Opened simulation requests 20090304

Updated on Wed, 2009-03-04 14:53. Originally created by jeromel on 2009-03-04 14:50.

Table summary - simulations in priority order.

PWG	Simulation request	Number of events	Required space (TB)		CPU time
PWG	Simulation request	Number of events	Geant	reco	Hours 1 CPU	Days for 100 CPU
Bulk	p+p 200GeV, pythia with year 2007 geometry	4.20 M	0.6	1.1	26250	10.9
Bulk	p+p 200GeV, pythia with year 2008 geometry	8.40 M	1.2	2.2	52500	21.88
Lfs	Au+Au 200GeV, Hijing minbias, geometry y2007g	0.15 M	0.5	1.4	12770	5.32
Lfs	Cu+Cu 200GeV, hijing minbias, geometry y2005g	0.20 M	0.25	0.42	5080	2.11
Spin	p+p 500 GeV, pythia with year 2009 geometry	2.30 M	0.6	1.2	27000	11.25
Total		-	3.15	6.32	-	51.46

Our current BNL resources should accomodate for 100 CPU slots as showed below:

The CAS nodes are all saturated.

QM starts April 30th - results should be ready 2 weeks prior which gives another 10 days to process data.

Conclusion: at 51.46 days on 100 CPU slots, we will not make it.

Plan:

Produce the 10.9 days sample on Amazon EC2
Produce the rest of 40.56 days on 300 CPUs instead, reducing the time to only 13.52 days
Production needs to start NOW ; resources cannot be taken by embeding

The STAR experiment