(Summers, Bauer, Sanders, student)

One critical path item for BABAR is the data format. This needs to be implemented and agreed upon so that data can uniformly be recorded from the five detector systems, Monte Carlo events, and reconstructed DST data types.

In BaBar Note 129, Steve Bracker has specified a framework for highly compressed bit-packed data. Compressed data leads to fewer tapes, fewer tape mounts, lower Federal Express bills for tape shipments, more events per disk for on-line access, and faster local network transfers of data. R&D needs to be done to develop the code for this method of storing data in consultation with the detector groups and with those doing Monte Carlo work. Some code has already been written in C++. FORTRAN is also posssible, but it lacks some nice features such as function in-lining and unsigned integers. A comparison then needs to be made between bit packed data and other possible formats (e.g. ZEBRA banks). The tests would naturally include:

1) compression factor,

2) speed of data packing and unpacking,

3) how easy the code is to read and understand,

4) ease of adding and subtracting data types,

5) flexibility in accepting integer, floating point, and other kinds of data, and

6) ability to run with minimal changes on any computer PEP-II might use

Krish Gounder was moved to SLAC and was working on the data format. He also intended to work on event-parallel processing software for the IBM workstation farm which was being put into place at SLAC. He had experience in this area as described in BABAR Note 128.

SIMULATION R&D

Doug Wright and Terry Wenaus of LBL had accounts with a Gigabyte of disk space on the Mississippi farm. They are did BBSIM calculations. Lucien Cremaldi and Don Summers are working with ASLUND in doing tracking resolution studies.