Phase 2 GTT TrackVertexAssociation Firmware limit #43278
Conversation
…ficial test vector validation. Add Converted tracks input tag and link-limit constant from firmware as configurable parameter. Prototype fills of remaining vtx words
…hod in codecs and L1TrackVertexAssociationProducer Fix L1TrackVertexAssociationProducer.cc etaSector calculation Centralize the gttLinkID method in the tracks codecs code format for codecs/tracks.h Add sectorsEtaPhi inversion func
…ationProducer.cc, apply code-format Remove prototype filled vertices Update L1Trigger/L1TTrackMatch/plugins/L1TrackVertexAssociationProducer.cc Remove leftover comment
|
+code-checks Logs: https://cmssdt.cern.ch/SDT/code-checks/cms-sw-PR-43278/37687
|
|
A new Pull Request was created by @NJManganelli (Nick) for master. It involves the following packages:
@AdrianoDee, @aloeliger, @epalencia, @srimanob, @cmsbuild can you please review it and eventually sign? Thanks. cms-bot commands are listed here |
|
Please test |
|
+1 Summary: https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-4d5e7a/35818/summary.html The following merge commits were also included on top of IB + this PR after doing git cms-merge-topic:
You can see more details here: Comparison SummarySummary:
|
|
+l1
|
|
+Upgrade |
cmsbuild
added
fully-signed
upgrade-approved
and removed
pending-signatures
upgrade-pending
labels
|
This pull request is fully signed and it will be integrated in one of the next master IBs (tests are also fine). This pull request will now be reviewed by the release team before it's merged. @antoniovilela, @rappoccio, @sextonkennedy (and backports should be raised in the release meeting by the corresponding L2) |
|
+1 |
PR description:
This PR implements matching of emulation to a firmware feature for the Global Track Trigger. For GTT, there are 18 optical links (9 phi sectors x 2 eta sectors) from the TrackFindingProcessors. Each link can send up to 104 tracks, but a limitation in the current firmware prevents all from being considered for association with the primary vertex: the limit is currently 94/104, and this limit is encoded in a configurable parameter in the l1tTrackVertexAssociation_cfi.py file. This has no impact on vertices (which are found using all (selected) tracks sent to GTT and strictly upstream of the TrackVertexAssociation), and in the cases of track-only objects, is expected to impact less than 1 in 10^6 ttbar (200PU) events (the original limit of 104 tracks per link was designed around such a figure of merit + an overhead, this PR decreases the overhead a small amount)
Bit-accurate outputs are maintained for board integration tests on 1008 events ttbar (200PU) events.
In anticipation of another PR, the code used to determine the link a track belongs to is encapsulated in a function in the codecs header, to be used commonly in the GTTFile{Reader,Writer} when updated.
PR validation:
This PR passed:
scram b
scram b code-checks
scram b code-format
Used to generate GTT test vectors for APx/Serenity on ~1000 events from TT (CMSSW_13_1) sample
Test vectors pass tests with corresponding update for TrackVertexAssociation in the LibHLS repo
runTheMatrix.py -l limited -i all --ibeos
Ports
This PR is a port from the cms-l1t-offline 13_3_0_pre3 integration branch, with commits squashed
cms-l1t-offline#1176