The decoding v5 for SciFi, requires to check every flagged cluster together with next one to know if it's a begin or an end fragment of a large cluster. This doesn't work well for multi-threading. To allow this, the data format is changed as following:
- small clusters are same as before, size flag is set to 0 - begin fragment of large has size flag 1 and fraction flag 1 [all first fragments have size 4 therefore fraction is always 1 anyway] - end fragment has size flag 1 and fraction flag 0. In order to be able to recover information of the real fraction, the channel number of the cluster is not the center channel but the last fired channel. Knowing the position of the begin fragment and the last channel, a fragmented large cluster can be recombined or missing middle fragments can be added as before.
The new encoding and decoding version is v6.
Following files are updated: FTClusterCreator in FT/FTDigitisation from Boole FTRawbankEncoder in FT/FTDAQ from LHCb [only coding version is changed to 6] FTRawbankDecoder in FT/FTDAQ from LHCb [addition to above changes, max cluster size is fixed to 4.]
This merge should be accepted together with LHCb!1709 (merged) (LHCb/dev-scifi-changeDataFormat)
Due to a bug in v5, small increase will be seen in hit efficiency with v6.