Progress towards a streaming DAQ for the ePIC collaboration at the EIC

The Electron Ion Collider (EIC) will be a unique machine colliding electrons with polarized protons and nuclei, constructed at Brookhaven National Laboratory (BNL) by a collaboration between the Thomas Jefferson National Laboratory (JLAB) and BNL. The EIC allows studies of multi-dimensional tomographic images of protons and nuclei, as well as collective effects of gluons in nuclei. It will be completed in the early 2030s. The "ePIC" collaboration was formed to design and construct the first general purpose detector to be ready at the beginning of the operation of the EIC to be sited at the IP6 interaction region of the RHIC/EIC accelerator complex. The physics calls for a 4π general purpose detector, plus far-forward and far-backward detectors highly integrated with the accelerator beam lines. More than 20 detector sub-systems will produce raw data rates in the order of 100 Tbps. The strategy of the ePIC DAQ is to reduce these raw data rates to a more manageable rate "to tape" around 100 Gbps, utilizing streaming readout technologies. The streaming readout concepts to be deployed at ePICs foresee to digitize all collision data without the use of an external trigger, aggressively zero suppress the data in "Front End boards" (FEB, containing the digitizing ASICs) and FPGA based Readout Boards (RDO), resulting in zero or very low deadtime. Data will be further reduced in several stages of the DAQ utilizing artificial intelligence and machine learning in programmable hardware serving as online data filters between the FEB and the data storage. Synchronization between the different detector sub-systems is facilitated by distributing the beam clock as well as accelerator information via a Global Timing Unit (GTU), which interfaces both to the collider, as well as run control and the DAQ hardware. The GTU is also the source for the distribution of a very low jitter (~10ps) clock for use in detector sub-systems requiring a precision clock such as the Time-of-Flight detector. In this talk I will describe the details of the ePIC DAQ proposal, including preliminary work on prototyping various aspects of the planned implementation.