Activities‎ > ‎

2011 - present

Computational Scientist and GPU Developer,
and PhD student in Computer Science (2011 - present)




Educational background

Jun 2012 - present

PhD candidate in Computer Science at Université de Rennes 1


Research activity at the INRIA/IRISA/CNRS ALF group within the Matisse doctoral school at Université de Rennes 1 with focus on the study of heuristics for automatic compiler optimisation of applications for performance portability across many-core architectures. Supervised by Prof. François Bodin and co-supervised by Mr. Gilles Civario and Dr. Erven Rohou.
Mar 2013 - present

Visiting researcher at the University of Delaware


Collaboration with Prof. John Cavazos in the study of heuristics and algorithms for automatic optimisation of parallel applications, aiming for performance portability across many-core architectures. This collaboration is being established with support from ICHEC, the AutoTune project and the HiPEAC Collaboration Grant.

Outcomes: organised a workshop

Professional Experience at Ireland's Supercomputing Centre

Jan 2011 - present

Computational Scientist at ICHEC


The Irish Centre for High-End Computing (ICHEC) is Ireland's national supercomputing centre. It is mainly responsible for the "research enablement" of the Irish community regarding the use of high-performance computing. The research enablement involves supporting and developing the academic and industrial communities in a whole, by means of research collaborations with European institutions; optimisation and porting of HPC applications; technology transfer and technical consultation, offering advice and expertise in HPC architectures; general user support, dealing with helpdesk issues; and education and training, e.g. organising and delivering training courses and modules for the industry and academia, supervising students and coordinating workshops. These activities are supported by a HPC infrastructure ICHEC maintains, which includes two compute clusters that sum up more than 4,000 processor cores and 10 TB RAM. More details can be found on the "About Me" page at ICHEC website.

Outcomes: bug fixes for OpenMX (patches 3.5.1 and 3.5.2) and Abinit 6.12.3 (fixed in Abinit 7.6.1)
Jan 2011 - present
GPU Developer at ICHEC

As part of ICHEC's GPU Activity, I work on the development, analysis, optimisation and porting of systems to many-core architectures, providing support to researchers and consultancy services to industrial clients. My past projects dealt with a variety of areas of science, including cryptography, financial simulation, geophysical analysis and molecular dynamics. I also worked on the European PRACE project where I was directly involved in enabling petascale computing on GPUs. I am a co-author of ICHEC's NVIDIA-certified CUDA training programme, which I have also delivered to industrial clients. The success of ICHEC's GPU activity were recognised by NVIDIA, for granting the status of CUDA Research Center in mid-2010 and naming the centre an "enabling partner" in GPU Computing early 2011; and by CAPS entreprise, for granting the status of HMPP Center of Competence late 2011.

Outcomes: a proceedings paper, a session at GTC 2012, an international award and a PRACE-2IP white paper and a PRACE-2IP deliverable
Jul 2011 - present
Project Manager at ICHEC

As a PRINCE2 accredited practitioner, managed projects of several team members and diverse backgrounds. Projects ranged from software development and deployment to organization of events and refurbishment of the institute’s procedures and strategies.

Outcomes: a certification and some projects managed, such as the Culture Night with DIAS and Met Éireann at the Dunsink Observatory.
Sep 2011 - Jun 2012
Lecturer on the NUI Galway Physics Structured PhD Programme

Formal, accredited teaching of graduate students enrolled in a structured PhD programme at Irish universities (NUI Galway, Tyndall Institute, UCD, DCU). Two modules (PH504: HPC and Parallel Programming, and PH502: Scientific Programming Concepts) were delivered so far.

Outcomes: some lectures delivered are Computer & HPC ArchitectureIntroduction to GPU ComputingIntroduction to Algorithms and Scripting with Python.
Oct 2011 - present
Researcher and Work Package leader on the AutoTune Project

Study about heuristics for automatic compiler tuning of applications for
multi- and 
many-core architectures, with the development of a plugin-driven distributed framework for profiling and automatically tuning runtime and energy performance of serial and parallel C/C++/Fortran codes. The AutoTune project is a European-funded FP7 project whose consortium includes European universities, supercomputing centers and multinational IT companies.

Outcomes: two posters (one at GTC 2012 and another at HiPEAC 2013), three deliverables (D4.1, D5.1 and D5.2) and a proceedings paper at PARA 2012
Oct 2011 - Apr 2012
Developer on the phiGEMM project

The
phiGEMM library is a numerical library that implements Level 3 BLAS GEMM routines to perform matrix-matrix multiplies taking advantage of heterogeneous multi-core compute nodes (or workstations) equipped with NVIDIA GPU cards. PhiGEMM splits large matrices to concurrently utilise CPUs and GPUs, achieving a sensible performance speedup when compared to CPU-only executions. Matrix-matrix multiplies are important for many scientific and engineering applications, which rely on the performance of mathematical libraries to deliver results in the least time possible due to their large problem size and time constraints.
Sep 2012
Lecturer and Organiser of the "Software Carpentry in the Cloud Boot Camp"

The "Software Carpentry in the Cloud Boot Camp" , organised in association with Amazon Web Services (AWS), covered the fundamentals of software carpentry, such as the tools and techniques that enhance the design, development, execution and testing of software, with particular reference to dealing with high-throughput and time-critical computing problems. Emphasis was placed on software deployment in the cloud using AWS to improve software scalability. The programme consisted of talks and hands-on tutorials on software carpentry tools and techniques with a view to deployment in the cloud using AWS. The boot camp finished with a "Grand Challenge" competition, where a problem was announced and each team had to devise, develop and deploy its solution using cloud computing resources.

Outcomes: ICHEC's press release, the event blog and presentation about Software Testing on the Cloud
Jan 2013
Exhibitor at the BT Young Scientist & Technology Exhibition (BTYSTE 2013)

At BTYSTE 2013, we showcased all things supercomputing to
primary and secondary level students in Ireland. At the stand in the World of Science & Technology Hall we displayed a fully configured live mini "supercomputer" built from the relatively cheap Raspberry Pi boards, 3D scientific visualisations of climate and weather data, presented the latest in accelerators and coprocessors (NVIDIA K20 and Intel Xeon Phi) and ran introductory sessions into HPC. Over 200 schools all across Ireland visited the ICHEC stand to understand the utility of supercomputing, including Donegal, Cork, Kerry, Galway, Dublin, and Antrim.

Outcomes: the event's promotional page, an ICHEC press release, a PRACE press release and a news item in the PRACE Annual Report 2012

Conference Service


Memberships

Nov 2011 - present

Association for Computing Machinery (ACM) 


Professional member of the world's largest and most prestigious scientific
and educational computing society. SIGs: SIGARCH.
Aug 2012 - present

Institute of Electrical and Electronics Engineers (IEEE)


The world's largest professional association for
the advancement of technology.