Hallo vubbers, while validating the skims I discovered that you are (unconsciously) running the production in an inefficient way. A "feature" sneaked in so that you are loading the micro for each event and only after loading the micro you filter on the tag database. Because of a similar problem, the filter on isBCMultihadron was not working, hence enhancing the load for the modes running on AllEvents. The fix is in SemiExclUser rf-120701 (or, if you are not using all the analysis-12 tags, the files you need to update are: SemiExclApp.cc TagFilterSequence.tcl SemiExclBdbFilter.tcl SemiExclkangaFilter.tcl ) As a check I am running with the D+ stream setup on the collection /groups/AllEvents/0001/9600/P10.0.3aV01fb/00019619/cb001/allevents I am currently queing up. Once the ntuple will come out in ~rfaccini/scra/8.6.1y/hbook/testFilter.hbook you will be able to compare it with the one already produced by Oliver. I estimate ~20% saving in time, but I will be able to quantify after the job has run... ciao ric