The recent rapid improvements in performance of commercially available digital signal processors (DSP) and data-driven array processors make their efficient integration into the front-end trigger and data acquisition electronics of future detectors for high energy physics (HEP) attactive. These devices provide an alternative to VLSI circuits such as ASICs (application specific integrated circuit) for this special purpose application. A parallel-processing system based on data-driven array processors, digital signal processors (DSP) and Transputers for trigger decision, data acquisition and compaction is described. The system is modular and suitable for all calorimeter sizes including the ones proposed for the LHC (large hadron collider) and SSC (superconducting supercollider) experiments. The aim is to give full programmability to the trigger decisions and data compaction, over the entire calorimeter at the single-channel granularity level, with no boundary limitation. The fast cluster finding system is based on two cooperating levels, the higher one utilizing FDPP (fast digital parallel processing modules), and the lower one, the new data-controlled array processor for video signal processing, DataWave. The proposed architecture assigns one DataWave processing element (PE) to each calorimeter channel and one FDPP module for every 256 DataWave PE. These numbers have been chosen for optimum performance but may be changed. Fast inter-processor communication within the DataWave array overcomes the problem of overlapping areas, and effectively provides a continuous array of PEs. Local maxima, along with their total energy, are found in a very short time by the data-driven array processors. These are easily selected and transfered, with the data of their region of interest, to a higher-level DSP array-processor system even if part of the data falls outside the region originally attributed to a particular FDPP.
Read full abstract