Massively parallel procecessing in high energy physics the CERN-MPPC project

Abstract

The next generation of high luminosity hadronic colliders (LHC and SSC) requires novel detectors, both highly time-sensitive and selective. The high selectivity will require devices which must allow to take on-line decisions (at ∼ 100 kHz) of a complexity comparable to what is today done in off-line analysis. Such decisions will be based on local or global fine-grain data (many Megabytes), and will require the on-line execution of algorithms in extremely fast computer-like devices. In order to solve this difficult task a Research and Development programme ”The MPPC Project” has been launched at CERN which includes ASPEX (UK), CEA/CEN-Saclay (F), CERN (CH), EPFL-Lausanne (CH),CNRS/IN2P3-LAL-Orsay (F), CRIP/KFKI Budapest (H), University of Brunel (UK), University of Geneva (CH) and Thomson-TMS (F). The MPPC collaboration concentrates its effort on the development of machines based on massive parallelism with thousands of integrated processing elements, arranged in string: the ASP (Associative String Processor), a new flexible and scalable architecture endowed with an intelligent and powerful communication network. The present goal of the project is to evaluate the ASP architecture on real time test benches. For this purpose, four 16K-APE (Associative Processing Element) machines are under construction. The basic element is a hybrid module (1K APE) based on an existing and fully tested 64 APE chip. The four 16K could be combined into a single 64K-APE machine, specially suited for on-line image processing. Seven applications are under detailed studies within the collaboration: three for LHC, one for SSC, two for fixed target high energy physics at CERN and one for HDTV. Preliminary results are presented. They show that the objectives should be reached with the use of the ASP architecture.