Skip to content

[BUG FIX] Skip contact pruning for degenerated buckets. #2836

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Draft
duburcqa wants to merge 1 commit into
base: main
Choose a base branch
from

Skip pruning for degenerated contact buckets.

02d7003
Select commit
Loading
Failed to load commit list.
Sign in for the full log view
Draft

[BUG FIX] Skip contact pruning for degenerated buckets. #2836

Skip pruning for degenerated contact buckets.
02d7003
Select commit
Loading
Failed to load commit list.
GitHub Actions / Benchmark Comparison succeeded May 28, 2026 in 0s

Benchmark Comparison

✅ No regressions detected. See tables below.

Details

Thresholds: runtime ± 8%, compile ± 30%, mem ± 4%

Runtime FPS

status env batch_size use_contact_island dtype backend gjk_collision constraint_solver current FPS baseline FPS [last (mean ± std)] (*1) Δ FPS (*2)
✅ anymal_random 30000 False field cuda - - 9,399,346 9,571,715 (9,455,587 ± 63,053) -1.8%
✅ anymal_random 30000 False ndarray cuda - - 7,671,405 7,718,057 (7,830,758 ± 148,614) -0.6%
✅ anymal_uniform 30000 False field cuda - - 12,415,188 12,301,404 (12,361,856 ± 133,917) +0.9%
✅ anymal_uniform 30000 False ndarray cuda - - 10,633,052 10,432,232 (10,527,428 ± 120,607) +1.9%
✅ anymal_uniform_kinematic 0 False field cpu - - 1,565 1,608 (1,586 ± 17) -2.7%
✅ anymal_uniform_kinematic 0 False ndarray cpu - - 940 964 (961 ± 8) -2.5%
✅ anymal_uniform_kinematic 30000 False field cuda - - 10,357,517 10,484,751 (10,406,225 ± 200,139) -1.2%
✅ anymal_uniform_kinematic 30000 False ndarray cuda - - 8,246,875 8,272,285 (8,239,523 ± 55,544) -0.3%
✅ anymal_zero 0 False field cpu - - 5,108 5,034 (5,059 ± 29) +1.5%
✅ anymal_zero 0 False ndarray cpu - - 3,124 3,118 (3,103 ± 13) +0.2%
✅ anymal_zero 30000 False field cuda - - 19,293,157 19,187,597 (19,315,923 ± 241,533) +0.6%
✅ anymal_zero 30000 False ndarray cuda - - 14,302,023 14,731,008 (14,534,886 ± 139,293) -2.9%
✅ box_pyramid_3 4096 False field cuda - - 1,554,745 1,566,977 (1,569,662 ± 17,173) -0.8%
✅ box_pyramid_3 4096 False ndarray cuda - - 1,109,694 1,132,121 (1,119,245 ± 8,325) -2.0%
✅ box_pyramid_4 4096 False field cuda - - 618,930 625,794 (621,446 ± 3,658) -1.1%
✅ box_pyramid_4 4096 False ndarray cuda - - 470,470 473,658 (470,580 ± 4,283) -0.7%
✅ box_pyramid_5 4096 False field cuda - - 217,426 217,197 (216,534 ± 808) +0.1%
✅ box_pyramid_5 4096 False ndarray cuda - - 173,138 173,558 (171,864 ± 2,417) -0.2%
✅ box_pyramid_6 4096 False field cuda False - 84,060 83,906 (83,079 ± 502) +0.2%
✅ box_pyramid_6 4096 False field cuda True - 87,058 86,771 (86,344 ± 675) +0.3%
✅ box_pyramid_6 4096 False ndarray cuda False - 71,821 72,994 (71,629 ± 1,037) -1.6%
✅ box_pyramid_6 4096 False ndarray cuda True - 73,554 74,669 (73,241 ± 1,025) -1.5%
✅ dex_hand 4096 False field cuda - - 24,556 25,335 (25,319 ± 167) -3.1%
✅ dex_hand 4096 False ndarray cuda - - 19,061 19,681 (19,327 ± 182) -3.2%
✅ duck_in_box_easy 30000 False field cuda False - 26,827,839 27,218,416 (26,810,599 ± 426,628) -1.4%
✅ duck_in_box_easy 30000 False field cuda True - 9,532,126 9,768,504 (9,638,895 ± 147,179) -2.4%
✅ duck_in_box_easy 30000 False ndarray cuda False - 20,160,883 20,109,181 (20,191,015 ± 127,565) +0.3%
✅ duck_in_box_easy 30000 False ndarray cuda True - 8,600,261 8,452,205 (8,463,033 ± 71,543) +1.8%
✅ duck_in_box_hard 0 False field cpu - - 3,997 4,003 (3,980 ± 25) -0.1%
✅ duck_in_box_hard 0 False ndarray cpu - - 2,689 2,708 (2,698 ± 37) -0.7%
✅ duck_in_box_hard 30000 False field cuda False - 10,720,805 10,969,343 (10,736,619 ± 125,956) -2.3%
✅ duck_in_box_hard 30000 False field cuda True - 3,305,821 3,251,795 (3,304,161 ± 45,787) +1.7%
✅ duck_in_box_hard 30000 False ndarray cuda False - 7,918,084 8,135,530 (7,971,738 ± 131,651) -2.7%
✅ duck_in_box_hard 30000 False ndarray cuda True - 2,728,734 2,857,094 (2,776,174 ± 45,387) -4.5%
✅ franka 30000 False field cuda - - 22,937,276 22,849,973 (22,373,178 ± 402,859) +0.4%
✅ franka 30000 False ndarray cuda - - 16,410,025 16,694,099 (16,399,939 ± 367,869) -1.7%
✅ franka_accessors 0 False field cpu - - 1,001 1,014 (1,006 ± 17) -1.3%
✅ franka_accessors 0 False ndarray cpu - - 786 816 (816 ± 16) -3.7%
✅ franka_accessors 30000 False field cuda - - 15,292,891 15,408,616 (15,263,310 ± 119,230) -0.8%
✅ franka_accessors 30000 False ndarray cuda - - 11,256,496 11,609,379 (11,361,044 ± 164,077) -3.0%
✅ franka_free 30000 False field cuda - - 32,931,705 33,609,214 (33,127,158 ± 394,385) -2.0%
✅ franka_free 30000 False ndarray cuda - - 21,130,291 21,114,552 (20,984,838 ± 134,296) +0.1%
✅ franka_random 0 False field cpu - - 4,423 4,457 (4,466 ± 32) -0.8%
✅ franka_random 0 False ndarray cpu - - 2,799 2,815 (2,847 ± 49) -0.6%
✅ franka_random 30000 False field cuda False - 17,012,174 16,686,276 (16,655,709 ± 263,790) +2.0%
✅ franka_random 30000 False field cuda True - 11,716,274 11,468,730 (11,447,665 ± 71,820) +2.2%
✅ franka_random 30000 False field cuda - CG 16,740,499 16,598,155 (16,674,191 ± 171,463) +0.9%
✅ franka_random 30000 False field cuda - Newton 16,549,627 16,909,629 (16,620,678 ± 202,701) -2.1%
✅ franka_random 30000 False ndarray cuda False - 12,475,164 12,785,875 (12,545,138 ± 125,759) -2.4%
✅ franka_random 30000 False ndarray cuda True - 8,714,448 8,843,781 (8,678,165 ± 91,967) -1.5%
✅ franka_random 30000 False ndarray cuda - CG 12,406,479 12,510,147 (12,570,312 ± 291,312) -0.8%
✅ franka_random 30000 False ndarray cuda - Newton 12,414,521 12,500,737 (12,650,850 ± 265,320) -0.7%
✅ g1_fall 4096 False field cuda - Newton 1,573,960 1,614,989 (1,617,449 ± 9,651) -2.5%
✅ g1_fall 4096 False ndarray cuda - Newton 1,117,079 1,143,268 (1,153,031 ± 8,548) -2.3%
✅ go2 4096 False field cuda False CG 4,671,094 4,635,553 (4,635,303 ± 36,410) +0.8%
✅ go2 4096 False field cuda False Newton 5,492,835 5,608,229 (5,490,291 ± 64,886) -2.1%
✅ go2 4096 False field cuda True - 3,751,346 3,747,260 (3,729,764 ± 37,288) +0.1%
✅ go2 4096 False ndarray cuda False CG 3,067,002 3,062,247 (3,071,425 ± 14,810) +0.2%
✅ go2 4096 False ndarray cuda False Newton 3,417,603 3,426,440 (3,441,942 ± 23,907) -0.3%
✅ go2 4096 False ndarray cuda True - 2,484,971 2,484,457 (2,470,472 ± 15,251) +0.0%
✅ shadow_hand_cubes 0 False field cpu - - 39 39 (39 ± 0) +0.0%
✅ shadow_hand_cubes 0 False ndarray cpu - - 31 31 (31 ± 0) +0.0%
✅ shadow_hand_cubes_sparse 0 False field cpu - - 62 62 (61 ± 0) +0.0%
✅ shadow_hand_cubes_sparse 0 False ndarray cpu - - 43 43 (42 ± 0) +0.0%

Baselines considered: 5 commits

Compile Time

status env batch_size use_contact_island dtype backend gjk_collision constraint_solver current compile baseline compile [last (mean ± std)] (*1) Δ compile (*2)
✅ anymal_random 30000 False field cuda - - 75.50 75.90 (76.00 ± 0.45) -0.5%
✅ anymal_random 30000 False ndarray cuda - - 69 69 (69 ± 1) +0.0%
✅ anymal_uniform 30000 False field cuda - - 75.60 75.60 (74.04 ± 1.43) +0.0%
✅ anymal_uniform 30000 False ndarray cuda - - 68.50 66.70 (68.52 ± 1.28) +2.7%
✅ anymal_uniform_kinematic 0 False field cpu - - 40.30 41.10 (40.14 ± 0.50) -1.9%
✅ anymal_uniform_kinematic 0 False ndarray cpu - - 34.90 34.50 (35.20 ± 0.66) +1.2%
✅ anymal_uniform_kinematic 30000 False field cuda - - 74.80 74.00 (74.90 ± 0.64) +1.1%
✅ anymal_uniform_kinematic 30000 False ndarray cuda - - 68.30 68.90 (68.06 ± 0.84) -0.9%
✅ anymal_zero 0 False field cpu - - 36.60 35.90 (36.02 ± 0.36) +1.9%
✅ anymal_zero 0 False ndarray cpu - - 31.30 30.80 (31.10 ± 0.16) +1.6%
✅ anymal_zero 30000 False field cuda - - 75 75 (76 ± 0) -1.1%
✅ anymal_zero 30000 False ndarray cuda - - 65.90 68.40 (69.06 ± 0.45) -3.7%
✅ box_pyramid_3 4096 False field cuda - - 72.50 73.60 (74.96 ± 1.60) -1.5%
✅ box_pyramid_3 4096 False ndarray cuda - - 69.60 70.30 (68.32 ± 1.77) -1.0%
✅ box_pyramid_4 4096 False field cuda - - 81.60 79.90 (87.96 ± 5.85) +2.1%
✅ box_pyramid_4 4096 False ndarray cuda - - 73.90 74.20 (79.42 ± 6.10) -0.4%
✅ box_pyramid_5 4096 False field cuda - - 92.80 93.50 (89.26 ± 5.71) -0.7%
✅ box_pyramid_5 4096 False ndarray cuda - - 88 86 (79 ± 6) +2.1%
✅ box_pyramid_6 4096 False field cuda False - 83.40 93.60 (91.52 ± 4.82) -10.9%
✅ box_pyramid_6 4096 False field cuda True - 72.70 60.40 (69.92 ± 4.67) +20.4%
✅ box_pyramid_6 4096 False ndarray cuda False - 74.90 85.80 (79.74 ± 4.91) -12.7%
✅ box_pyramid_6 4096 False ndarray cuda True - 65.30 54.40 (57.18 ± 4.45) +20.0%
✅ dex_hand 4096 False field cuda - - 103.10 102.20 (104.72 ± 1.76) +0.9%
✅ dex_hand 4096 False ndarray cuda - - 97.30 95.90 (96.54 ± 0.49) +1.5%
✅ duck_in_box_easy 30000 False field cuda False - 71.60 71.20 (73.36 ± 1.12) +0.6%
✅ duck_in_box_easy 30000 False field cuda True - 51.10 50.90 (50.82 ± 0.35) +0.4%
✅ duck_in_box_easy 30000 False ndarray cuda False - 68.30 69.00 (67.84 ± 1.26) -1.0%
✅ duck_in_box_easy 30000 False ndarray cuda True - 45.20 44.10 (44.78 ± 0.51) +2.5%
✅ duck_in_box_hard 0 False field cpu - - 39.20 38.80 (39.06 ± 0.23) +1.0%
✅ duck_in_box_hard 0 False ndarray cpu - - 34.70 34.30 (34.32 ± 0.35) +1.2%
✅ duck_in_box_hard 30000 False field cuda False - 72.20 73.00 (72.30 ± 0.97) -1.1%
✅ duck_in_box_hard 30000 False field cuda True - 51.20 50.70 (51.54 ± 0.64) +1.0%
✅ duck_in_box_hard 30000 False ndarray cuda False - 64.70 66.80 (66.20 ± 0.43) -3.1%
✅ duck_in_box_hard 30000 False ndarray cuda True - 44.50 44.70 (45.60 ± 0.67) -0.4%
✅ franka 30000 False field cuda - - 71.80 70.10 (70.38 ± 0.42) +2.4%
✅ franka 30000 False ndarray cuda - - 63.90 63.00 (63.10 ± 0.42) +1.4%
✅ franka_accessors 0 False field cpu - - 36.70 36.60 (36.46 ± 0.30) +0.3%
✅ franka_accessors 0 False ndarray cpu - - 32.50 31.60 (32.28 ± 0.59) +2.8%
✅ franka_accessors 30000 False field cuda - - 71.10 72.10 (71.34 ± 1.18) -1.4%
✅ franka_accessors 30000 False ndarray cuda - - 62.80 63.90 (63.78 ± 0.49) -1.7%
✅ franka_free 30000 False field cuda - - 69.20 69.30 (69.46 ± 0.92) -0.1%
✅ franka_free 30000 False ndarray cuda - - 62.50 62.70 (62.78 ± 0.34) -0.3%
✅ franka_random 0 False field cpu - - 35.90 35.80 (36.14 ± 0.33) +0.3%
✅ franka_random 0 False ndarray cpu - - 31.20 30.80 (31.16 ± 0.43) +1.3%
✅ franka_random 30000 False field cuda False - 69.80 68.50 (69.36 ± 0.49) +1.9%
✅ franka_random 30000 False field cuda True - 49.20 48.40 (48.76 ± 0.36) +1.7%
✅ franka_random 30000 False field cuda - CG 69.70 68.90 (69.48 ± 0.73) +1.2%
✅ franka_random 30000 False field cuda - Newton 69.50 70.50 (70.56 ± 0.33) -1.4%
✅ franka_random 30000 False ndarray cuda False - 62.90 63.40 (63.32 ± 0.76) -0.8%
✅ franka_random 30000 False ndarray cuda True - 41.90 42.40 (41.92 ± 0.24) -1.2%
✅ franka_random 30000 False ndarray cuda - CG 61.50 62.10 (62.92 ± 0.65) -1.0%
✅ franka_random 30000 False ndarray cuda - Newton 63.40 62.90 (63.50 ± 0.69) +0.8%
✅ g1_fall 4096 False field cuda - Newton 75.50 70.80 (74.46 ± 2.63) +6.6%
✅ g1_fall 4096 False ndarray cuda - Newton 65.60 69.30 (67.88 ± 1.31) -5.3%
✅ go2 4096 False field cuda False CG 69.70 68.90 (68.90 ± 0.49) +1.2%
✅ go2 4096 False field cuda False Newton 92.40 82.00 (84.52 ± 4.65) +12.7%
✅ go2 4096 False field cuda True - 59 72 (69 ± 4) -18.4%
✅ go2 4096 False ndarray cuda False CG 63.60 63.70 (63.48 ± 0.65) -0.2%
✅ go2 4096 False ndarray cuda False Newton 73.90 86.40 (82.34 ± 6.24) -14.5%
✅ go2 4096 False ndarray cuda True - 53.30 65.00 (58.46 ± 5.52) -18.0%
✅ shadow_hand_cubes 0 False field cpu - - 37.70 37.70 (37.94 ± 0.29) +0.0%
✅ shadow_hand_cubes 0 False ndarray cpu - - 32.80 33.30 (32.96 ± 0.29) -1.5%
✅ shadow_hand_cubes_sparse 0 False field cpu - - 37.30 37.30 (37.06 ± 0.33) +0.0%
✅ shadow_hand_cubes_sparse 0 False ndarray cpu - - 32.70 32.40 (32.40 ± 0.39) +0.9%

Baselines considered: 5 commits

Memory usage

status env batch_size backend dtype gjk_collision constraint_solver current memory baseline memory [last (mean ± std)] (*1) Δ memory (*2)
✅ anymal_random 30000 gpu field - - 12,690 12,690 (12,690 ± 0) +0.0%
✅ anymal_random 30000 gpu ndarray - - 12,658 12,658 (12,659 ± 3) +0.0%
✅ anymal_uniform 30000 gpu field - - 12,690 12,690 (12,690 ± 0) +0.0%
✅ anymal_uniform 30000 gpu ndarray - - 12,658 12,658 (12,658 ± 0) +0.0%
✅ anymal_uniform_kinematic 0 cpu field - - 952 952 (952 ± 0) +0.0%
✅ anymal_uniform_kinematic 0 cpu ndarray - - 984 1,016 (1,009 ± 12) -3.1%
✅ anymal_uniform_kinematic 30000 gpu field - - 12,962 13,026 (13,026 ± 0) -0.5%
✅ anymal_uniform_kinematic 30000 gpu ndarray - - 13,026 13,026 (13,026 ± 0) +0.0%
✅ anymal_zero 0 cpu field - - 920 920 (920 ± 0) +0.0%
✅ anymal_zero 0 cpu ndarray - - 1,016 1,016 (1,016 ± 0) +0.0%
✅ anymal_zero 30000 gpu field - - 12,690 12,690 (12,690 ± 0) +0.0%
✅ anymal_zero 30000 gpu ndarray - - 12,658 12,570 (12,640 ± 34) +0.7%
✅ box_pyramid_3 4096 gpu field - - 2,048 2,048 (2,048 ± 0) +0.0%
✅ box_pyramid_3 4096 gpu ndarray - - 2,048 2,048 (2,048 ± 0) +0.0%
✅ box_pyramid_4 4096 gpu field - - 2,982 2,950 (2,956 ± 12) +1.1%
✅ box_pyramid_4 4096 gpu ndarray - - 2,950 2,950 (2,956 ± 12) +0.0%
✅ box_pyramid_5 4096 gpu field - - 6,598 6,598 (6,598 ± 0) +0.0%
✅ box_pyramid_5 4096 gpu ndarray - - 6,598 6,598 (6,598 ± 0) +0.0%
✅ box_pyramid_6 4096 gpu field False - 9,628 9,628 (9,628 ± 0) +0.0%
✅ box_pyramid_6 4096 gpu field True - 10,182 10,182 (10,182 ± 0) +0.0%
✅ box_pyramid_6 4096 gpu ndarray False - 9,628 9,628 (9,628 ± 0) +0.0%
✅ box_pyramid_6 4096 gpu ndarray True - 10,182 10,182 (10,182 ± 0) +0.0%
✅ dex_hand 4096 gpu field - - 9,082 9,082 (9,082 ± 0) +0.0%
✅ dex_hand 4096 gpu ndarray - - 9,050 9,078 (9,066 ± 13) -0.3%
✅ duck_in_box_easy 30000 gpu field False - 3,386 3,378 (3,378 ± 0) +0.2%
✅ duck_in_box_easy 30000 gpu field True - 4,026 4,018 (4,018 ± 0) +0.2%
✅ duck_in_box_easy 30000 gpu ndarray False - 3,378 3,378 (3,379 ± 3) +0.0%
✅ duck_in_box_easy 30000 gpu ndarray True - 3,986 3,986 (3,986 ± 0) +0.0%
✅ duck_in_box_hard 0 cpu field - - 8 8 (8 ± 0) +0.0%
✅ duck_in_box_hard 0 cpu ndarray - - 8 8 (8 ± 0) +0.0%
✅ duck_in_box_hard 30000 gpu field False - 13,746 13,746 (13,749 ± 3) +0.0%
✅ duck_in_box_hard 30000 gpu field True - 14,386 14,394 (14,387 ± 3) -0.1%
✅ duck_in_box_hard 30000 gpu ndarray False - 13,746 13,746 (13,747 ± 3) +0.0%
✅ duck_in_box_hard 30000 gpu ndarray True - 14,354 14,386 (14,379 ± 12) -0.2%
✅ franka 30000 gpu field - - 16,580 16,580 (16,580 ± 0) +0.0%
✅ franka 30000 gpu ndarray - - 16,612 16,548 (16,567 ± 15) +0.4%
✅ franka_accessors 0 cpu field - - 1,304 1,304 (1,304 ± 0) +0.0%
✅ franka_accessors 0 cpu ndarray - - 1,400 1,400 (1,400 ± 0) +0.0%
✅ franka_accessors 30000 gpu field - - 16,572 16,700 (16,610 ± 50) -0.8%
✅ franka_accessors 30000 gpu ndarray - - 16,540 16,540 (16,572 ± 48) +0.0%
✅ franka_free 30000 gpu field - - 16,196 16,196 (16,213 ± 21) +0.0%
✅ franka_free 30000 gpu ndarray - - 16,196 16,196 (16,185 ± 17) +0.0%
✅ franka_random 0 cpu field - - 1,304 1,304 (1,304 ± 0) +0.0%
✅ franka_random 0 cpu ndarray - - 1,400 1,400 (1,400 ± 0) +0.0%
✅ franka_random 30000 gpu field False - 16,196 16,196 (16,226 ± 41) +0.0%
✅ franka_random 30000 gpu field True - 16,890 16,890 (16,915 ± 36) +0.0%
✅ franka_random 30000 gpu field - CG 16,226 16,290 (16,290 ± 39) -0.4%
✅ franka_random 30000 gpu field - Newton 16,578 16,578 (16,616 ± 50) +0.0%
✅ franka_random 30000 gpu ndarray False - 16,164 16,196 (16,189 ± 12) -0.2%
✅ franka_random 30000 gpu ndarray True - 16,890 16,890 (16,890 ± 0) +0.0%
✅ franka_random 30000 gpu ndarray - CG 16,226 16,194 (16,219 ± 23) +0.2%
✅ franka_random 30000 gpu ndarray - Newton 16,578 16,546 (16,565 ± 25) +0.2%
✅ g1_fall 4096 gpu field - Newton 4,768 4,768 (4,768 ± 0) +0.0%
✅ g1_fall 4096 gpu ndarray - Newton 4,768 4,768 (4,768 ± 0) +0.0%
✅ go2 4096 gpu field False CG 3,616 3,616 (3,616 ± 0) +0.0%
✅ go2 4096 gpu field False Newton 3,968 3,968 (3,968 ± 0) +0.0%
✅ go2 4096 gpu field True - 4,280 4,280 (4,280 ± 0) +0.0%
✅ go2 4096 gpu ndarray False CG 3,616 3,616 (3,616 ± 0) +0.0%
✅ go2 4096 gpu ndarray False Newton 3,968 3,968 (3,968 ± 0) +0.0%
✅ go2 4096 gpu ndarray True - 4,280 4,280 (4,273 ± 12) +0.0%
✅ shadow_hand_cubes 0 cpu field - - 1,344 1,344 (1,344 ± 0) +0.0%
✅ shadow_hand_cubes 0 cpu ndarray - - 1,440 1,408 (1,420 ± 15) +2.3%
✅ shadow_hand_cubes_sparse 0 cpu field - - 1,304 1,304 (1,304 ± 0) +0.0%
✅ shadow_hand_cubes_sparse 0 cpu ndarray - - 1,400 1,400 (1,400 ± 0) +0.0%

Baselines considered: 5 commits

  • Commit 1: 6007f26

  • Commit 2: d7fabbb

  • Commit 3: 192b528

  • Commit 4: 78dc129

  • Commit 5: 93a064f

  • (*1) last: last commit on main, mean/std: stats over commit hashes 5 commits if available.

  • (*2) Δ: relative difference between PR and last commit on main, i.e. (PR - main) / main * 100%.

Artifact: Download raw data

View more details on GitHub Actions