Implement a spacepoint weight cut on seeds

[stephenswat]

This PR implements a cut that requires a seed to be the best seed for at least $n$ of its constituent spacepoints, which mirrors the cut that is used in ACTS. This cut is a specialisation of the cut in #1082, implemented in a way that should allow it to run faster.

[stephenswat]

Depends on #1097.

[stephenswat]

I discussed this with @andiwand and I found out that the cut that ACTS implements isn't actually quite this, but rather a greedy algorithm that updates after every step. That means that the ACTS implementation is essentially impossible to efficiently reproduce on GPU.

[stephenswat]

This now depends on #1113; if weights are strictly positive we can add a relative minimum weight to recover the efficiencies.

[sonarqubecloud]

Quality Gate passed

Issues
1 New issue
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud

[stephenswat]

Physics performance summary

Here is a summary of the physics performance effects of this PR. Command used:

traccc_seeding_example_cuda --input-directory=/data/Acts/odd-simulations-20240506/geant4_ttbar_mu200 --digitization-file=geometries/odd/odd-digi-geometric-config.json --detector-file=geometries/odd/odd-detray_geometry_detray.json --grid-file=geometries/odd/odd-detray_surface_grids_detray.json --use-detray-detector=on --input-events=10 --use-acts-geom-source=on --check-performance --truth-finding-min-track-candidates=5 --truth-finding-min-pt=1.0 --truth-finding-min-z=-150 --truth-finding-max-z=150 --truth-finding-max-r=10 --seed-matching-ratio=0.99 --track-matching-ratio=0.5 --track-candidates-range=5:100 --seedfinder-vertex-range=-150:150

Seeding performance

Total number of seeds went from 336240 to 264849 (-21.2%)

Track finding performance

Total number of found tracks went from 153828 to 126058 (-18.1%)

Note

This is an automated message produced on the explicit request of a human being.

[stephenswat]

Performance summary

Here is a summary of the performance effects of this PR:

Graphical

Tabular

Kernel	a2dbda9	640ef09	Delta
fit_forward	32.08 ms	24.04 ms	-25.1%
propagate_to_next_surface	28.66 ms	18.84 ms	-34.3%
fit_backward	13.53 ms	9.68 ms	-28.4%
count_triplets	2.06 ms	2.06 ms	0.1%
find_tracks	2.69 ms	2.00 ms	-25.6%
count_doublets	1.01 ms	1.00 ms	-0.2%
ccl_kernel	871.35 μs	871.86 μs	0.1%
find_doublets	801.36 μs	795.18 μs	-0.8%
find_triplets	571.92 μs	580.58 μs	1.5%
Thrust::sort	457.09 μs	441.24 μs	-3.5%
fit_prelude	197.19 μs	138.39 μs	-29.8%
select_seeds	102.54 μs	125.56 μs	22.5%
remove_duplicates	170.60 μs	122.82 μs	-28.0%
collect_highest_weight_per_spacepoint	—	47.34 μs	—
build_tracks	54.78 μs	39.64 μs	-27.6%
apply_interaction	42.24 μs	30.79 μs	-27.1%
update_triplet_weights	29.87 μs	29.74 μs	-0.4%
populate_grid	23.88 μs	23.84 μs	-0.2%
estimate_track_params	38.07 μs	23.23 μs	-39.0%
count_grid_capacities	22.59 μs	22.60 μs	0.0%
unknown	20.48 μs	20.42 μs	-0.3%
fill_finding_propagation_sort_keys	26.01 μs	18.64 μs	-28.3%
form_spacepoints	12.47 μs	12.53 μs	0.5%
fill_finding_duplicate_removal_sort_keys	10.78 μs	8.33 μs	-22.7%
reduce_triplet_counts	6.83 μs	6.86 μs	0.4%
make_barcode_sequence	1.01 μs	1.01 μs	-0.3%
fill_fitting_sort_keys	1.15 μs	900.02 ns	-21.6%
fill_prefix_sum	171.94 ns	171.95 ns	0.0%
Total	83.49 ms	60.99 ms	-26.9%

Important

All metrics in this report are given as reciprocal throughput, not as wallclock runtime.

Warning

At least one kernel incurred a significant performance regression.

Note

This is an automated message produced on the explicit request of a human being.