The ACP multiprocessor system at Fermilab

Study of theηc(11Armstrong1, Bettoni2, Bharadwaj3 et al. 1995Phys. Rev. D33141View full textAdd to dashboardCite

“…The level-1 trigger logic required a total time of about 350 ns, and the event readout from the CAMAC front end took about 100 psec, resulting in a 90% live time. The software trigger was performed by 26 Fermilab ACP [21] processors, whose main tasks were to convert the recorded pulse heights to energies, to reconstruct shower positions and energies, and to calculate the invariant mass of all candidate photon pairs. The clustering algorithm used in the ACP was a simplified version of the one used in the off-line analysis (to be described later).…”

The Parallel Processing Compute Server

“…A good 654 review of these machines can be found in [6]. Similar farms based on machine architectures other than System/370 have been built at other laboratories [15][16][17]. These farms generally do not have operating systems running on the satellite nodes, but rather very primitive kernels sufficient to execute FORTRAN programs.…”

“…The matrix switch provides a high-speed interconnection network that allows any-to-any processor communication. This permits the PPCS to be used for more general parallel applications, such as lattice gauge calculations [16]. A consistent approach to System/370 architecture and system software (Parallel VM and Distributed VS FORTRAN) is used throughout the system and provides a large set of functions that operate on the I-host as well as on the satellite processors.…”