I am Senior Researcher at the Technical University of Berlin (Technische Universität Berlin) where I work with Prof. Ben Juurlink in the AES group. My current research interests include compiler technology, high-performance computing, software optimization and their intersection with various fields such as graphics, artificial intelligence, physics, chemistry and engineering.
From 2011 to 2015, I have been Post-Doctoral Researcher at the University of Innsbruck (Universität Innsbruck) where I worked with Prof. Thomas Fahringer in the DPS group. I contributed to the Insieme Compiler, heterogeneous computing with OpenCL, and on the libwater distributed runtime system. I was also involved in the DK-CIM programme at the Scientific Computing platform on multidisciplinary research approaches, together with leading scientists from various fields such as physics, chemists and engineers.
I received my Ph.D. from the University of Salerno (Università di Salerno) in March 2011, supervised by Prof. Vittorio Scarano. During my Ph.D., I was the recipient of a several grants and scholarships (HPC-Europa2, HPC-Europa++, DAAD, Cineca ISCRA) and I worked at the HLRS Supercomputing Center in Stuttgart and I was hosted by the University of Stuttgart (Universität Stuttgart) under the supervision of Prof. Carsten Dachsbacher and by Prof. Thomas Ertl.
News
- I am Guest Editor of Future Generation Computer System's Special Issue: On The Road to Exascale II: Advances in High Performance Computing and Simulations. Consider to submit your manuscript.
- I will be at ScalPerf 2018 in Bertinoro to present my work on autotuning
- I will be at Euro-Par 2018 in Turin to present our domain specific language for agent-based simulations and chairing the Application session
- on June 1st I will be at the University of Salerno to present my recent work on autotuning
- I am hiring! We are looking for either one post-doctoral researcher or PhD student to work with us on the CELERITY project.
- New paper in collaboration with CERN accepted at PDP 2018!
- I'll give a talk about autotuning with machine learning at the LLVM Berlin Meetup on Nov 30 in the Mozilla Berlin Community Space
- I am PI of the new DFG-funded project CELERITY on programming models, runtime system and modeling for HPC
- New paper on approximate computing accepted at CGO 2018!
- I will be speaker at the ARM Research Summit in Cambridge, UK
- I will moderate a panel on Programming Model for the Exascale Era at HPCS 2017 in Genoa, Italy
- I will be at SCOPES 2017 in Sankt Goar, Germany
- New paper about stencil autotuning with machine learning accepted at IPDPS 2017. I will present it on May 30, 2017 in Orlando, Florida
- New paper about PHP 7 static optimization accepted at CC 2017. I will present it on Febreuary 5th, 2017 in Austin, Texas
Projects
[Dec.2017 - 2020] Principal Investigator of the DFG project CELERITY: advanCed modELing for scalablE distRIbuted runTime sYstems, an international DACH project between TU Berlin and the University of Innsbruck. DFG funding 413.230 Euro, overall (DFG+FWF) 698.000 euro.
[2011 - 2013] Work Package coordinator of the project Automatic Portable Performance for Heterogeneous Multi-cores, Austrian Science Fund (FWF) TRP 220-N23, P.I. Prof. Thomas Fahringer.
Selected Publications (See publication list for the full list)
- Cosenza, Durillo, Ermon, and Juurlink. Autotuning Stencil Computations with Structural Ordinal Regression Learning. In IEEE International Parallel and Distributed Processing Symposium (IPDPS), pages 287-296, 2017.
(Acceptance rate: 22%) - Popov, Cosenza, Juurlink, and Stogov. Static Optimization in PHP 7. In Proceedings of the 26th International Conference on Compiler Construction (CC), pages 65-75, 2017. (Acceptance rate: 24.5)
This work has been included in the offical release of PHP 7.1 by Nikita Popov - Grasso, Pellegrini, Cosenza, and Fahringer. libwater: Heterogeneous distributed computing
made easy. In Proceedings of the 27th ACM International Conference on Supercomputing (ICS),
pages 161–172, 2013.
(Acceptance rate: 21%) Extended version in Journal of Parallel and Distributed Computing (JPDC), 74(12): 3228-3239, 2014 - Kofler, Grasso, Cosenza, and Fahringer. An automatic input-sensitive approach for heterogeneous task partitioning. In Proceedings of the 27th ACM International Conference on Supercomputing,
(ICS), pages 149-160.
(Acceptance rate: 21%)
Early version in ACM PPoPP 2013: 281-282 - Kofler, Cosenza, and Fahringer. Automatic Data Layout Optimizations for GPUs. In International European Conference on Parallel and Distributed Processing (Euro-Par), pages 263-274, 2015.
(Acceptance rate: 27%) - Hunger, Cosenza, Kimeswenger, and Fahringer. Random fields generation on the gpu
with the spectral turning bands method. In International European Conference on Parallel and Distributed Processing (Euro-Par), 2014.
(Acceptance rate: 25%, Best paper selection)
Extended version in the journal Concurrency and Computation: Practice and Experience (CCPE), 27(16): 4122-4136, 2015
Teaching (at TU Berlin)
- Compiler Design (WS 2017-18, WS 2016-17, WS 2015-16)
- Avanced Computer Architecture, lab (SoSe 2018, SoSe 2017, SoSe 2016)
- AES Seminars (WS 2015, WS 2016, WS 2017, WS 2018)
- Recent Advances in Computer Architectures (WS 2017-18, WS 2018-19)