News
Gastvortrag: Software Engineering Challenges of the H2020 DEEP-EST Modular Supercomputing Architecture
Herr Prof. Dr. Helmut Neukirchen der University of Iceland wird am 07.11.2017 um 16:00 Uhr (c.t.) im Seminarraum 2.101 des Instituts für Informatik einen Vortrag über das H2020-Projekt DEEP-EST halten. Gäste sind herzlich willkommen.
Vortragstitel: Software Engineering Challenges of the H2020 DEEP-EST Modular Supercomputing Architecture
Abstract: While Moore's law still gives as a doubling of the number of transistors in an integrated circuit every two years, this does not anymore yield as well an associated doubling of clock speed and performance per clock cycle every two years. Hence, the only way to speed-up nowadays computations is parallel processing. However, due to non-parallel overheads, parallel processing does not scale infinitely and the state-of-the-art High-Performance Computing (HPC) approach to achieve high processing speed are heterogeneousarchitectures, using accelerators such as Graphics Processing Units (GPUs) or Many Integrated Cores (MIC) in addition to general purpose Central Processing Units (CPUs). Existing HPC codes however need to be rewritten to exploit the accelerators. This talks give a brief overview on the heterogeneous, Modular Supercomputing Architecture (MSA) of the European H2020 research project DEEP-EST (Dynamical Exascale Entry Platform – Extreme Scale Technologies) and the resulting Software Engineering challenges.
No office hours of Prof. Grabowski on 26.10.2017
informatik.uni-goettingen.de or studienberatunginformatik.uni-goettingen.de.Gastvortrag: Softwareentwicklung bei Mahr
Later office hours of Prof. Grabowski on 19.10.2017
No office hours of Prof. Grabowski on 07., 14., and 21.09.2017
informatik.uni-goettingen.de or studienberatunginformatik.uni-goettingen.de.Comment article accepted by IEEE Transactions on Software Engineering
Paper accepted at Springer Empirical Software Engineering Journal
We are happy to announce that our paper "Addressing Problems with Replicability and Validity of Repository Mining Studies through a Smart Data Platform" was accepted for publication by Springer Empirical Software Engineering. Within this article, four major problems in the state of practice of software repository mining that threaten the external validity of research results, i.e., the generalizability of drawn conclussions, are discussed. This article is an extension to our MSR publication.
The pre-print of the article is available online at Springer. A free view-only version of the article is available here.
No office hours of Prof. Grabowski on 03.08.2017
informatik.uni-goettingen.de or studienberatunginformatik.uni-goettingen.de.Paper accepted for IEEE Transactions on Software Engineering
We are happy to announce that our paper A Comparative Study to Benchmark Cross-project Defect Prediction Approaches was accepted for publication by the IEEE Transactions on Software Engineering - the top journal for the field of software engineering.
Within this article, we compared 24 approaches for cross-project defect prediction proposed between 2008 and 2015. Most of these approaches were never replicated or compared to each other before. Our suprising findings show that very little actual progress was made in the state of the the art since 2009, if we considered multiple performance metrics and data sets.
The pre-print of the article is available in IEEE Xplore and thanks to the green open access policy of the IEEE on our homepage.