Simple best track creator
This MR adds two new algorithms:
-
create_reduced_scifi_hit_container_t: Gets a scifi hit container and a list of used hits and creates a new reduced scifi hit container with only unused hits. -
combine_long_track_containers_t: Combines two long track containers. The combination is done without any smart filters applied. In particular, it doesn't remove clones.
With the use of these two algorithms, one can combine the outputs of the two main algorithms that currently produce long tracks: Seeding+matching and Forward tracking. There are other manners one could think of joining the outputs of these containers and the intent of this MR is not to achieve all of them.
The following new configuration options are now available in setup_hlt1_node, which now configures the tracking with an enum with the following options:
-
FORWARD,MATCHING: Options that refer to preexisting tracking modes. -
FORWARD_THEN_MATCHING: Configures a sequence where forward runs first, then seeding+matching is run over unused scifi hits and unused velo tracks, and finally both long track containers are combined. -
MATCHING_THEN_FORWARD: Vice versa of the previous sequence.
Performance
The impact on throughput depends on the sequence. From a first look, the two newly added algorithms have a negligible effect on throughput, with measured contributions of 1% for create_reduced_scifi_hit_container_t and 0.06% for combine_track_views. The throughput impact comes then from running the additional tracking algorithm. Having more long tracks also impacts other subsequent algorithms, like secondary vertex finding. For instance, going from hlt1_pp_default to hlt1_pp_forward_then_matching results in a 20-30% slowdown, and the top contributors look as follows:
lf_triplet_seeding │██████████████████████████████████████████████ 11.63 %
run_lines │██████████████████████████████████ 8.59 %
seed_xz │█████████████████████████████████ 8.44 %
velo_search_by_triplet │██████████████████████████████ 7.63 %
velo_calculate_permutations │█████████████████████ 5.34 %
ut_find_permutation │██████████████████ 4.51 %
ut_search_windows │███████████████ 3.79 %
seed_confirmTracks │██████████████ 3.72 %Hereby some throughput tests. The hlt1_pp_forward_then_matching configurations yield many more tracks (see below), and hence are slower than hlt1_pp_matching_then_forward in the tested scenarios:
* Settings for tests of all configs:
Testing card: NVIDIA RTX A5000
Config line: CUDA_VISIBLE_DEVICES=1 ./toolchain/wrapper ./Allen --sequence <seq> --mdf /scratch/allen_data/mdf_input/upgrade_mc_minbias_scifi_v5_retinacluster_000_v1_newLHCbID.mdf -t 16 -n500 -m700 -r 200
* UT configurations:
hlt1_pp_no_gec: 124061.103389 ev/s
hlt1_pp_matching_then_forward_no_gec: 90470.560748 ev/s
hlt1_pp_forward_then_matching_no_gec: 86547.350429 ev/s
* No UT configurations:
hlt1_pp_no_ut_no_gec: 123192.291047
hlt1_pp_matching_no_gec: 124656.504279 ev/s
hlt1_pp_matching_then_forward_no_gec_no_ut: 90173.703443 ev/s
hlt1_pp_forward_then_matching_no_ut_no_gec: 80347.166548 ev/sA positive effect on efficiency is observed all around. A more in-depth study will follow, but from a first look fakes increase slightly and reco efficiency improves all around. Worth noting is that by construction these configurations don't increase clones.
- With UT:
Dataset: Upgrade_BsPhiPhi_MD_FTv4_DIGI_retinacluster_v1_newLHCbID.mdf
Configuration: No GEC
Forward (hlt1_pp_no_gec_validation):
long_validator validation:
TrackChecker output : 1741/ 42511 4.10% ghosts
for P>3GeV,Pt>0.5GeV : 935/ 26717 3.50% ghosts
01_long : 38567/ 75133 51.33% ( 52.84%), 346 ( 0.89%) clones, pur 99.09%, hit eff 95.36%
02_long_P>5GeV : 31931/ 47271 67.55% ( 68.91%), 277 ( 0.86%) clones, pur 99.21%, hit eff 96.38%
03_long_strange : 1178/ 3460 34.05% ( 33.74%), 8 ( 0.67%) clones, pur 98.60%, hit eff 94.62%
04_long_strange_P>5GeV : 860/ 1609 53.45% ( 54.39%), 3 ( 0.35%) clones, pur 98.72%, hit eff 95.99%
05_long_fromB : 3304/ 4413 74.87% ( 76.33%), 27 ( 0.81%) clones, pur 99.19%, hit eff 96.57%
06_long_fromB_P>5GeV : 3087/ 3641 84.78% ( 85.11%), 27 ( 0.87%) clones, pur 99.22%, hit eff 96.87%
07_long_electrons : 755/ 5807 13.00% ( 13.91%), 23 ( 2.96%) clones, pur 98.36%, hit eff 95.07%
08_long_electrons_P>5GeV : 659/ 2901 22.72% ( 24.04%), 19 ( 2.80%) clones, pur 98.51%, hit eff 96.00%
09_long_fromB_electrons : 72/ 212 33.96% ( 39.10%), 3 ( 4.00%) clones, pur 98.49%, hit eff 96.19%
10_long_fromB_electrons_P>5GeV : 69/ 138 50.00% ( 57.08%), 3 ( 4.17%) clones, pur 98.51%, hit eff 96.49%
long_P>5GeV_AND_Pt>1GeV : 8273/ 9402 87.99% ( 89.35%), 86 ( 1.03%) clones, pur 99.30%, hit eff 96.83%
long_fromB_P>5GeV_AND_Pt>1GeV : 2073/ 2284 90.76% ( 90.49%), 24 ( 1.14%) clones, pur 99.28%, hit eff 97.18%
11_noVelo_UT : 0/ 8097 0.00% ( 0.00%), 0 ( 0.00%) clones, pur -nan%, hit eff -nan%
12_noVelo_UT_P>5GeV : 0/ 3231 0.00% ( 0.00%), 0 ( 0.00%) clones, pur -nan%, hit eff -nan%
13_long_PT>2GeV : 2058/ 2270 90.66% ( 91.52%), 21 ( 1.01%) clones, pur 99.31%, hit eff 97.23%
14_long_from_B_PT>2GeV : 913/ 997 91.57% ( 91.34%), 11 ( 1.19%) clones, pur 99.22%, hit eff 97.21%
15_long_strange_P>5GeV : 860/ 1609 53.45% ( 54.39%), 3 ( 0.35%) clones, pur 98.72%, hit eff 95.99%
16_long_strange_P>5GeV_PT>500MeV : 460/ 558 82.44% ( 82.76%), 2 ( 0.43%) clones, pur 98.67%, hit eff 95.63%
Matching (hlt1_pp_matching_no_gec_validation):
long_validator validation:
TrackChecker output : 5406/ 57197 9.45% ghosts
for P>3GeV,Pt>0.5GeV : 1346/ 25646 5.25% ghosts
01_long : 48539/ 75133 64.60% ( 67.28%), 1 ( 0.00%) clones, pur 99.47%, hit eff 96.08%
02_long_P>5GeV : 37762/ 47271 79.88% ( 83.41%), 1 ( 0.00%) clones, pur 99.48%, hit eff 96.68%
03_long_strange : 1862/ 3460 53.82% ( 54.78%), 0 ( 0.00%) clones, pur 99.38%, hit eff 96.03%
04_long_strange_P>5GeV : 1245/ 1609 77.38% ( 80.09%), 0 ( 0.00%) clones, pur 99.35%, hit eff 96.82%
05_long_fromB : 3506/ 4413 79.45% ( 80.39%), 0 ( 0.00%) clones, pur 99.47%, hit eff 96.74%
06_long_fromB_P>5GeV : 3176/ 3641 87.23% ( 87.92%), 0 ( 0.00%) clones, pur 99.47%, hit eff 97.01%
07_long_electrons : 1850/ 5807 31.86% ( 33.39%), 0 ( 0.00%) clones, pur 98.98%, hit eff 95.67%
08_long_electrons_P>5GeV : 1385/ 2901 47.74% ( 50.56%), 0 ( 0.00%) clones, pur 99.00%, hit eff 96.28%
09_long_fromB_electrons : 97/ 212 45.75% ( 49.01%), 0 ( 0.00%) clones, pur 99.40%, hit eff 96.27%
10_long_fromB_electrons_P>5GeV : 85/ 138 61.59% ( 65.93%), 0 ( 0.00%) clones, pur 99.40%, hit eff 96.58%
long_P>5GeV_AND_Pt>1GeV : 7890/ 9402 83.92% ( 86.87%), 1 ( 0.01%) clones, pur 99.49%, hit eff 96.92%
long_fromB_P>5GeV_AND_Pt>1GeV : 2004/ 2284 87.74% ( 87.84%), 0 ( 0.00%) clones, pur 99.51%, hit eff 97.27%
11_noVelo_UT : 0/ 8097 0.00% ( 0.00%), 0 ( 0.00%) clones, pur -nan%, hit eff -nan%
12_noVelo_UT_P>5GeV : 0/ 3231 0.00% ( 0.00%), 0 ( 0.00%) clones, pur -nan%, hit eff -nan%
13_long_PT>2GeV : 1938/ 2270 85.37% ( 87.04%), 1 ( 0.05%) clones, pur 99.49%, hit eff 97.24%
14_long_from_B_PT>2GeV : 875/ 997 87.76% ( 88.11%), 0 ( 0.00%) clones, pur 99.45%, hit eff 97.12%
15_long_strange_P>5GeV : 1245/ 1609 77.38% ( 80.09%), 0 ( 0.00%) clones, pur 99.35%, hit eff 96.82%
16_long_strange_P>5GeV_PT>500MeV : 465/ 558 83.33% ( 84.37%), 0 ( 0.00%) clones, pur 99.12%, hit eff 96.42%
Matching then forward (hlt1_pp_matching_then_forward_no_gec_validation):
long_validator validation:
TrackChecker output : 5804/ 62205 9.33% ghosts
for P>3GeV,Pt>0.5GeV : 1570/ 27977 5.61% ghosts
01_long : 52747/ 75133 70.20% ( 72.29%), 35 ( 0.07%) clones, pur 99.37%, hit eff 95.28%
02_long_P>5GeV : 39897/ 47271 84.40% ( 86.78%), 17 ( 0.04%) clones, pur 99.40%, hit eff 96.11%
03_long_strange : 2008/ 3460 58.03% ( 59.02%), 0 ( 0.00%) clones, pur 99.28%, hit eff 95.29%
04_long_strange_P>5GeV : 1289/ 1609 80.11% ( 82.66%), 0 ( 0.00%) clones, pur 99.31%, hit eff 96.45%
05_long_fromB : 3715/ 4413 84.18% ( 85.07%), 2 ( 0.05%) clones, pur 99.36%, hit eff 96.23%
06_long_fromB_P>5GeV : 3322/ 3641 91.24% ( 91.68%), 2 ( 0.06%) clones, pur 99.36%, hit eff 96.55%
07_long_electrons : 1949/ 5807 33.56% ( 35.00%), 3 ( 0.15%) clones, pur 98.88%, hit eff 95.24%
08_long_electrons_P>5GeV : 1449/ 2901 49.95% ( 52.30%), 3 ( 0.21%) clones, pur 98.91%, hit eff 95.93%
09_long_fromB_electrons : 101/ 212 47.64% ( 51.18%), 0 ( 0.00%) clones, pur 99.26%, hit eff 96.16%
10_long_fromB_electrons_P>5GeV : 89/ 138 64.49% ( 69.47%), 0 ( 0.00%) clones, pur 99.24%, hit eff 96.45%
long_P>5GeV_AND_Pt>1GeV : 8433/ 9402 89.69% ( 91.48%), 7 ( 0.08%) clones, pur 99.37%, hit eff 96.21%
long_fromB_P>5GeV_AND_Pt>1GeV : 2112/ 2284 92.47% ( 92.43%), 2 ( 0.09%) clones, pur 99.40%, hit eff 96.77%
11_noVelo_UT : 0/ 8097 0.00% ( 0.00%), 0 ( 0.00%) clones, pur -nan%, hit eff -nan%
12_noVelo_UT_P>5GeV : 0/ 3231 0.00% ( 0.00%), 0 ( 0.00%) clones, pur -nan%, hit eff -nan%
13_long_PT>2GeV : 2047/ 2270 90.18% ( 90.89%), 1 ( 0.05%) clones, pur 99.40%, hit eff 96.66%
14_long_from_B_PT>2GeV : 921/ 997 92.38% ( 92.51%), 0 ( 0.00%) clones, pur 99.36%, hit eff 96.61%
15_long_strange_P>5GeV : 1289/ 1609 80.11% ( 82.66%), 0 ( 0.00%) clones, pur 99.31%, hit eff 96.45%
16_long_strange_P>5GeV_PT>500MeV : 489/ 558 87.63% ( 88.31%), 0 ( 0.00%) clones, pur 99.08%, hit eff 95.91%
Forward then matching (hlt1_pp_forward_then_matching_no_gec_validation):
long_validator validation:
TrackChecker output : 6258/ 65663 9.53% ghosts
for P>3GeV,Pt>0.5GeV : 1836/ 30590 6.00% ghosts
01_long : 55087/ 75133 73.32% ( 74.54%), 346 ( 0.62%) clones, pur 99.18%, hit eff 95.20%
02_long_P>5GeV : 42073/ 47271 89.00% ( 90.09%), 277 ( 0.65%) clones, pur 99.25%, hit eff 96.17%
03_long_strange : 2084/ 3460 60.23% ( 60.74%), 8 ( 0.38%) clones, pur 98.98%, hit eff 95.01%
04_long_strange_P>5GeV : 1357/ 1609 84.34% ( 86.19%), 3 ( 0.22%) clones, pur 99.00%, hit eff 96.21%
05_long_fromB : 3824/ 4413 86.65% ( 87.39%), 27 ( 0.70%) clones, pur 99.21%, hit eff 96.38%
06_long_fromB_P>5GeV : 3431/ 3641 94.23% ( 94.33%), 27 ( 0.78%) clones, pur 99.22%, hit eff 96.76%
07_long_electrons : 2045/ 5807 35.22% ( 36.42%), 23 ( 1.11%) clones, pur 98.66%, hit eff 94.87%
08_long_electrons_P>5GeV : 1529/ 2901 52.71% ( 54.90%), 19 ( 1.23%) clones, pur 98.69%, hit eff 95.62%
09_long_fromB_electrons : 107/ 212 50.47% ( 54.49%), 3 ( 2.73%) clones, pur 98.86%, hit eff 95.65%
10_long_fromB_electrons_P>5GeV : 95/ 138 68.84% ( 74.34%), 3 ( 3.06%) clones, pur 98.78%, hit eff 96.02%
long_P>5GeV_AND_Pt>1GeV : 8879/ 9402 94.44% ( 95.13%), 86 ( 0.96%) clones, pur 99.30%, hit eff 96.71%
long_fromB_P>5GeV_AND_Pt>1GeV : 2193/ 2284 96.02% ( 95.67%), 24 ( 1.08%) clones, pur 99.29%, hit eff 97.13%
11_noVelo_UT : 0/ 8097 0.00% ( 0.00%), 0 ( 0.00%) clones, pur -nan%, hit eff -nan%
12_noVelo_UT_P>5GeV : 0/ 3231 0.00% ( 0.00%), 0 ( 0.00%) clones, pur -nan%, hit eff -nan%
13_long_PT>2GeV : 2166/ 2270 95.42% ( 95.84%), 21 ( 0.96%) clones, pur 99.32%, hit eff 97.17%
14_long_from_B_PT>2GeV : 957/ 997 95.99% ( 95.82%), 11 ( 1.14%) clones, pur 99.23%, hit eff 97.16%
15_long_strange_P>5GeV : 1357/ 1609 84.34% ( 86.19%), 3 ( 0.22%) clones, pur 99.00%, hit eff 96.21%
16_long_strange_P>5GeV_PT>500MeV : 513/ 558 91.94% ( 92.12%), 2 ( 0.39%) clones, pur 98.75%, hit eff 95.65%- Without UT:
Dataset: Upgrade_BsPhiPhi_MD_FTv4_DIGI_retinacluster_v1_newLHCbID.mdf
Configuration: No UT, No GEC
Forward (hlt1_pp_no_gec_no_ut_validation):
long_validator validation:
TrackChecker output : 1482/ 13109 11.31% ghosts
for P>3GeV,Pt>0.5GeV : 1482/ 13109 11.31% ghosts
01_long : 11136/ 75133 14.82% ( 15.32%), 110 ( 0.98%) clones, pur 99.55%, hit eff 96.27%
02_long_P>5GeV : 11071/ 47271 23.42% ( 24.04%), 110 ( 0.98%) clones, pur 99.56%, hit eff 96.29%
03_long_strange : 164/ 3460 4.74% ( 4.69%), 1 ( 0.61%) clones, pur 99.11%, hit eff 95.10%
04_long_strange_P>5GeV : 162/ 1609 10.07% ( 10.54%), 1 ( 0.61%) clones, pur 99.22%, hit eff 95.15%
05_long_fromB : 2350/ 4413 53.25% ( 56.33%), 25 ( 1.05%) clones, pur 99.56%, hit eff 97.06%
06_long_fromB_P>5GeV : 2341/ 3641 64.30% ( 66.71%), 25 ( 1.06%) clones, pur 99.56%, hit eff 97.07%
07_long_electrons : 115/ 5807 1.98% ( 2.12%), 4 ( 3.36%) clones, pur 98.90%, hit eff 96.60%
08_long_electrons_P>5GeV : 115/ 2901 3.96% ( 4.37%), 4 ( 3.36%) clones, pur 98.90%, hit eff 96.60%
09_long_fromB_electrons : 35/ 212 16.51% ( 19.07%), 2 ( 5.41%) clones, pur 99.10%, hit eff 96.33%
10_long_fromB_electrons_P>5GeV : 35/ 138 25.36% ( 28.54%), 2 ( 5.41%) clones, pur 99.10%, hit eff 96.33%
long_P>5GeV_AND_Pt>1GeV : 8334/ 9402 88.64% ( 89.53%), 89 ( 1.06%) clones, pur 99.56%, hit eff 96.90%
long_fromB_P>5GeV_AND_Pt>1GeV : 2088/ 2284 91.42% ( 90.75%), 24 ( 1.14%) clones, pur 99.55%, hit eff 97.25%
11_noVelo_UT : 0/ 8097 0.00% ( 0.00%), 0 ( 0.00%) clones, pur -nan%, hit eff -nan%
12_noVelo_UT_P>5GeV : 0/ 3231 0.00% ( 0.00%), 0 ( 0.00%) clones, pur -nan%, hit eff -nan%
13_long_PT>2GeV : 2108/ 2270 92.86% ( 93.40%), 18 ( 0.85%) clones, pur 99.59%, hit eff 97.26%
14_long_from_B_PT>2GeV : 937/ 997 93.98% ( 94.12%), 10 ( 1.06%) clones, pur 99.52%, hit eff 97.20%
15_long_strange_P>5GeV : 162/ 1609 10.07% ( 10.54%), 1 ( 0.61%) clones, pur 99.22%, hit eff 95.15%
16_long_strange_P>5GeV_PT>500MeV : 162/ 558 29.03% ( 28.42%), 1 ( 0.61%) clones, pur 99.22%, hit eff 95.15%
Matching (hlt1_pp_matching_no_gec_validation):
long_validator validation:
TrackChecker output : 5406/ 57197 9.45% ghosts
for P>3GeV,Pt>0.5GeV : 1346/ 25646 5.25% ghosts
01_long : 48539/ 75133 64.60% ( 67.28%), 1 ( 0.00%) clones, pur 99.47%, hit eff 96.08%
02_long_P>5GeV : 37762/ 47271 79.88% ( 83.41%), 1 ( 0.00%) clones, pur 99.48%, hit eff 96.68%
03_long_strange : 1862/ 3460 53.82% ( 54.78%), 0 ( 0.00%) clones, pur 99.38%, hit eff 96.03%
04_long_strange_P>5GeV : 1245/ 1609 77.38% ( 80.09%), 0 ( 0.00%) clones, pur 99.35%, hit eff 96.82%
05_long_fromB : 3506/ 4413 79.45% ( 80.39%), 0 ( 0.00%) clones, pur 99.47%, hit eff 96.74%
06_long_fromB_P>5GeV : 3176/ 3641 87.23% ( 87.92%), 0 ( 0.00%) clones, pur 99.47%, hit eff 97.01%
07_long_electrons : 1850/ 5807 31.86% ( 33.39%), 0 ( 0.00%) clones, pur 98.98%, hit eff 95.67%
08_long_electrons_P>5GeV : 1385/ 2901 47.74% ( 50.56%), 0 ( 0.00%) clones, pur 99.00%, hit eff 96.28%
09_long_fromB_electrons : 97/ 212 45.75% ( 49.01%), 0 ( 0.00%) clones, pur 99.40%, hit eff 96.27%
10_long_fromB_electrons_P>5GeV : 85/ 138 61.59% ( 65.93%), 0 ( 0.00%) clones, pur 99.40%, hit eff 96.58%
long_P>5GeV_AND_Pt>1GeV : 7890/ 9402 83.92% ( 86.87%), 1 ( 0.01%) clones, pur 99.49%, hit eff 96.92%
long_fromB_P>5GeV_AND_Pt>1GeV : 2004/ 2284 87.74% ( 87.84%), 0 ( 0.00%) clones, pur 99.51%, hit eff 97.27%
11_noVelo_UT : 0/ 8097 0.00% ( 0.00%), 0 ( 0.00%) clones, pur -nan%, hit eff -nan%
12_noVelo_UT_P>5GeV : 0/ 3231 0.00% ( 0.00%), 0 ( 0.00%) clones, pur -nan%, hit eff -nan%
13_long_PT>2GeV : 1938/ 2270 85.37% ( 87.04%), 1 ( 0.05%) clones, pur 99.49%, hit eff 97.24%
14_long_from_B_PT>2GeV : 875/ 997 87.76% ( 88.11%), 0 ( 0.00%) clones, pur 99.45%, hit eff 97.12%
15_long_strange_P>5GeV : 1245/ 1609 77.38% ( 80.09%), 0 ( 0.00%) clones, pur 99.35%, hit eff 96.82%
16_long_strange_P>5GeV_PT>500MeV : 465/ 558 83.33% ( 84.37%), 0 ( 0.00%) clones, pur 99.12%, hit eff 96.42%
Matching then forward (hlt1_pp_matching_then_forward_no_gec_no_ut_validation):
long_validator validation:
TrackChecker output : 5529/ 57949 9.54% ghosts
for P>3GeV,Pt>0.5GeV : 1469/ 26398 5.56% ghosts
01_long : 49133/ 75133 65.39% ( 67.86%), 4 ( 0.01%) clones, pur 99.47%, hit eff 95.96%
02_long_P>5GeV : 38354/ 47271 81.14% ( 84.32%), 4 ( 0.01%) clones, pur 99.48%, hit eff 96.53%
03_long_strange : 1871/ 3460 54.08% ( 54.96%), 0 ( 0.00%) clones, pur 99.38%, hit eff 95.98%
04_long_strange_P>5GeV : 1254/ 1609 77.94% ( 80.52%), 0 ( 0.00%) clones, pur 99.35%, hit eff 96.75%
05_long_fromB : 3597/ 4413 81.51% ( 82.51%), 2 ( 0.06%) clones, pur 99.46%, hit eff 96.52%
06_long_fromB_P>5GeV : 3266/ 3641 89.70% ( 90.35%), 2 ( 0.06%) clones, pur 99.45%, hit eff 96.75%
07_long_electrons : 1860/ 5807 32.03% ( 33.51%), 0 ( 0.00%) clones, pur 98.98%, hit eff 95.67%
08_long_electrons_P>5GeV : 1395/ 2901 48.09% ( 50.86%), 0 ( 0.00%) clones, pur 99.00%, hit eff 96.28%
09_long_fromB_electrons : 98/ 212 46.23% ( 49.32%), 0 ( 0.00%) clones, pur 99.41%, hit eff 96.31%
10_long_fromB_electrons_P>5GeV : 86/ 138 62.32% ( 66.81%), 0 ( 0.00%) clones, pur 99.40%, hit eff 96.62%
long_P>5GeV_AND_Pt>1GeV : 8322/ 9402 88.51% ( 90.27%), 4 ( 0.05%) clones, pur 99.46%, hit eff 96.37%
long_fromB_P>5GeV_AND_Pt>1GeV : 2089/ 2284 91.46% ( 91.45%), 2 ( 0.10%) clones, pur 99.49%, hit eff 96.87%
11_noVelo_UT : 0/ 8097 0.00% ( 0.00%), 0 ( 0.00%) clones, pur -nan%, hit eff -nan%
12_noVelo_UT_P>5GeV : 0/ 3231 0.00% ( 0.00%), 0 ( 0.00%) clones, pur -nan%, hit eff -nan%
13_long_PT>2GeV : 2039/ 2270 89.82% ( 90.70%), 1 ( 0.05%) clones, pur 99.48%, hit eff 96.69%
14_long_from_B_PT>2GeV : 917/ 997 91.98% ( 92.33%), 0 ( 0.00%) clones, pur 99.43%, hit eff 96.71%
15_long_strange_P>5GeV : 1254/ 1609 77.94% ( 80.52%), 0 ( 0.00%) clones, pur 99.35%, hit eff 96.75%
16_long_strange_P>5GeV_PT>500MeV : 474/ 558 84.95% ( 85.59%), 0 ( 0.00%) clones, pur 99.12%, hit eff 96.24%
Forward then matching (hlt1_pp_forward_then_matching_no_ut_no_gec_validation):
long_validator validation:
TrackChecker output : 6614/ 59801 11.06% ghosts
for P>3GeV,Pt>0.5GeV : 2560/ 28257 9.06% ghosts
01_long : 49736/ 75133 66.20% ( 68.28%), 111 ( 0.22%) clones, pur 99.47%, hit eff 95.76%
02_long_P>5GeV : 38963/ 47271 82.42% ( 85.01%), 111 ( 0.28%) clones, pur 99.48%, hit eff 96.29%
03_long_strange : 1871/ 3460 54.08% ( 54.73%), 1 ( 0.05%) clones, pur 99.36%, hit eff 95.74%
04_long_strange_P>5GeV : 1255/ 1609 78.00% ( 80.05%), 1 ( 0.08%) clones, pur 99.33%, hit eff 96.51%
05_long_fromB : 3663/ 4413 83.00% ( 84.05%), 25 ( 0.68%) clones, pur 99.49%, hit eff 96.56%
06_long_fromB_P>5GeV : 3337/ 3641 91.65% ( 92.14%), 25 ( 0.74%) clones, pur 99.50%, hit eff 96.81%
07_long_electrons : 1872/ 5807 32.24% ( 33.51%), 4 ( 0.21%) clones, pur 98.95%, hit eff 95.47%
08_long_electrons_P>5GeV : 1404/ 2901 48.40% ( 50.93%), 4 ( 0.28%) clones, pur 98.98%, hit eff 96.05%
09_long_fromB_electrons : 102/ 212 48.11% ( 51.08%), 2 ( 1.92%) clones, pur 99.29%, hit eff 96.04%
10_long_fromB_electrons_P>5GeV : 89/ 138 64.49% ( 68.58%), 2 ( 2.20%) clones, pur 99.35%, hit eff 96.26%
long_P>5GeV_AND_Pt>1GeV : 8776/ 9402 93.34% ( 93.92%), 90 ( 1.02%) clones, pur 99.54%, hit eff 96.89%
long_fromB_P>5GeV_AND_Pt>1GeV : 2162/ 2284 94.66% ( 94.15%), 24 ( 1.10%) clones, pur 99.53%, hit eff 97.23%
11_noVelo_UT : 0/ 8097 0.00% ( 0.00%), 0 ( 0.00%) clones, pur -nan%, hit eff -nan%
12_noVelo_UT_P>5GeV : 0/ 3231 0.00% ( 0.00%), 0 ( 0.00%) clones, pur -nan%, hit eff -nan%
13_long_PT>2GeV : 2151/ 2270 94.76% ( 95.49%), 19 ( 0.88%) clones, pur 99.57%, hit eff 97.26%
14_long_from_B_PT>2GeV : 949/ 997 95.19% ( 95.50%), 10 ( 1.04%) clones, pur 99.49%, hit eff 97.18%
15_long_strange_P>5GeV : 1255/ 1609 78.00% ( 80.05%), 1 ( 0.08%) clones, pur 99.33%, hit eff 96.51%
16_long_strange_P>5GeV_PT>500MeV : 480/ 558 86.02% ( 86.88%), 1 ( 0.21%) clones, pur 99.13%, hit eff 95.84%Memory usage of the sequence increases. Again, this is not due to the introduced algorithms but rather due to the increase in tracks.
Changelist:
- Algorithms
create_reduced_scifi_hit_container_tandcombine_long_track_containers_tadded. - Added
resizeargument operation, which allows resizing a HOST or DEVICE output insideoperator(). - Added
track_container_offset()function to Velo, UT and SciFi consolidated Track types. - Changed
sizebytestosize_bytes, which is more standardized. - Make Scheduler never configure the release of HOST variables.
Goes with Moore!1967 (merged)