|Home | Research | Publications | Teaching | Contact|
My current research interests include:
performance-oriented software engineering; high-level modeling and
specification of parallel programs for large-scale computing
systems; cognitive computing techniques for parallel programming; system design and optimization using meta-heuristics and machine learning.
I also enjoy studying the properties of various kinds of systems (such as parallel computers, flexible manufacturing systems or health-care systems) using innovative modeling and simulation techniques.
In my parallel computing lab at LNU we use for research DISA that comprises 19 DELL EMC nodes (14 PowerEdge R740 and 5 PowerEdge R740XD); heterogeneous nodes of DISA contain two 20-core CPUs (Intel Xeon Gold 6148) and four NVIDIA GPUs.
|- Suejb Memeti (currently PostDoc at LiU)|
|- Iryna Talamanova (MS student)|
|- Yasser Alsouda (MS student)|
|- Pranav Patel (BS student)|
|- Corne Roozemond (BS student)|
|- Shishengxiong Zhong (BS student)|
|- Meng Li (BS student)|
European ICT COST Action (IC1406) on High-Performance Modelling and Simulation for
Big Data Applications (cHiPSet).
Contributions: Parallel Programming Models (WG2)
|- ETP4HPC: The European Technology Platform for High Performance Computing|
|Contributions: Associate member|
European Network on High Performance and Embedded
Architecture and Compilation.
Contributions: Member of the HiPEAC network of excellence.
|- Software Technology for Self-Adaptive Systems. Funded by the Swedish Knowledge Foundation. Project period 2015 - 2019. This project consists of three subprojects (A, B, and C).|
|Contributions: Principal Investigator for subproject C.|
|- SciChallenge: Next Generation Science Challenges Using Participatory Techniques and Digital Media. Research and Innovation Action (RIA) funded by the EU research and innovation programme Horizon 2020 (H2020).|
|Contributions: Work Package Leader, Member of the Scientific Steering Board|
|- iGlass. Sub-project of the SHS project funded by VINNOVA; in collaboration with Glafo, Sigma IT and Sigma Connectivity. iGlass studies intelligent transparent building elements of smart homes enriched with computational, communication, sensing and interacting capabilities that can adapt to user preferences and environment conditions.|
|Contributions: Project Coordinator|
|- PEPPHER: Performance Portability and Programmability for Heterogeneous Many-core Architectures. EU FP7 STREP Project. PEPPHER has been top-ranked during the evaluation and no other project proposal that addressed the topic ICT-2009.3.6 received more points. The total budget of PEPPHER is 3.44 million.|
|Contributions: Project Coordinator during the project implementation. Also coordinated the project proposal preparation.|
Automatic Online Tuning. EU FP7 STREP Project.
Contributions: performance modeling techniques.
EU FP7 Support Action.
Contributions: member of the PlanetHPC network. Contribution to a roadmap for future HPC research.
MY SCIENCE: EU FP7 Project.
Contributions: training the young European journalists to report about the EU ICT-related research. Contributed the ICT-related part to the project proposal. Devised the program and coordinated the implementation of the ICT-related workshop.
Advanced Data Mining and Integration Research for Europe. EU FP7 STREP Project.
Contributions: process designer; CRISP-DMI.
Integrated Biomedical Informatics for the Management of Cerebral
Aneurysms. EU FP6 IP Project.
Contributions: high-level specification of QoS-aware grid workflows.
HiPro: high-level programming of multi-core computing systems.
HiPro aims to develop a high-level program development model for
multi-core computing systems, which hides the complex details of
low-level multi-threading. Our approach supports a compositional
program development process, where programs are constructed from
smaller parallel building blocks that are adapted and tuned for a
specific multi-core platform.
Responsibilities: high-level programming of multi-core computing systems.
Contributions: an intelligent compositional program development approach for multi-core computing systems.
Networked European Software and Services Initiative (NESSI) is the
European Technology Platform on Software and Services.
Contributions: Grid Vision and Strategic Research Agenda (V.3.0).
a holistic service-oriented environment for
QoS-aware Grid workflows. Amadeus considers user
requirements (in form of QoS constraints) during workflow
specification, planning, and execution. A distinguishing feature of
Amadeus is the support of a comprehensive set of QoS requirements,
that consider in addition to performance and economical aspects also
legal and security aspects. Amadeus project introduced the concept
of Location Affinity, which allows the user to express preferences
regarding the location of Grid resources where an activity should be
executed, by specifying the Grid site, organizational, or
Responsibilities: specification of QoS-aware Grid workflows.
Contributions: (1) development of a UML-based domain specific language (DSL) for QoS-aware Grid workflows and the corresponding tool-support; (2) development of the concept of Location Affinity.
a tool set for cluster and Grid computing.
Responsibilities: (1) performance modeling and prediction; (2) specification of Grid workflows.
Contributions: (1) development of a hybrid approach for performance modeling and prediction, which combines mathematical modeling and discrete-event simulation; (2) development of a UML-based domain specific language for performance-oriented parallel and distributed programs; (3) development of the Abstract Grid Workflow Language (AGWL), which is an XML-based workflow language; (4) development of a UML-based domain specific language for Grid workflows.
Aurora: Advanced Models, Applications and Software Systems for High Performance Computing.
A special research program (SFB) funded by the Austrian Science Fund (FWF), 1997--2007.
The special research program Aurora was a distinctively interdisciplinary project, based
upon a highly synergetic cooperation of the participating
institutions across a broad range of disciplines that include:
parallel languages, compilers, runtime systems, performance
evaluation tools, numerical algorithms, computational finance,
theoretical chemistry, and atto-second physics.
Responsibilities: performance modeling and prediction.
Contributions: development of an approach and the corresponding tool-support for the performance modeling and prediction of parallel and distributed computing systems.
a performance modeling and prediction system for
parallel and distributed programs. Our approach supports the
graphical specification of performance model in a human-intuitive
fashion on one hand, and on the other hand is amenable to
machine-efficient model evaluation. The model transformation, from
the graphical human-intuitive form (that is, UML representation), to
the form that can be efficiently evaluated by a machine (that is, C++
representation), is performed automatically. Our methodology for
performance prediction of computing systems combines mathematical
modeling and simulation. We use mathematical modeling to develop
parameterized performance models (that is, cost functions) for
components of the system. Thereafter, we use discrete-event
simulation to describe the structure of system. The aim is to combine the model evaluation
efficiency of mathematical performance models with the structure
awareness of simulation models.
Responsibilities: principal investigator.
Contributions: design and implementation of Performance Prophet.
a platform-independent and extensible graphical editor for the UML-based
modeling of parallel and distributed programs. The main
components of Teuta are: Model Checker, Model Traverser, and the
Graphical User Interface (GUI). The GUI of Teuta is used for the
development of UML model. The Model Checker is used to verify
whether the model conforms to the UML specification. For the
generation of different model representations (such as XML or C++)
the Model Traverser is used.
Responsibilities: principal investigator.
Contributions: design and implementation of Teuta.