evaluation of TpcRS (new TPX slow simu)

A)----- Au+Au events @  sqrt(s)=200 GeV, no pileup simulation

Events samples were generated by Wei-Ming using: sqrt(s)=200 GeV, Relativistic Quantum Molecular Dynamics (RQMD), geometry 2009

Events's sets, 20 events apiece:

minB : Impact parameter b goes from 0 to 15 fm 

peripheral (b > 13 fm); mid (9 fm < b < 11 fm), and   cent (b < 5 fm)

AuAu events, BFC w/ TpcRs
event set   avr # of good glob   CPU per eve   
 minB  770 tracks/eve  20 min  
peripheral  12 tracks/eve  44 sec  
mid  360 tracks/eve  11 min  
central  2400 tracks/eve  53 min  

The largest time consumers in BFC w/ TpcRS for minB data set of 20 events:

QA :INFO  - QAInfo:Chain            StBFChain::bfc                  Ast =23789.48        Cpu =23538.49
QA :INFO  - QAInfo:Maker          StTreeMaker::outputStream         Ast = 12.87( 0.1%) Cpu = 12.49( 0.1%)
QA :INFO  - QAInfo:Maker    StMcAnalysisMaker::StMcAnalysisMaker    Ast = 40.81( 0.2%) Cpu = 40.25( 0.2%)
QA :INFO  - QAInfo:Maker   StAssociationMaker::StAssociationMaker   Ast = 14.21( 0.1%) Cpu = 14.05( 0.1%)
...
    INFO  - QAInfo:Maker        StdEdxY2Maker::dEdxY2               Ast = 13.22( 0.1%) Cpu = 13.06( 0.1%)
QA :INFO  - QAInfo:Maker             StiMaker::Sti                  Ast =209.78( 0.9%) Cpu =207.62( 0.9%)
...
QA :INFO  - QAInfo:Maker     StTpcRTSHitMaker::tpc_hits             Ast =686.62( 2.9%) Cpu =679.61( 2.9%)
QA :INFO  - QAInfo:Maker       StMcEventMaker::StMcEventMaker       Ast =  8.77( 0.0%) Cpu =  8.72( 0.0%)
QA :INFO  - QAInfo:Maker         StTpcRSMaker::TpcRS                Ast =22781.93(95.8%) Cpu =22550.06(95.8%)

 

The following BFC chain was used:

 bfc.C(20,"MakeEvent,ITTF,NoSsdIt,NoSvtIt,Idst,VFMinuit,-EventQA,-dstout,tags,Tree,EvOut,analysis,dEdxY2,
IdTruth,useInTracker,-hitfilt,tpcDB,TpcHitMover,TpxClu,McAna,fzin,y2009,tpcrs,sdt20090428.141700",
"/star/u/wzhang/links/ksu/auau200/fzdFile/au200f.fzd")

 

B)----- U+U events @  sqrt(s)=200 GeV, no pileup simulation

Events samples were generated by We Kejun in IOPP using: sqrt(s)=200 GeV,  RQMD, geometry 2009, the deformation for U hasn't been considered, minB collisions. 

 

U+U  events, BFC w/ TpcRs
event set   avr # of good glob   CPU per eve   
 minB  707 tracks/eve  22 min  
       

 

 

C)---- P+P events @ sqrt(s)=500 GeV, includes pileup simulation

Estimation of time needed to simulate pp500 events w/ pileup. Based on T. Roser CAD projection we expect peak lumi as follows. 

Table 1  CPU per M-C pp500 events w/ pileup
year  test 1  test 2  2009   2011   2012  2013 2014
peak lumi (10^30/cm^/s) 4 20 85 169 250 338 417
# of minB inter. per bXing  0.01  0.05 0.21   0.42  0.63 0.85  1.0 
 # of good glob tracks  50-150  360-500  1.9K        13-15K
 CPU time per event (BFC w/ TpcRS) *)  3 min  11 min  42 min        3.5 hours
 CPU time per event (BFC w/ TRS) **)  0.27 min  0.9 min  3.2 min        0.26 hours

*) averaged over 5 events. **)  averaged over 100,20,10,10 events, respectively.

All timing was measured on rcas6001 where I got ~100% CPU per process, on the same Pythia events w/ partonic pt>20 GeV. Pileup were Pythia events w/ partonic pt>1 GeV. Events were generated with y2009 geometry as in dev on Nov. 1 2009.

This BFC chain was used to simulate 5% pileup BFC w/ TpcRS (new TPX sloe simu):

 root4star -q -b 'bfc.C(5,"MakeEvent,ITTF,NoSsdIt,NoSvtIt,Idst,VFPPVnoCTB,logger,-EventQA,-dstout,tags,
Tree,EvOut,analysis,dEdxY2,IdTruth,useInTracker,-hitfilt,tpcDB,TpcHitMover,TpxClu,McAna,fzin,y2009,tpcrs,
sdt20090428.141700,geant,geantout,beamLine,eemcDb,McEvOut,bigbig,emcY2,EEfs,bbcSim,ctf,CMuDST",
"../TpcRS-rev2/test_QCDprod.fzd; gfile B ../TpcRS-rev2/test_QCDminb.fzd; mode BTOF back 1; 
mode TPCE back 3761376; gback 470 470 0.05 106. 1.5; rndm 10 1200"
)' >&LogP2

 More details about pileup simu: Jan blog, Xin:PPV+BTOF, Jan:ppLMV


 Similar test were done using BFC w/ TRS (current TPC slow simu) , 1% pileup:

root4star -b -q bfc.C'(10,"trs -ssd y2008  Idst IAna l0 tpcI fcf -ftpc Tree logger ITTF Sti StiRnd 
 -IstIT NoSvtIt SvtCL,svtDb -SsdIt MakeEvent McEvent geant evout geantout IdTruth  bbcSim emcY2 
EEfs bigbig -dstout fzin -MiniMcMk McEvOut clearmem -ctbMatchVtx VFPPVnoCTB eemcDb beamLine",
"../TpcRS-rev2/test_QCDprod.fzd; gfile B ../TpcRS-rev2/test_QCDminb.fzd; mode BTOF back 1; 
mode TPCE back 3761376; gback 470 470 0.01 106. 1.5; rndm 10 1200"
)' >&LogP1

 

 

Groups: