data size and io performance report

11
Ilija Vukotic [email protected] Data size and Data size and IO performance IO performance report report ATLAS Software & Computing Workshop

Upload: sylvester-evans

Post on 04-Jan-2016

20 views

Category:

Documents


0 download

DESCRIPTION

Data size and IO performance report. ATLAS Software & Computing Workshop. Data size MC http://athena-infoioperformance.web.cern.ch/athena-infoIOperformance/MCsizes.html Streams http://athena-infoioperformance.web.cern.ch/athena-infoIOperformance/StreamsSizes.html - PowerPoint PPT Presentation

TRANSCRIPT

Page 1: Data size and  IO performance report

Ilija Vukotic [email protected]

Data size and Data size and IO performance reportIO performance report

ATLAS Software & Computing Workshop

Page 2: Data size and  IO performance report

• Data size

– MC http://athena-infoioperformance.web.cern.ch/athena-infoIOperformance/MCsizes.html

– Streams http://athena-infoioperformance.web.cern.ch/athena-infoIOperformance/StreamsSizes.html

– Categories http://athena-infoioperformance.web.cern.ch/athena-infoIOperformance/T0streams.html

• IO performance

• Possibilities for improvement

20/04/23 Ilija Vukotic 2

Page 3: Data size and  IO performance report

MC event size

20/04/23 Ilija Vukotic 3

Page 4: Data size and  IO performance report

MC event size

20/04/23 Ilija Vukotic 4

Note: trigger size is not realistic as used trigger menu was not realistic. Problem rectified with addition of new aliases:Physics_default -> Physics_pp_v1MC_loose_default -> MC_pp_v1_loose_mc_prescaleMC_tight_default -> MC_pp_v1_tight_Mc_prescale (see bug report #74712).

Size AOD Size ESD

RecoTrf 572 +27.17%* 1880 +33.09%

stdSim 381 +11.18% 1620 +24.03%

fullSim 381 +11.18% 1620 +24.03%

physSim 320 +11.96% 1520 +24.79%

*Compared to 22 Aug. 2010

Page 5: Data size and  IO performance report

Streams sizes

20/04/23 Ilija Vukotic 5

Only runs with 5k+ eventsand good LBs

Shows:•AODs,ESDs and all DESDs•Events/total size/ev. size

Page 6: Data size and  IO performance report

Streams sizes

20/04/23 Ilija Vukotic 6

Stream Size[GB]

ESD 1023834

AOD 92319

DESDM_EGAMMA 55288

DESD_SGLEL 44191

DESD_SGLMU 39375

DESDM_TRACK 26548

DESD_MBIAS 16841

DESD_CALJET 12370

DESDM_MET 5108

DESD_PHOJET 4281

DESD_SGLMUDESD_SGLEL

DESDM_EGAMMAAODESD

Page 7: Data size and  IO performance report

20/04/23 Ilija Vukotic 7

Categories

Page 8: Data size and  IO performance report

Categories

20/04/23 Ilija Vukotic 8

EgammaL1CaloMuonsJetTauETMiss

https://spreadsheets.google.com/ccc?key=0AiPvgbRljNCodG5IY1BseHRHa2R5WjZoMGRnLXp0dHc&hl=en

Details and list of collections in categories can be found at:

Page 9: Data size and  IO performance report

20/04/23 Ilija Vukotic 9

IO performance

Slow

Bad CF

KEEP IN MINDIn real data TrackCollections are even larger !

Page 10: Data size and  IO performance report

20/04/23 Ilija Vukotic 10

IO performance

• Do we really need all of these?

• No way to say if object is ever used.

• What is so terrible about having a name attached to each object we store?

Page 11: Data size and  IO performance report

Possibilities for improvement

• TrackCollection/TrackParticle container

– Maybe 50kb in size and factor 2 in time

– Few MB in memory

• CaloShowerContainer

– Probably can go down in size

• Calo(Topo)TowerContainer

– Gets recreated at read time

– Must be possible to optimize further

• TRT_Drift_Circle_Container

– Must be faster and smaller

20/04/23 Ilija Vukotic 11

My estimate:

In total between 50 and 200kB/ev can be shaved in converters

Should be possible >100ms/ev (currently at 480 ms/ev)

Other options:

Object removal.

Cuts tuning