Ruben Shahoyan

+test

Hi, I agree with @sawenzel , I would consider converting it to CCDB object.

Merging after the fix and local test.

> Ok, this I somehow expected. The overflow pixels are indeed just "pushed_back" into the chip digits vector. Would a 2-D sorting (row-col) done at the end of squashing work?...

Thanks, the ratio of tracks is reasonable but the eta shape is not as flat is it was in Oct.2021 test. For the clusters the effect is not evident from...

@mconcas thanks! In principle, yes, but do you have CPU benchmark for e.g. 500 kHz data? We must match the processing rate at least 0.5Hz of incoming TF rate at...

Hi @mconcas why not use for the neighbour definition the same maxRowColDiffToMask? The meaning is exactly the same. As for the number of ROFs to squash, I would not add...

@mconcas yes, you can add a logic ``` int rofBC = mClusterer->isContinuousReadOut() ? alpParams.roFrameLengthInBC : (alpParams.roFrameLengthTrig / o2::constants::lhc::LHCBunchSpacingNS); // ROF length in BC int nbcSeparation= rofBC + clParams.maxBCDiffToMaskBias; // enable...

The ROFs are not guaranteed to be contiguous even in continuous readout (e.g. empty ROF may be absent), one should always check if 2 ROFs are adjascent.

@mconcas ready to be merged?