Event reconstruction in high-energy physics

Cooperative Processes Software (CPS), a parallel programming toolkit developed at Fermilab for processing CPU-intensive experimental data in the area of high-energy physics, is described. CPS allows physicists to solve computationally intensive problems on a collection of reduced-instruction-set-computer (RISC) workstations that work in parallel to provide solutions within a reasonable time frame. This mode of parallel processing provides a cost effective and easily scalable solution to CPU-bound computational problems that can be classified as large grained and loosely coupled. The performance of CPS on a typical experiment of Fermilab, which has 1000 to 2000 8 mm tapes of raw event data that needs to be reconstructed, is discussed.<
>