|Universidade da Coruña Rúa da Maestranza, 9 15001 A Coruña, Spain||
|Computer Architecture Group (GAC) Ramón Doallo Facultade de Informática, Campus de Elviña, s/n 15071 A Coruña, Spain|
The University of A Coruña was founded under the Galician Universities Law 11/1989 on July 20th, 1989. The Constituent Assembly passed the University Statutes on the February 4th, 1992 and these were published in the Official Bulletin (DOG) on September 17th, 1992. It is territorially situated in the cities of A Coruña and Ferrol, both located in Galicia, at the Noth-West of Spain. The university has around 19000 undergraduate students, 1100 graduate students, 1400 PhD students, and a teaching staff of 1450 members.
Research Group Background
The Computer Architecture Group (GAC, see http://gac.udc.es/english), at the Universidade da Coruña (Spain), has a long experience in the area of High Performance Computing (HPC), covering a wide range of topics such as parallel and distributed computing and cloud computing, high performance microprocessors, high speed communications, computer graphics and scientific visualization, mobile robotics and computer vision, and geographic information systems.
The GAC is composed of 15 (PhD) professors and 18 PhD students, and it has achieved R&D funding of 4.208.533€ in the last 5 years from 18 competitive public research projects, 33 contracts with institutions and companies, and 14 fellowships from competitive public calls of human resources, all of them with a PI from the GAC. In this period the group published 62 papers in international top ranking journals (listed in ISI-JCR 2012) and 97 in international conferences, 42 of them in the categories CORE A or B. Also, 9 PhD theses were completed within the group in this period in the context of a PhD program awarded by the Ministry of Education with the national “Excellence Mention”. Three of the new PhDs were transferred to research centers abroad and two of them to technological companies.
The GAC participates in several European projects such as the HiPEAC3 Network of Excellence (FP7 ICT-287759) and its previous editions (HiPEAC and HiPEAC2) and the recently finished Open European Network for High Performance Computing on Complex Environments (ComplexHPC, COST Action IC0805) that preceded the NESUS COST Action IC1305.
The group is strongly committed to technology transfer. This is reflected by the large number of software packages related to HPC it has developed on its own or in collaboration with other leading international research groups, many of which have been made publicly available under open source licenses. In the last 5 years the GAC registered six software products in the IP Registry and one US patent. The GAC has also given place to a spin-off to commercialize part of its HPC results.
Major Interest in Action Scientific Topics
The preceding experience and current interest of the GAC in this Action Scientific Topics focuses in the working groups ‘WG2: Programming models and runtimes’ and ‘WG3: Resilience of applications and runtime environments’. This way, regarding WG2, GAChas a long experience in the design and development of tools based on libraries, runtimes and compilers to enhance programmability of parallel systems. These tools have been oriented both to shared, distributed and hybrid memory systems based on regular processors and the novel accelerators that play a key role in nowadays large-scale complex systems. Another WG2 point of interest in which GAC has been working is in the automation of the adaption of codes to the characteristics of the underlying hardware in order to maximize the performance achieved with minimal user time costs as the hardware available evolves. As regards WG3, the group has been working in fault tolerance of parallel applications for more than 10 years with special focus on the field of checkpointing optimization. As a result of this work, CPPC (ComPiler for Portable Checkpointing, http://cppc.des.udc.es), an application-level checkpointing tool for long-running message-passing applications, was developed. Currently the GAC is exploring the possibility of implementing malleability of MPI applications as an extension to the CPPC tool, so that the final approach enables transparent reconfiguration of parallel applications. Malleable applications allow automatically adapting the parallel execution to the available processors, coping with hardware errors and improving the use of resources. This will have a direct effect on the energy consumption required for the execution of applications, resulting in both cost savings and a greener computing.