PostDoc and PhD positions on Faculty of Informatics, TU Wien, Austria
For a new, 3-year research project to commence in Summer/Autumn 2013, funded by the Austrian Science Fund (FWF), the Research Group for Parallel Computing at the Institute of Information Systems, Faculty of Informatics, Vienna University of Technology (TU Wien) is offering
* A fully funded, 40 hour/week Postdoc position (40 hours)
* Two 30 hour/week PhD positions
for research, tool development, and extensive hypothesis testing/benchmarking within a new project on "Verifying Self-consistent MPI performance guidelines". The project aims at formulating rules that can guide the quality and performance portability of MPI (Message-Passing Interface) implementations (possibly extending the ideas towards other parallel languages, interfaces and libraries), and develop means to automatically verify and detect violations to such rules. Important aspects of the project are rule-directed benchmarking, computer system modeling, communication algorithms, automated, rule-based (statistical) result analysis, and include analysis and development of statistically well-founded, fast, accurate, comprehensive and extensible (customizable) benchmarking procedures for MPI. The project can build on previous work in the MPI community [1,2,3,4,5], and is expected to significantly extend on, broaden and substantiate such results. A detailed project description, that will be the starting point for the research, will be available to candidates considered for the positions.
The positions are research positions and come with no teaching obligations, although general interest in the Parallel Computing group's activities is expected; teaching on a case-by-case basis will be possible and is encouraged (especially for the Post Doctoral researcher). Salaries for both types of positions (40 and 30 hour) are in accordance with the Austrian standard rate for University employment. For the PhD positions the goal of the project is to provide sufficient technical material for PhD dissertations by the end of the project. The project will offer good opportunities for strong and significant interaction with the MPI community. The project will be led by Jesper Larsson Traff, in close cooperation with the Post Doctoral researcher, who is expected to assist significantly in supervising the project and guiding the research.
The focus of the research group for Parallel Computing at the Vienna University of Technology is on means and methods for efficiently utilizing different, real (e.g., clusters and shared memory systems with accelerators and other HW support) as well as idealized, (e.g., fully connected communication networks, PRAM) parallel architectures for the solution of given computational problems. This includes the design, development and implementation of parallel algorithms and data structures for fundamental, basic problems (e.g., communication, reduction) and problems incurred by applications, of parallel programming models, interfaces, languages and libraries and their algorithmic support, and the study of parallel architectures.
Required qualifications (PhD and PostDoc positions):
- Good background in computer science, especially in Parallel computing, Computer Systems or architecture, Algorithms and data structures, also including theoretical foundations (complexity and computability)
- Strong prior knowledge and experience with MPI, either at application or implementation level, appreciation of the MPI standard per se
- Good programming skills in C/C++, preferably also Fortran, knowledge of other paradigms and languages
- Good software engineering skills
- Mathematical and analytical skills, especially in statistics and probability, including computer system and performance modeling.
- Strong interest and enthusiasm for research and development in parallel computing, including practical implementation and experimentation
- An idea of the direction of the project as indicated in [1,2,3,4,5]
Additional qualifications for Postdoc applicants:
- A good PhD in computer science, electrical engineering, mathematics, or other related area
- Very good background in statistics, probability and computer system benchmarking and analysis
- A qualitatively strong publication record in one or more parallel computing themes, practically or theoretically oriented, e.g., in parallel interfaces, programming models, their implementation and run-time support, parallel algorithms, performance modeling and benchmarking
The Parallel Computing Group of the Vienna University of Technology is conveniently and attractively located in central Vienna, and currently consists of a professor, two PhD students, two Post Doctoral researchers, and a system administrator. Speaking German is not required, but fluency in English (written and spoken) is a must. Willingness to learn German for foreign applicants is a plus. In addition to documenting the required qualifications the application should include a short statement of (general) interests, a suggested, project related research outline, and the name of one (PhD applicant) to two (PostDoc) references than can be contacted for supporting information.
For further questions please contact Prof. Dr. Jesper Larsson Traff, traff@par.tuwien.ac.at
Deadline for application is 3rd of June, 2013. The positions will remain open until filled.
Please direct your application to kamper@par.tuwien.ac.at and steinbuch@par.tuwien.ac.at, including an outline of project related research (maximum one page), CV, PhD and MSc certificates highlighting relevant background, and list of publications with 1-3 highlighted main results (for the Post Doctoral position).
[1] Jesper Larsson Traff, William D. Gropp, Rajeev Thakur:
Self-Consistent MPI Performance Guidelines.
IEEE Trans. Parallel Distrib. Syst. 21(5): 698-709 (2010)
[2] William Gropp, Torsten Hoefler, Rajeev Thakur, Jesper Larsson Traff:
Performance Expectations and Guidelines for MPI Derived Datatypes.
EuroMPI 2011: 150-159
[3] William D. Gropp, Dries Kimpe, Robert B. Ross, Rajeev Thakur, Jesper Larsson Traff:
Self-consistent MPI-IO Performance Requirements and Expectations.
PVM/MPI 2008: 167-176
[4] Jesper Larsson Traff:
MPIcroscope:
Towards an MPI Benchmark Tool for Performance Guideline Verification.
EuroMPI 2012: 100-109
[5] Ralf Reussner, Peter Sanders, Jesper Larsson Traff:
SKaMPI: a comprehensive benchmark for public benchmarking of MPI.
Scientific Programming 10(1): 55-65 (2002)
Research Group Parallel Computing
Faculty of Informatics
 Vienna University of Technology (TU Wien)
Vienna, Austria
www.par.tuwien.ac.at