The performance of the online system of the PHENIX experiment in the RHIC run 5

PHENIX is one of two large experiments at Brookhaven National Laboratory\´s Relativistic Heavy Ion Collider (RHIC). Previously, the PHENIX experiment had completed a very successful run 4 in early 2004, where we had been able to quadruple the data rate that had been achieved before. Run 5 consisted of a heavy-ion part, where copper beams were accelerated, followed by a polarized proton-proton part. We were able to further expand the data rates to a maximum of about 600 MB/s. By fully commissioning the so-called multi-event buffering in the data acquisition front-end, we were able to decrease the deadtime of the data acquisition while at the same increasing the event rate that we could sustain. In addition, the handling of the data was more efficient, which led to a faster turnaround between data taking and the subsequent calibration and reconstruction steps. During the proton-proton part of run 5, we were able to overcome a serious problem with the available local computing resources at Brookhaven National Laboratory. Using Grid tools, we sent 270 TB of raw data to the PHENIX "CCJ" computing center in Japan, through the network