti kouluutetulle älylle. Häämatkallaan 1894 ja
vielä myöhemminkin hän oli käyttänyt Ibsenin elämänviisautta
melkein juristin tavoin todistaessaan lakkaamatta, kuinka väärässä
Ida Aalberg kulloinkin oli, mutta jo 1898 hänen ainoana
tarkoituksenaan tuntuu olleen vain luoda selvyyttä »Pohjan Velhon»
henkilöluomiin. Hänen saksalaisuutensa ei ilmene vain tavattomassa
perinpohjaisuudessa, jolla hän tutustuu Ibsenin draamoihin, vaan
myöskin kaavamaisuudessa ja hirvittävässä ja väsyttävässä
monisanaisuudessa, jolla hän tekee Ida Aalbergille selvää
tutkimustensa tuloksista. Niinpä esim. »Hedda Gabler-selostus
vuodelta 1898 on oiva näyte saksalaisesta kaavamaisuudesta ja
perinpohjaisuudesta. Se alkaa:

»Hedda. Synteettinen taulu.

(Kuinka yksityisistä luonnepiirteistä muodostuu


kokonaisuus ja kokonaiskuva Heddan kompliseeratusta
luonteesta).
1) Heddan (koko toiminnan aikana yhä kasvava)
pessimismi, elämäänkyllästyminen ja inho.
2) Rotu.
3) Ylhäinen olemus.
4) Halu loistoon ja komeuteen.
5) Vallanhimo.
6) Äly; iva ja pilkka.
7) Epätoivo; demoninen viileys.
8) Mustasukkaisuus.

*****

1) Pääkohdat, joissa tämä Heddan pessimismi, inho ja


elämäänkyllästyminen näyttäytyvät, on merkitty erikoiseen paperiin.
»(Tuohon paperiin on todella koottu hartaalla vaivalla kaikki yksityiset
lauseet, missä nuo ominaisuudet voivat ilmetä).

»2) Inhosta aiheutuu pyrkimys saada lohtua kauneudesta. Mutta


todellisuudessa Hedda ei siihen usko. (Hedda ei ole sentimentaali).
Hänen älynsä pilkkaa hänen omaa inhoansakin. Todellisuudessa
Hedda uskoo vaan valtaan ja voimaan eikä estetiikkaan ja
kauneuteen. Tästä inhosta aiheutuu myöskin inho odotettavissa
olevaa skandaalia kohtaan. —

3) Ylhäinen olemus. Rotu on jotakin fyysillistä…. $en pitää ilmetä


koko olemuksessa, käynnissä liikkeissä, kynsien puhdistamisessa
j.n.e. Ylhäinen olemus tulee esiin seuraavissa kohdissa:
tohvelihistoriassa» — — — — — — j.n.e., j.n.e.

*****
Ida Aalbergin taiteilijaluonne ei ollut älyllinen. Vapaaherra Uexküll-
Gyllenbandin intellektualismilla jota tarjottiin ylen runsaina
annoksina, oli omat vaaransa Ida Aalbergille. Neuvoja oli liian paljon,
jotta Ida Aalberg olisi voinut niitä kaikkia edullisesti hyödykseen
käyttää. Ne voivat muodostua raskaaksi painolastiksi, joka voi
hävittää hänen taiteensa suurimman vaikutusvoiman: välittömyyden
tunneilmauksissa. Hän tarvitsi Bergbomilta vain pienen vihjauksen
tai lyhykäisen lauseen ohjauksekseen, heti sanotaan hänen
joutuneen inspiratsionin valtaan, joka voimallaan ja hehkullaan ylitti
kaikki toivomukset, ja pystyneen tällöin luomaan sellaista, mitä
ohjaaja omassa mielikuvituksessaan ei ollut nähnyt, mutta johon hän
täysin yhtyi. Tuommoista hetkelliseen inspiratsioniin perustuvaa
taidetta voidaan syyttää pintapuolisuudesta, mutta älyn ylivalta vie
tehon taiteelta ja tekee esityksen kuolleeksi.

Mikä vaikutus oli vapaaherra Uexküll-Gyllenbandin ensimmäisien


vuosien aikana antamalla ohjauksella Ida Aalbergin taiteeseen?
Vaikuttiko hän siihen edullisesti vai ehkäisevästi?…

Vapaaherra Uexküll-Gyllenband oli itse varma silta, että hänen


vaikutuksensa oli mitä siunauksellisin. Vuonna 1903 hän laati Ida
Aalbergin taiteesta kirjoituksen saksalaista »Die Woche» lehteä
varten kirjoituksen, joka kai olisi ollut jo aivan liian pitkäkin
voidakseen tulla julkaistuksi. Siinä hän erottaa Aalbergin taiteessa
kaksi suurta, määräävää ominaisuutta, joista toinen on eräänlainen
sisäisyys, tunneilmausten välittömyys, mutta toinen ja hänen
mielestään paljon tärkeämpi, se johdonmukaisuus, konsekvenssi,
jolla Ida Aalberg muka on tunkeutunut runoilijan ajatusten
perimmäisiinkin sopukoihin ja sulattanut kaiken eläväksi ja
yhdenmukaiseksi ja todelliseksi taideteokseksi.
Jotkut niistä, jotka ovat tunteneet Ida Aalbergin taiteen jo
1880-luvulla, m.m. Bertha Forsman ovat väittäneet, että nuori Ida
Aalberg antoi taiteilijana paljon enemmän kuin vapaaherratar Ida
Aalberg-Uexküll-Gyllenband. Huomattava on kuitenkin, että Bertha
Forsman syytti Ida Aalbergia liiasta hillinnästä jo 1891.

Sanomalehdissä ja aikakausjulkaisuissa esiintyvät kritiikit tietävät


kertoa sangen vähän Ida Aalbergin taiteen älyllisestä puolesta.
Tietysti siitä sanotaan silloin tällöin jokin ylimalkainen fraasi, mutta
kun hänen myöhempääkin taidettaan kiitetään todellisella
innostuksella, soivat sanat temperamentin, tunnevoiman,
intoutumisen ylistystä eivätkä suinkaan loogillisten avujen.

Epäilemättä Ida Aalberg kuitenkin alusta alkaen sai jotakin hyötyä


näistä suurella ilolla ja hartaudella annetuista neuvoista, joskaan
hyöty kaiken todennäköisyyden mukaan ei muodostunut niin
ratkaisevaksi kuin antaja toivoi ja uskoi. Vielä enemmän hän
varmaan hyötyi vapaaherra Uexküll-Gyllenbandin ohjauksesta, kun
tämä kehittyi ja alkoi saada todellista ammattimiehen silmää.
Vuosien kuluessa vapaaherra vapautui yhä enemmän
filosofeeraamisesta ja kirjallisesta saivartelemisesta, ja hänen
ohjauskirjansa alkoivat muistuttaa muiden ohjaajien kirjoja:
merkinnät alkoivat yhä enemmän koskea lausuntaa, mielentiloja,
asemia, ilmeitä, liikkeitä ja saada yhä käytännöllisemmän sävyn.
Niissä ei ainakaan tapaa niin paljoa kuin varhaisempien vuosien
ohjeissa perinpohjaisia selityksiä siitä, mikä on itsestään selvää tai
mitä ei ainakaan tarvitsisi monisanaisesti todistaa. Ne paksut
paperipinkat, joihin hänen muistiinpanonsa on tehty, eivät todista
ainoastaan hänen innostustaan ja tarmoaan ja älyllisen erittelynsä
terävyyttä, vaan niistä voi löytää myöskin paljon sellaista, joka
osoittaa hänellä olleen taiteilijan mielikuvituksen ja herkän
havaintokyvyn.

Eläessään yhdessä vapaaherra Uexküll-Gyllenbandin kanssa Ida


Aalberg luki paljon ja luki varmaan järjestelmällisemmin kuin
milloinkaan ennen. Mutta vaikka hän näin viljeli omaa henkeään,
kadotti hän itsenäisen ajattelemisen lahjan suorastaan
hämmästyttävässä määrässä. Vuoden 1904:n jälkeen hän puhui
usein ja paljon teatterista ja taiteesta, kirjoitti niistä myöskin monesti,
mutta melkein poikkeuksetta on kaikki se, mitä hän on puhunut tai
kirjoittanut, ollut vapaaherra Uexküll-Gyllenbandin todistettavasti
hänelle dikteeraamaa.

Vaikkakin Ida Aalbergin voi sanoa elämänsä loppupuolella


toimineen ehkä liian voimakkaan henkisen painostuksen alaisena,
tiedetään hänen harvoin sitä valittaneen. Vuosien kuluessa hän oppi
entistä enemmän luottamaan mieheensä, hän alistui, hän tyytyi
henkisen itsenäisyyden menetykseen, ja silloin oli kaikki hyvin, hyvin
ainakin heidän kahden kesken.
XVII.

UUDEN KOULUN VAIKUTUKSESTA IDA AALBERGIN TAITEESEEN.

Kevätkaudella 1897 Suomalaiseen teatteriin tekemänsä vierailun


jälkeen Ida Aalberg pysyi poissa näyttämöltä lähes kolme vuotta.
Näihin aikoihin hän oli epätoivoissaan terveytensä tilasta, saman
vuoden syksyllä hän sai Helsingissä kestää leikkauksen ja
myöhemmin hän koki etsiä parannusta useista ulkomaisista kylpy- ja
terveydenhoitolaitoksista. Kirjeenvaihdosta näkee, että hän oleskeli
milloin Lausannessa, milloin Dresdenissä, milloin missäkin.
Menetettyä terveyttään hän ei koskaan saanut täysin takaisin, ja
myöhemminkin tarkoittivat lakkaamatta toistuvat matkat etelään
pikemmin ruumiinhoitoa kuin henkistä virkistymistä tai huvittelua.

On täysi syy uskoa, että vapaaherra Uexküll-Gyllenbandilla oli


varsin vähäinen osuus siihen taiteeseen, jota Ida Aalberg pitkällisen
sairausaikansa jälkeen esitti vierailunäytännöissään Suomalaisessa
teatterissa. Vapaaherran tutkimukset koskivat melkein yksinomaan
Ibseniä, hänen papereissaan on ainakin tältä ajalta verraten vähän
yrityksiä etsiä johdonmukaisuutta ja syvyyttä toisten runoilijain
teoksista. Suomalaisessa teatterissa Ida Aalberg esiintyi eräissä
entisissä loisto-osissaan; uusia tehtäviä hänellä oli vain kahdessa
Sardoun kappaleessa,»Fedorassa» ja »Theodorassa», joissa
kumpaisessakin hän häikäisi helsinkiläisiä arvostelijoita ja yleisöä
teknillisellä taituruudellaan. »Fedoran» johdosta Jalmari Hahl sai
aiheen »Valvojassa» verrata Ida Aalbergin tekniikkaa Sarah
Bernhardiin tekniikkaan ja todeta, että kumpainenkin on yhtä
kehittynyttä, mutta että Sarah'n näyttelemisessä on enemmän
maneeria ja asentojen liiallista vaativaisuutta. Samaa ylivoimaisen
tekniikan ylistystä sisältää esim. Bertel Gripenbergin »Euterpeen»
»Teodorasta» kirjoittama kritiikki.

Nämä vierailunäytännöt Suomessa olivat luonteeltaan verraten


tilapäisiä ja ikäänkuin puolella sydämellä tehtyjä. Ida Aalbergin
mielessä kangasteli edelleenkin suurempi päämaali, eurooppalaisen
maineen voittaminen, ja vapaaherra Uexküll-Gyllenband tuki kaikin
voimin hänen pyrkimyksiään tässä suhteessa. Parannusmatkoilla
ollessaan Ida Aalberg vielä kerran koetti perehtyä saksankielen
ääntämiseen ja Pietarissa hän sai opetusta eräältä lausujalta, rouva
Marie Strauch-Spettiniltä.

Vapaaherra Uexküll-Gyllenband oli siis muodostanut oman


teoriansa siitä, miten Ibseniä ja draamaa yleensäkin oli näyttämöltä
esitettävä. Tämän teorian ydin oli siinä, että näyttämöltä oli esitettävä
runoilijaa eikä näyttelijää. Näyttämöesityksen tuli pohjautua
perusteelliseen tutkimukseen ja erittelyyn runoluomasta, runoilijan
tarkoituksista ja ajatuksista, joita hän oli kätkenyt teokseensa. Se
hylkäsi kaiken ulkonaisen efektintavoittelun.

Vapaaherra Uexküll-Gyllenbandin teoria ei sinänsä ole mitään


uutta eikä vallankumouksellista oppia. Se on vain sangen yleisen
vaatimuksen tehostamista. Kukapa näyttelijä tai ohjaaja väittäisi
ilman muuta olevansa välinpitämätön runoilijan tarkoituksista tai

You might also like