The SuperB asymmetric energy e+e- collider and detector [1] to be built at the newly founded Nicola Cabibbo Lab [2] will provide a uniquely sensitive probe of New Physics in the flavor sector of the Standard Model. Studying minute effects in the heavy quark and heavy lepton sectors requires a data sample of 75 ab-1 and a peak luminosity of 1036 cm-2s-1. These parameters require a substantial growth in computing requirements and performances: we roughly estimate that in few years of operations we will have to cope with near half EB of raw data and that the CPU required for processing will be close to 6000 KHep-Spec06 per year. The SuperB collaboration is thus investigating the advantages of new CPU architectures (multi and many cores, now largely available on the market), with the aim to be able to treat this amount of data both efficiently and within reasonable amounts of time. At the same time the collaboration is analyzing the current software, in large part inherited from previous experiments (mainly BaBar), to understand the underlying parallelism level, how to exploit it and to find the better mapping to emergent hardware architectures. In this work we first present the measurements done on the analysis and simulation software, then the Framework architecture we are designing. We complete the presentation with a description of our Framework prototype and some preliminary performance measurements. © 2012 IEEE.
A parallel framework for the SuperB super flavor factory
CORVO, Marco;FELLA, Armando;LUPPI, Eleonora;MANZALI, Matteo;TOMASSETTI, Luca
2012
Abstract
The SuperB asymmetric energy e+e- collider and detector [1] to be built at the newly founded Nicola Cabibbo Lab [2] will provide a uniquely sensitive probe of New Physics in the flavor sector of the Standard Model. Studying minute effects in the heavy quark and heavy lepton sectors requires a data sample of 75 ab-1 and a peak luminosity of 1036 cm-2s-1. These parameters require a substantial growth in computing requirements and performances: we roughly estimate that in few years of operations we will have to cope with near half EB of raw data and that the CPU required for processing will be close to 6000 KHep-Spec06 per year. The SuperB collaboration is thus investigating the advantages of new CPU architectures (multi and many cores, now largely available on the market), with the aim to be able to treat this amount of data both efficiently and within reasonable amounts of time. At the same time the collaboration is analyzing the current software, in large part inherited from previous experiments (mainly BaBar), to understand the underlying parallelism level, how to exploit it and to find the better mapping to emergent hardware architectures. In this work we first present the measurements done on the analysis and simulation software, then the Framework architecture we are designing. We complete the presentation with a description of our Framework prototype and some preliminary performance measurements. © 2012 IEEE.I documenti in SFERA sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
