SLAC PEP-II
BABAR
SLAC<->RAL
Babar logo CM2 logo
HEPIC E,S & H Databases PDG HEP preprints
Organization Detector Computing Physics Documentation
Personnel Glossary Sitemap Search Hypernews

Rates and Performances

Last Updated: 8 April 2004
Daniele del Re, UC San Diego

Skim Rates

Detailed documentation of skims in release 14 can be found in this link.

A summary of rates (and stability over time) is here.

A 7% rule is applied: if rate is below 7% threshold the skim is a deep-copy skim, if above it is a pointer skim, as specified in this version of FilterTools/defineMiniSkims.tcl.

Sizes

  1. plain new mini size

    size(Kb*ev-1)
    Mini Micro
    generic B+-B- ~16.5 ~7.6
    generic B0-B0bar ~16.5 ~7.6
    generic c-cbar ~15.2 ~6.9
    generic uds ~14.7 ~6.4
    data ~7.3 ~2.6

  2. size of skimmed data for few skims

    size(Kb*ev-1)
    deep copy Mini deep copy Micro adding composites adding UsrData pointer skim
    BSemiExcl ~14.4 ~5.5 +0.2 +0.1 ~1.3
    BRecoToDDstar ~12.3 ~4.9 +0.9 +0.4 ~1.9
    BToDlnu ~11.8 ~4.6 +0.4 +0.1 ~1.1
    XSLBtoXulnuFilter ~11.8 ~4.6 +0.2 +0.1 -

CPU time

As described in the introduction, CM2 allows the user to run on skimmed data in different ways. Summarizing there are the following approaches:
  1. redo combinatorics. This is the CM1-like style. The skims are used just as subsamples and the candidates for the final analysis are redone. The CPU time in this case is very dependent on the particular reconstruction the user is doing.
  2. reading back composites
  3. reading back composites and UsrData
  4. reskimming of existing skims. To allow the user to create small samples for the final analysis
  5. interactive Beta. Reading composite and make use of Beta interactively in root.
The following numbers have been calculated running on BSemiExcl skims on a noma batch machine.

ms*ev-1
read micro read compos. read compos.+UsrData reskimming int. Beta
- - - - -
vertexing per B - - - - -

CPU time for Elf, Moose and BetaMiniApp

Some more detailed information on release 14 BetaMiniApp, Moose and Elf applications and on each contributing module can be found here: BetaMiniApp, Moose and Elf.