The following is a summary of a software micro-workshop at BNL (KH + FV + JHLee).
STAF
1. BRAHMS will probabally use STAF for at least the 1st stage of analysis,
ie raw data to processed data.
a) STAF is a real analysis framework.
b) Has the ability to write out multiple tables and read them back in
keeping the correlation between events.
c) PHENIX and STAR are using it -> major support + probable RCF support
2. Flemming generated two STAF modules:
a) Read GBRAHMS C-stream files and generate hits tables
b) Use these hits tables as input into a prototype TOF STAF module and
perform operations on these tables and generate new "calibrated"
tables.
c) write these tables out as STAF XDR files.
3. Short term plans
a) Clean up STAF example modules described in (2.)
b) Set up STAF access in a uniform way on the RCF to make it workable
for everyone.
c) Generate a recommendation for the module interface.
d) Interface c-stream files with dio instead of using the reading routine
as an analysis module as is currently done.
GBRAHMS
1. Generated files for the following angles and field settings of the
forward spectrometer.
angle field current location file size
2.3 1/2 /brahms/mds/gbrahms_output/bench1/bench2.cdat 5.69 Mb
2.3 1/4 /brahms/mds/gbrahms_output/bench1/bench3.cdat 4.63 Mb
6.0 1/4 /brahms/mds/gbrahms_output/bench1/bench4.cdat 2.54 Mb
SONATA++
1. Program status
a) Digitize, Decode and local tracking routines written and for the
most part debugged
b) Can compile on most platforms at the RCF using gcc
c) Is fairly stable
2. Used the program to read GEANT c-stream files for various angle and field
settings. This was done for the purpose of refining estimates for
the BRAHMS CPU requirements to supply to the Requirements Task Force
for the RCF.
3. Benchmarking Results
angle field time per event
2.3 1/2 4.4 sec
2.3 1/4 5.4 sec
6.0 1/4 1.6 sec
It was noted that many events in the 2.3 setting went very fast, but some
of the more difficult events to deal with took a very long time. The
numbers above reflect a condition inserted into the program to abort
the event every time it found more than 1000 local tracks.
4. Short term plans
a) Continue stabilization procedure
b) Construct the TSonata object to inherit from TObject in ROOT so
that it can be used with ROOT as a real simulation package.
This archive was generated by hypermail 2b29 : Tue Feb 01 2000 - 20:35:19 EST