Skip to content

Conversation

@niermann999
Copy link
Contributor

@niermann999 niermann999 commented Nov 21, 2025

Since it just retrieves the first measurement of the seed, the first Kalman gain matrix call can be skipped

@stephenswat

This comment was marked as outdated.

@stephenswat

This comment was marked as outdated.

@niermann999 niermann999 force-pushed the feat-sjip-first-measurement branch from 67919d3 to 3826cfb Compare November 27, 2025 14:09
@stephenswat

This comment was marked as outdated.

@niermann999 niermann999 force-pushed the feat-sjip-first-measurement branch from 3826cfb to b2e90b5 Compare November 27, 2025 14:30
@stephenswat

This comment was marked as outdated.

@niermann999 niermann999 force-pushed the feat-sjip-first-measurement branch from b2e90b5 to 743cd94 Compare November 27, 2025 14:47
@niermann999 niermann999 force-pushed the feat-sjip-first-measurement branch 2 times, most recently from 1acf19e to 703ac82 Compare December 4, 2025 15:53
@stephenswat
Copy link
Member

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 --material-file=geometries/odd/odd-detray_material_detray.json --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 298345 to 298345 (+0.0%)

Seeding plots



Track finding performance

Total number of found tracks went from 50185 to 50194 (+0.0%)

Finding plots









Track fitting performance

Fitting plots





















Seeding to track finding relative performance

Seeding to track finding plots



Track finding to track fitting relative performance

Track finding to track fitting plots



Note

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

@stephenswat
Copy link
Member

Performance summary

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

Graphical

Tabular

KernelReciprocal ThroughputParallelism
a244e2d703ac82Deltaa244e2d703ac82
propagate_to_next_surface8.39 ms8.37 ms-0.2%3.483.49
fit_backward2.55 ms2.55 ms0.1%7.347.33
fit_forward1.99 ms1.99 ms0.3%7.347.34
find_tracks1.39 ms1.16 ms-16.3%1.811.86
ccl_kernel823.85 μs824.51 μs0.1%1.371.37
count_doublets628.35 μs617.58 μs-1.7%1.611.61
count_triplets590.86 μs588.64 μs-0.4%1.021.02
find_doublets451.25 μs450.01 μs-0.3%3.083.08
Thrust::sort384.93 μs385.04 μs0.0%7.367.36
find_triplets173.41 μs171.75 μs-1.0%1.311.32
select_seeds53.48 μs53.33 μs-0.3%1.341.34
remove_duplicates23.60 μs23.56 μs-0.2%26.1026.08
populate_grid23.49 μs23.45 μs-0.2%1.221.22
count_grid_capacities22.22 μs22.20 μs-0.1%1.221.22
fill_sorted_measurements19.71 μs19.74 μs0.2%1.131.13
update_triplet_weights15.05 μs15.11 μs0.4%1.271.27
apply_interaction14.26 μs14.26 μs-0.0%6.726.71
estimate_track_params14.22 μs14.25 μs0.2%2.152.15
fit_prelude9.80 μs9.81 μs0.1%21.7721.78
fill_finding_propagation_sort_keys8.84 μs8.85 μs0.2%7.617.59
form_spacepoints7.86 μs7.85 μs-0.2%1.481.48
reduce_triplet_counts6.28 μs6.28 μs-0.0%3.083.08
unknown5.04 μs5.05 μs0.2%4.284.27
build_tracks4.12 μs4.11 μs-0.1%14.7814.77
fill_finding_duplicate_removal_sort_keys1.57 μs1.57 μs0.2%37.9737.93
fill_prefix_sum172.00 ns171.98 ns-0.0%341.30341.30
fill_fitting_sort_keys153.43 ns153.56 ns0.1%22.0822.08
Total17.59 ms17.34 ms-1.4%4.194.22

Important

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

Note

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

@stephenswat

This comment was marked as outdated.

@niermann999 niermann999 force-pushed the feat-sjip-first-measurement branch from c4305df to 0de875c Compare December 4, 2025 16:35
@stephenswat
Copy link
Member

Performance summary

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

Graphical

Tabular

KernelReciprocal ThroughputParallelism
a244e2d0de875cDeltaa244e2d0de875c
propagate_to_next_surface8.38 ms8.33 ms-0.5%3.493.45
fit_backward2.55 ms2.16 ms-15.1%7.347.83
fit_forward1.98 ms1.77 ms-10.5%7.357.83
find_tracks1.38 ms1.15 ms-16.9%1.811.90
ccl_kernel826.14 μs825.59 μs-0.1%1.371.37
count_doublets623.83 μs624.21 μs0.1%1.611.61
count_triplets590.41 μs591.80 μs0.2%1.021.02
find_doublets446.30 μs445.42 μs-0.2%3.083.08
Thrust::sort384.70 μs384.16 μs-0.1%7.367.34
find_triplets172.58 μs171.77 μs-0.5%1.311.31
select_seeds53.48 μs53.75 μs0.5%1.341.34
populate_grid23.48 μs23.46 μs-0.1%1.221.22
remove_duplicates23.58 μs22.60 μs-4.2%26.0627.03
count_grid_capacities22.16 μs22.18 μs0.1%1.221.22
fill_sorted_measurements19.80 μs19.78 μs-0.1%1.131.13
update_triplet_weights15.06 μs14.98 μs-0.5%1.271.27
estimate_track_params14.27 μs14.30 μs0.2%2.152.15
apply_interaction14.26 μs14.15 μs-0.8%6.726.72
fit_prelude9.85 μs8.76 μs-11.0%21.8023.26
fill_finding_propagation_sort_keys8.86 μs8.76 μs-1.1%7.607.62
form_spacepoints7.94 μs7.92 μs-0.3%1.491.48
reduce_triplet_counts6.29 μs6.28 μs-0.2%3.083.08
unknown5.06 μs5.05 μs-0.1%4.274.28
build_tracks4.10 μs3.89 μs-5.0%14.7815.83
fill_finding_duplicate_removal_sort_keys1.57 μs1.50 μs-4.6%37.9339.45
fill_prefix_sum171.94 ns171.97 ns0.0%341.30341.30
fill_fitting_sort_keys152.73 ns143.43 ns-6.1%22.1123.56
Total17.56 ms16.69 ms-5.0%4.194.22

Important

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

Note

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

@stephenswat
Copy link
Member

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 --material-file=geometries/odd/odd-detray_material_detray.json --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 298345 to 298342 (-0.0%)

Seeding plots



Track finding performance

Total number of found tracks went from 50179 to 47951 (-4.4%)

Finding plots









Track fitting performance

Fitting plots





















Seeding to track finding relative performance

Seeding to track finding plots



Track finding to track fitting relative performance

Track finding to track fitting plots



Note

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

@niermann999 niermann999 force-pushed the feat-sjip-first-measurement branch from 0de875c to 66ac85b Compare December 5, 2025 13:51
@stephenswat
Copy link
Member

Performance summary

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

Graphical

Tabular

KernelReciprocal ThroughputParallelism
a244e2d66ac85bDeltaa244e2d66ac85b
propagate_to_next_surface8.39 ms8.46 ms0.9%3.483.57
fit_backward2.54 ms2.58 ms1.4%7.347.14
fit_forward1.98 ms2.14 ms7.8%7.347.14
find_tracks1.38 ms1.43 ms3.8%1.811.81
ccl_kernel824.33 μs824.09 μs-0.0%1.371.37
count_doublets625.34 μs618.70 μs-1.1%1.611.61
count_triplets590.36 μs588.27 μs-0.4%1.021.02
find_doublets449.16 μs446.93 μs-0.5%3.073.08
Thrust::sort384.97 μs385.55 μs0.2%7.367.39
find_triplets172.12 μs172.01 μs-0.1%1.321.31
select_seeds53.66 μs53.22 μs-0.8%1.341.34
remove_duplicates23.62 μs24.16 μs2.3%26.0925.42
populate_grid23.44 μs23.46 μs0.1%1.221.22
count_grid_capacities22.59 μs22.19 μs-1.8%1.221.22
fill_sorted_measurements19.77 μs19.84 μs0.3%1.131.13
update_triplet_weights15.08 μs15.16 μs0.5%1.271.27
apply_interaction14.40 μs14.36 μs-0.3%6.736.75
estimate_track_params14.24 μs14.32 μs0.6%2.152.15
fit_prelude9.83 μs10.68 μs8.7%21.7721.24
fill_finding_propagation_sort_keys8.82 μs8.97 μs1.6%7.627.60
form_spacepoints7.97 μs7.92 μs-0.6%1.481.48
reduce_triplet_counts6.33 μs6.30 μs-0.5%3.083.08
unknown5.05 μs5.07 μs0.4%4.274.27
build_tracks4.10 μs4.28 μs4.5%14.7614.47
fill_finding_duplicate_removal_sort_keys1.56 μs1.61 μs3.2%38.1437.32
fill_prefix_sum171.89 ns171.88 ns-0.0%341.30341.30
fill_fitting_sort_keys152.61 ns157.67 ns3.3%22.0821.48
Total17.57 ms17.87 ms1.7%4.184.20

Important

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

Note

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

@stephenswat
Copy link
Member

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 --material-file=geometries/odd/odd-detray_material_detray.json --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 298345 to 298342 (-0.0%)

Seeding plots



Track finding performance

Total number of found tracks went from 50179 to 50478 (+0.6%)

Finding plots









Track fitting performance

Fitting plots





















Seeding to track finding relative performance

Seeding to track finding plots



Track finding to track fitting relative performance

Track finding to track fitting plots



Note

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

@niermann999 niermann999 force-pushed the feat-sjip-first-measurement branch from 66ac85b to 2d959e0 Compare December 5, 2025 15:24
@sonarqubecloud
Copy link

sonarqubecloud bot commented Dec 5, 2025

@stephenswat
Copy link
Member

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 --material-file=geometries/odd/odd-detray_material_detray.json --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 298342 to 298341 (-0.0%)

Seeding plots



Track finding performance

Total number of found tracks went from 50190 to 50252 (+0.1%)

Finding plots









Track fitting performance

Fitting plots





















Seeding to track finding relative performance

Seeding to track finding plots



Track finding to track fitting relative performance

Track finding to track fitting plots



Note

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

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants