Error in <TTreeCache::FillBuffer>: Inconsistency [...] and conflicting results

Dear experts,

When computing the sum of gen weights for a set of files in a TChain with a friend chain using RDataFrame I’m getting this message:

Error in TTreeCache::FillBuffer: Inconsistency: fCurrentClusterStart=0 fEntryCurrent=85705 fNextClusterStart=103062 but fEntryCurrent should not be in between the two

Also, if I compare the results with my loop-style computation using SetBranchAddress I get different numbers. Not sure what’s going on. I have tried setting the chain cache size to 0 as per but it didn’t make any difference.

Here’s the relevant code for both ways:

SetBranchAddress and loop over events:

TChain * data_gen = new TChain("ntuples_gbm/genT");
// Add files to gen TChain
data_gen->SetBranchStatus("gen_wgt", 1);
Float_t gen_wgt;
data_gen->SetBranchAddress("gen_wgt", &gen_wgt);
Float_t sum_gen_wgt = 0;
for (int i = 0; i < data_gen->GetEntries(); i++) {
    sum_gen_wgt += gen_wgt;

cout << "sample: " << sample << ", sum_gen_wgt: " << sum_gen_wgt << endl;

RDataFrame with .Sum:

TChain * data_reco = new TChain("ntuples_gbm/recoT");
TChain * data_gen = new TChain("ntuples_gbm/genT");
// Add files to both

ROOT::RDataFrame df(*data_reco);
auto df_wgts = df.
       Define("Zwgt", calcZsf, {"gen_ID", "gen_pt"}).
       Define("Wwgt", calcWsf, {"gen_ID", "gen_pt"}).
       Define("Twgt", calcTsf, {"gen_ID", "gen_pt"}).
       Define("PUwgt", calcPUsf, {"gen_pu_true"}).
       Define("wgt", calcTotalWgt, {"Zwgt", "Wwgt", "Twgt", "PUwgt", "gen_wgt"});
auto df_sumgenwgts = df_wgts.Sum("gen_wgt");

cout << "Sum_gen_wgts as inferred from RDF: " << *df_sumgenwgts << endl;

Just as an example, I get sum_gen_wgt: 1.67772e+07 for a particular sample using the first way and 1.00094e+08 using the second.

Any help in figuring this out will be greatly appreciated!

Thank you,

ROOT Version: 6.18/00
Platform: x86_64 slc6
Compiler: gcc8

Try with: Double_t sum_gen_wgt = 0.;

1 Like

Thanks! That does seem to match up the results from both methods. I was under the impression that given the max number that floats can hold (something like 10^38) this wouldn’t be a problem, but I guess the precision is what matters here.

Regarding the TTreeCache::FillBuffer error, is it a harmless message?

Thanks again,

Hi Andre,
and welcome to the ROOT forum.


let’s ping our TTree expert @pcanal .


Can you try with v6.18/04? If it still show the error message, can you provide a reproducer (and fill report bug report in JIRA). Thanks.

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.