sPHENIX discussion of software

[pinkenburg <pinkenburg AT bnl.gov>]

    Hello everybody,
    
    This is a reminder that we will have a sPHENIX simulation and
    software meeting tomorrow September 10th 1:00PM ET. 
    
    The Agenda for this meeting is: https://indico.bnl.gov/event/24674/
    
    We have to talk seriously about coordinating resource usage. The
    free for all is coming to an end (or better a screeching halt). We
    kind of addressed the cpu usage with last weeks limit to 23k jobs
    only to encounter the disk as the next bottleneck. Just try to read
    from or write to /sphenix/tg/tg01 or /sphenix/user and you'll see
    that our current approach is just not workable. Each of them can do
    10+GB/s which isn't shabby but 20k condor jobs can overwhelm this
    easily, see the attached snapshots from today. Even lustre with its
    100-200GB/s capabilities doesn't stand a chance in the longer run
    (once we try to go over full datasets).
    
    We need to agree on priorities - what is absolutely essential
    (verification of our incoming data), essential (preparing our data
    for physics analysis, calibrations), important (analysis prep, pp
    sims), nice to have (Au+Au related analysis),...
    
    The place to set those are the TGs - one option would be that
    projects using a lot of our resources need to be endorsed by the
    respective TG convenors (or detector groups). I don't have a good
    idea where to set the bar for this (1k jobs, 10k,...?) but we need
    to do something that the investment of resources to run these jobs
    benefits sPHENIX at large.
    
    The other issue which has to be addressed are the simulations. I
    have been looking of what is being run right now and it's basically
    a waste of time. We run at different crossing angle(s), have likely
    different vertex distributions and definitely different luminosities
    which go then into our pileup sims. The primary calorimeter output
    is still the crude energy to adc conversion with some noise applied
    and a zero suppression nobody believes in and then there is the
    tracking (which is understandably in flux and really hard). A small
    group of people tackling this can make a real difference here.
    
    
    sPHENIX members should have write access, the access key is
    babar1008
    
    Don't forget the office hours at 3:00pm-3:30pm (Help will always be
    given at sPHENIX for those who ask for it). 
    
    https://app.gather.town/invite?token=jG1_BxRHToWd-FNKTitF
    
    Here are the zoom coordinates

Join ZoomGov Meeting
https://bnl.zoomgov.com/j/1609732193?pwd=K2tZVStqZTBxdm9zdVlFVDh1MFF3dz09

Meeting ID: 160 973 2193
Passcode: 404116
One tap mobile
+16692545252,,1609732193#,,,,,,0#,,404116# US (San Jose)
+16468287666,,1609732193#,,,,,,0#,,404116# US (New York)

Dial by your location
        +1 669 254 5252 US (San Jose)
        +1 646 828 7666 US (New York)
        +1 669 216 1590 US (San Jose)
        +1 551 285 1373 US
Meeting ID: 160 973 2193
Passcode: 404116
Find your local number: https://bnl.zoomgov.com/u/abPqekC4i

    Looking forward to talking to you,

Cheers,

    Chris
-- 
*************************************************************

Christopher H. Pinkenburg	;    pinkenburg AT bnl.gov
				;    http://www.phenix.bnl.gov/~pinkenbu

Brookhaven National Laboratory	;    phone: (631) 344-5692
Physics Department Bldg 510 C	;    fax:   (631) 344-3253
Upton, NY 11973-5000

*************************************************************

Attachment: gpfs.png
Description: PNG image

Attachment: tg.png
Description: PNG image