The HyperCP data acquisition system is the fastest in the world, with a trigger rate of up to 80 kHz and a throughput to tape of 13 MB/s. Simplicity and parallelism are its hallmarks. The average dead time per event was 3 microseconds, and the average event size during standard running was 550 bytes. With the trigger rate listed above, HyperCP produced data at 44 MB/s during the spill for a continuous rate to tape of 13 MB/s. Tests with improved DAQ software made towards the end of the run gave a maximum sustainable data rate to tape of 17 MB/s.

The figure shows the structure of the data-acquisition system. Information from the detectors was digitized and sparsified by two front-end systems: one based on the Nevis protocol and used to read out the wire chambers, latches and calorimeter ADC's, the other custom system designed by the Taiwan group and used to read out the muon chambers. The data were then transmitted via optical fibers to the VDAS (video data-acquisition system) spill buffers in the control room. Five event-building systems operating in parallel accessed the VDAS to assemble events and record them on tape. Each event-building system was housed in a single VME crate and was composed of three MVME167 processors, five Event Buffer Interfaces (to provide access to the VDAS), and three Ciprico RF3573 SCSI host adapters which each control three Exabyte 8505 tape drives. Control of the system was provided via Ethernet links to each `Boss' MVME167 processor and to two booting and monitoring processors (not shown in the diagram) located in the Electronics Hall.