INtelligent, Fast, Interconnected and Efficient devices for Frontier Exploitation in Research and Industry
FOR ASTROPHYSICS, MEDICAL PHYSICS AND PARTICLE PHYSICS
This topic is evolving impressively fast over the last years. Indeed the Exascale and Big Data fields are offering a number of advanced new technologies on the firmware and hardware related aspects.
Because the highly challenging applications the network is addressing in the Astrophysics (especially SKA) in HEP with the High Luminosity LHC case and in new Medical Imaging techniques, WP4 is a unique interdisciplinary Forum to explore, test and further develop these new technologies. It a very highly interdisciplinary WP; all the fields of application in INFIERI have a strong and common need on such high performing computing (HPC) systems. INFIERI allows to study in a collaborative approach the best technical solutions with very fruitful exchanges between partners and an instrumental cooperation with industrial partners even not part of this network.
The INFIERI project initiated a new series of International Summer Schools (see WP7) on “Intelligent Signal Processing for Frontier Research and Industry” with lectures and Lab sessions given by worldwide experts in high performance and Massive Parallel computing technologies as well as from the Industrial World (IBM, INTEL, GPGPU etc.).
And a new series of Workshop on “High Performance System Processing”, HPSP2014, has been launched similarly this last October.
The new processors we are considering are new products on the shell, namely: novel FPGAs produced by XILINX and ALTERA, new developments on GPGPU’s (nVIDIA), new INTEL Xeon Phi and some other new processors from other industrial firms. A number of related firmware tools are being studied as well with benefitial expertise from some INFIERI partners.
In parallel, a full custom solution based on a new version of the Associative Memory technology originally developed in HEP for the CDF experiment is pursued by FNAL and INFN for application to the track triggering at the LHC.
This is a very active area in INFIERI. During the first half of the project, it took us some time to review and to decide the best ways to address and further develop this topic, in a very innovative and efficient way, establishing new fruitful contacts with industrial firms or experts not part of the project. This is why some partners decided to delay consequently the hiring of the young fellows and indeed to favour co-funded fellowships (ex: the newly appointed ESR1 (CNRS)).
We are now reviewing the main achievements in WP4 for the various applications covered by INFIERI, including the modeling activity part in 4.4.