The following is provided for your comment. The FFREAD parameters have
been
changed slightly, as well as revised estimates for running times and
files sizes (adjusted
for 1.6 GeV running and other effects).
- Will
-----------------------------------------------------------------------------
The following summarizes the current proposal for the first long GSIM
runs.
Proposal version 5
We have a distributed farm consisting of some fraction of time from the
following cpu's: (managers/contacts)
MIT - 10 DEC Alphas, 500 MHz (David Rowntree/Jianguo Zhao)
UVA - 13 Linux/Intel 300 MHz (Cole Smith)
FSU - 8 Linux/Intel, more to come (Larry Dennis)
OU - 1 Linux/Intel, (+Linux/Alpha's, others) (John O'Donnell/Allena
Opper)
ODU - 2 DEC Alphas, 500 MHz (Junho Yun/Andi Klein)
3 Linux/Intel 300 MHz
Calculate 3 types of events:
1) elastic scattering (10% of available farm)
2) CELEG with resonances+elastics (60% of available farm)
3) e-p-pi0 events from AO (30% of available farm)
1.6 GeV, 60% magnetic field
Plan to run for 3-4 weeks, re-evaluate as we go along.
Analyze the runs with the 'Dec97' reconstruction program at JLAB.
File sizes: whatever you get from 24 hours of running, at first.
E.g., Linux/Intel @ 300 MHz, expect 0.5 sec/event & 1.0 kbyte/event.
=> ~ 170 Mbytes, ~170,000 events
For DEC Alphas, expect ~0.10 sec/event, => ~ 860 Mbytes, ~860,000 events
(These numbers assume a 100% dedicated machine.)
After we gain some experience, the runs could go longer; because of the
2
Gbyte file limit of unix, the Linux runs are limited to ~11 days and the
DEC Alpha runs are limited to ~2.3 days. (We may eventually need a
multiple
file output option from GSIM, as the DAQ has.)
The above numbers depend GREATLY on the FFREAD and command line
parameters.
FFREAD options to be used:
AUTO 1
CUTS 5.e-3 5.e-3 5.e-3 5.e-3 5.e-3
CCCUTS 1.e-3 1.e-3 1.e-3 1.e-3 1.e-3
DCCUTS 1.e-4 1.e-4 1.e-4 1.e-4 1.e-4
ECCUTS 5.e-4 5.e-4 5.e-4 5.e-4 5.e-4
SCCUTS 1.e-4 1.e-4 1.e-4 1.e-4 1.e-4
MAGSCALE 0.5829 0.75
NOGEOM 'ST'
NOSEC 'OTHE'
STOP
Typical startup script, with parameters:
../bin/Linux/gsim \
-ffread longrun1.ffread \
-bosout longrun1.evt \
-nomcdata \
-mcin 2.4gev_p_tgt_1.evt \
-trig 100 \
>& longrun1.out &
Version of GSIM to be used: release-1-16.
ISSUE: to run with secondaries turned off in volume OTHE or not? We're
trying to evaluate sample files. With OTHE turned on, we get more junk
(which looks like the real data), but it takes twice as long to run,
and the file size is twice as big. The current proposal is to leave it
turned off, but this is clearly worth discussing.
Any comments are welcome.