CMS is planning to reduce the expected 40 MHz interaction rate to 100 kHz by the first level trigger and by a further factor of 1000 with a large multiprocessor farm leading to a maximum rate to tape of 100 Hz. This data reduction will require use of a powerful data acquisition system capable of collecting the data from 1000 front-end data sources into a processor farm of 1000 processors. A distinct feature of the CMS DAQ and Higher-level Trigger system is the absence of a dedicated Level-2 hardware trigger. Instead, the processor farm is expected to deliver the equivalent of a Level-2 reduction rate by running dedicated physics algorithms on a subset of the detector data. A successful Level-2 decision will be followed by the transfer of the rest of the event data into the Farm for a final Level-3 decision. It is considered that this approach will give much greater flexibility in terms of choice of algorithm than a more classical Level-2 decision based on special purpose processors for specific triggers.
CERN currently has a large R&Deffort in various generic directions. CMS has not yet decided on specific technologies or readout architectures. There is a very good intellectual opportunity for the US groups to play a major role in the selection of technologies and the design of this readout system. The US groups presently working on this system are Fermilab, Iowa State University, MIT and UCSD.
Our contribution to CMS DAQ/High-Level Triggers so far has been in the design of a prototype for the front-end memories. We currently plan to participate in the design and simulation of the full system. We will investigate, in collaboration with CERN, various switching technologies. In addition, we will develop fast Level-2 software algorithms to demonstrate the flexibility of the proposed system and determine basic throughput requirements.
The rapid development and reduction in price of processor technologies argues for a late procurement of the processor farm. It is expected that US industry will continue to be the market leader in this field at the end the century. It is therefore proposed that the US groups should supply the processor farm to CMS and that this would constitute part of the US commitment to the common fund of the experiment.
We also plan to participate in the overall CMS computing effort. Valuable contributions can be made in the areas of system architecture; detector-specific algorithms, particularly for pattern recognition, for all subsystems; and global algorithms for track matching and gamma/electron identification. Much work has been done in modeling neutron and gamma backgrounds by US groups (e.g. LANL), and comparison with CMS efforts would be quite valuable.
There are also issues of World Wide Web (WWW) networking, teleconferencing, transfer of CAD drawings, and information distribution in general. The US CMS groups will contribute to the development and maintenance of a distributed CMS computing environment.
It may also become possible to establish a remote control room in the US, such that US physicists can stand shifts within the US. An ``existence proof'' is the remote operation of the Sloan Digital Sky Survey from Fermilab. Clearly, all control rooms are remote; it is simply a question of cable length. The possibility of remote operation would enhance the capabilities of US groups.
List of Interested Groups: