Restructure TrackletProcessorDisplaced #297
Conversation
L1Trigger/TrackFindingTracklet/interface/TrackletProcessorDisplaced.h
Outdated
Show resolved
Hide resolved
|
Comments above were addressed, we can let @tomalin merge when he has also taken a look. |
aehart
changed the base branch from
L1TK-dev-14_0_0_pre2
to
L1TK-dev-14_2_0_pre4
aehart
force-pushed
the
restructure_TPD_PR
branch
from
4378c34 to
9d75886
Compare
|
I took the liberty of changing the target branch to the new default branch (L1TK-dev-14_2_0_pre4). I think this was the cause of the failing CI jobs. I also rebased the PR branch onto the new default branch and added a commit re-adding the L1Trigger/TrackFindingTracklet/data/ directory. Hopefully this all allows the CI to pass again. |
| @@ -0,0 +1,2721 @@ | |||
| AllInnerStubs: AS_D1PHIA_DM [42] | |||
There was a problem hiding this comment.
This file is also changed by #301 . Why do we have two PRs changing the same file?
There was a problem hiding this comment.
I guess this reflects the course of development. The changes here are a minor bug fix to the wiring, while #301 includes more substantial changes.
L1Trigger/TrackFindingTracklet/data/processingmodules_hourglassExtendedAllCombined.dat
Outdated
Show resolved
Hide resolved
| @@ -666,7 +667,8 @@ namespace trklet { | |||
| int chisqphifactbits_{14}; | |||
| int chisqzfactbits_{14}; | |||
|
|
|||
| std::array<unsigned int, N_SEED> teunits_{{5, 2, 5, 3, 3, 2, 3, 2, 0, 0, 0, 0}}; //teunits used by seed | |||
| std::array<unsigned int, N_SEED> teunits_{{5, 2, 5, 3, 3, 2, 3, 2, 0, 0, 0, 0}}; //teunits used by seed | |||
| std::array<unsigned int, N_SEED> trpunits_{{0, 0, 0, 0, 0, 0, 0, 0, 10, 10, 10, 10}}; //trpunits used by seed | |||
There was a problem hiding this comment.
This indicates the number of triplet engine units, distinct from tracklet engine units
L1Trigger/TrackFindingTracklet/src/TrackletProcessorDisplaced.cc
Outdated
Show resolved
Hide resolved
L1Trigger/TrackFindingTracklet/src/TrackletProcessorDisplaced.cc
Outdated
Show resolved
Hide resolved
|
This needs rebasing, now I've merged your other PR, since of them changed the same wiring map files, so we now have a git conflict. |
cgsavard
force-pushed
the
restructure_TPD_PR
branch
from
dae9c97 to
84293ed
Compare
aehart
force-pushed
the
restructure_TPD_PR
branch
from
84293ed to
3af747e
Compare
|
@cgsavard Have you checked that your PR to update the wiring map in central CMSSW cms-data repo still corresponds to the wiring here, giving you/Andrew have mucked around with the wiring a bit to resolve the conflicts between your two PRs? |
Changed geometry for latest CMSSW release version Run combined modules by default (#265) * Make combined modules default * tweak * Improve USEHYBRID ifdef range * Fix compiler error for pure Tracklet algo * Move fitpattern.txt refs, so only used for pure Tracklet algo * code format Bin TQ MVA output (#272) * add TQ MVA output binning scheme * remove code no longer used * move logistic sigmoid into TTTrack class * fix for proper precision after MVA conversion in track word * label all MVA variables before logistic sigmoid with 'Pre' * formatting * fix filling of track word before running TQ MVA * switch to post-sigmoid mva bins * remove any onnx instances * fix MVA1 initialization * bin TQ MVA in pre-sigmoided bins in KFout * move pre-sigmoid bins to track quality class * change minimum bin squash c++20 and merge fixes code-format run scripts update. run script clean up. rebase fix TrackFindingProcessor for hybrid adapted. script fix little fixes. furthrt little fixes TQ outputs altered. TM mux order configurable New wiring for TPD (#301) * new wiring for TPD * fix seed 6 duplicates * code formatting Restructure TrackletProcessorDisplaced (#297) * initial add of restructured TPD * clean up new TPD code * remove unnecessary includes * remove magic numbers auto code format (#303) Eliminated MatchCalculator vs MonteCarlo ambiguity
PR description:
This PR restructures TrackletProcessorDisplaced to look like TrackletProcessor which was adapted to better follow the firmware implementation. The new structure is explained in this presentation here and requires the addition of a new TripletEngineUnit script (mimicking the TrackletEngineUnit in the TrackletProcessor case) tasked with creating triplets from different stub combinations.
In addition to this restructuring, a bug in the wiring map was fixed which was causing duplicate stub memories to be passed into the same TPD instance.
PR validation:
No changes were made which affect the performance of the tracking. I have checked the make sure that the performance (efficiency, fake rate, duplicate rate, etc) is the same before and after these changes.