Skip to content

[BugFix] Fix Habitat CI by upgrading to Python 3.10 and building habitat-sim from source#3346

Merged
vmoens merged 13 commits into
mainfrom
fix-habitat-claude2
Jan 20, 2026
Merged

[BugFix] Fix Habitat CI by upgrading to Python 3.10 and building habitat-sim from source#3346
vmoens merged 13 commits into
mainfrom
fix-habitat-claude2

Conversation

@vmoens

@vmoens vmoens commented Jan 19, 2026

Copy link
Copy Markdown
Collaborator

Summary

  • Updates the Habitat CI workflow to use Python 3.10 (tensordict now requires Python >= 3.10)
  • Switches from conda-based habitat-sim installation to building from source via pip (conda packages only support Python < 3.10)
  • Adds additional build dependencies for building habitat-sim from source (ninja-build, EGL dev libraries, etc.)

Test plan

  • Verify Habitat CI job runs successfully
  • Verify habitat-sim builds correctly from source
  • Verify torchrl and tensordict install correctly with Python 3.10
  • Verify Habitat tests pass

@vmoens
[BugFix] Fix Habitat CI by upgrading to Python 3.10 and building habi…
40bd933 
…tat-sim from source

tensordict now requires Python >= 3.10, but habitat-sim conda packages
only support Python < 3.10. This change:

- Updates the Habitat CI workflow to use Python 3.10
- Switches from conda-based habitat-sim installation to building from source via pip
- Adds additional build dependencies (ninja-build, EGL dev libraries, etc.)
@pytorch-bot

pytorch-bot Bot commented Jan 19, 2026 •
edited
Loading

Copy link
Copy Markdown

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/rl/3346

Note: Links to docs will display an error until the docs builds have been completed.

❌ 1 New Failure, 13 Pending

As of commit f29da05 with merge base a7e4b69 (image):

NEW FAILURE - The following job has failed:

This comment was automatically generated by Dr. CI and updates every 15 minutes.

@meta-cla meta-cla Bot added the CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed. label Jan 19, 2026
vmoens added 4 commits January 19, 2026 15:15
@vmoens
[BugFix] Use system cmake instead of pip cmake for habitat-sim build
bf2f668 
pip cmake 4.x is incompatible with habitat-sim's CMake files that use
old cmake_minimum_required. Using system cmake 3.22 instead.
@vmoens
[BugFix] Remove libstdcxx-ng installation that conflicts with PyTorch
11909ad 
The conda installation of libstdcxx-ng=12 was causing PyTorch C extensions
to fail to load. This was needed for conda-based habitat-sim but is not
needed when building from source.
@vmoens
[BugFix] Download Habitat test datasets after building habitat-sim
e869b56 
The pip-based habitat-sim installation doesn't include the test datasets
that were bundled with conda packages. Download replica_cad_dataset and
rearrange_task_assets for the tests to work.
@vmoens
[BugFix] Update git-lfs to v3.4.0 for dataset download compatibility
eb7a315 
The habitat dataset download uses `git lfs prune -f --recent` which
requires git-lfs >= 3.0. Updated from v2.9.0 to v3.4.0.
@vmoens vmoens added the bug Something isn't working label Jan 19, 2026
@vmoens
[BugFix] Fix git-lfs installation path for v3.4.0
8ba5624 
The v3.4.0 tar file extracts to git-lfs-3.4.0/ directory,
not directly to git-lfs.
@github-actions

github-actions Bot commented Jan 19, 2026 •
edited
Loading

Copy link
Copy Markdown
Contributor

$\color{#D29922}\textsf{\Large&amp;#x26A0;\kern{0.2cm}\normalsize Warning}$ Result of GPU Benchmark Tests

Total Benchmarks: 148. Improved: $\large\color{#35bf28}18$. Worsened: $\large\color{#d91a1a}8$.

Expand to view detailed results
Name Max Mean Ops Ops on Repo HEAD Change
test_tensor_to_bytestream_speed[pickle] 81.3310μs 80.4185μs 12.4349 KOps/s 12.4664 KOps/s $\color{#d91a1a}-0.25\%$
test_tensor_to_bytestream_speed[torch.save] 0.1399ms 0.1397ms 7.1568 KOps/s 7.1658 KOps/s $\color{#d91a1a}-0.13\%$
test_tensor_to_bytestream_speed[untyped_storage] 0.1073s 0.1072s 9.3270 Ops/s 9.2218 Ops/s $\color{#35bf28}+1.14\%$
test_tensor_to_bytestream_speed[numpy] 2.4317μs 2.4268μs 412.0589 KOps/s 408.2537 KOps/s $\color{#35bf28}+0.93\%$
test_tensor_to_bytestream_speed[safetensors] 38.0509μs 37.0932μs 26.9591 KOps/s 27.3410 KOps/s $\color{#d91a1a}-1.40\%$
test_simple 0.9082s 0.8140s 1.2285 Ops/s 1.2085 Ops/s $\color{#35bf28}+1.65\%$
test_transformed 1.5251s 1.4326s 0.6980 Ops/s 0.6927 Ops/s $\color{#35bf28}+0.76\%$
test_serial 2.3984s 2.3058s 0.4337 Ops/s 0.4315 Ops/s $\color{#35bf28}+0.51\%$
test_parallel 2.0993s 2.0244s 0.4940 Ops/s 0.5104 Ops/s $\color{#d91a1a}-3.21\%$
test_step_mdp_speed[True-True-True-True-True] 0.2393ms 46.1662μs 21.6609 KOps/s 22.0853 KOps/s $\color{#d91a1a}-1.92\%$
test_step_mdp_speed[True-True-True-True-False] 56.6420μs 25.2265μs 39.6408 KOps/s 39.0103 KOps/s $\color{#35bf28}+1.62\%$
test_step_mdp_speed[True-True-True-False-True] 62.4310μs 26.0277μs 38.4206 KOps/s 39.3895 KOps/s $\color{#d91a1a}-2.46\%$
test_step_mdp_speed[True-True-True-False-False] 50.7110μs 14.0009μs 71.4237 KOps/s 70.4343 KOps/s $\color{#35bf28}+1.40\%$
test_step_mdp_speed[True-True-False-True-True] 85.5220μs 48.5752μs 20.5866 KOps/s 20.3868 KOps/s $\color{#35bf28}+0.98\%$
test_step_mdp_speed[True-True-False-True-False] 59.6620μs 27.9034μs 35.8379 KOps/s 35.9891 KOps/s $\color{#d91a1a}-0.42\%$
test_step_mdp_speed[True-True-False-False-True] 65.8610μs 28.2463μs 35.4028 KOps/s 35.6222 KOps/s $\color{#d91a1a}-0.62\%$
test_step_mdp_speed[True-True-False-False-False] 45.6310μs 16.3295μs 61.2390 KOps/s 58.3177 KOps/s $\textbf{\color{#35bf28}+5.01\%}$
test_step_mdp_speed[True-False-True-True-True] 89.9620μs 52.0904μs 19.1974 KOps/s 19.4688 KOps/s $\color{#d91a1a}-1.39\%$
test_step_mdp_speed[True-False-True-True-False] 53.7010μs 31.2518μs 31.9982 KOps/s 32.0262 KOps/s $\color{#d91a1a}-0.09\%$
test_step_mdp_speed[True-False-True-False-True] 65.8010μs 29.0762μs 34.3924 KOps/s 36.0038 KOps/s $\color{#d91a1a}-4.48\%$
test_step_mdp_speed[True-False-True-False-False] 44.8310μs 16.9480μs 59.0039 KOps/s 59.6561 KOps/s $\color{#d91a1a}-1.09\%$
test_step_mdp_speed[True-False-False-True-True] 83.5620μs 54.5888μs 18.3188 KOps/s 18.7449 KOps/s $\color{#d91a1a}-2.27\%$
test_step_mdp_speed[True-False-False-True-False] 65.0920μs 33.7068μs 29.6676 KOps/s 30.4847 KOps/s $\color{#d91a1a}-2.68\%$
test_step_mdp_speed[True-False-False-False-True] 68.4410μs 31.6953μs 31.5504 KOps/s 33.6830 KOps/s $\textbf{\color{#d91a1a}-6.33\%}$
test_step_mdp_speed[True-False-False-False-False] 63.9920μs 19.6189μs 50.9712 KOps/s 51.6868 KOps/s $\color{#d91a1a}-1.38\%$
test_step_mdp_speed[False-True-True-True-True] 89.3220μs 51.7659μs 19.3178 KOps/s 19.7584 KOps/s $\color{#d91a1a}-2.23\%$
test_step_mdp_speed[False-True-True-True-False] 58.8620μs 30.6697μs 32.6054 KOps/s 32.8586 KOps/s $\color{#d91a1a}-0.77\%$
test_step_mdp_speed[False-True-True-False-True] 69.3510μs 32.1233μs 31.1300 KOps/s 32.3033 KOps/s $\color{#d91a1a}-3.63\%$
test_step_mdp_speed[False-True-True-False-False] 46.3410μs 18.8118μs 53.1581 KOps/s 53.9222 KOps/s $\color{#d91a1a}-1.42\%$
test_step_mdp_speed[False-True-False-True-True] 2.7125ms 55.0130μs 18.1775 KOps/s 18.3685 KOps/s $\color{#d91a1a}-1.04\%$
test_step_mdp_speed[False-True-False-True-False] 62.4410μs 33.9043μs 29.4948 KOps/s 29.9453 KOps/s $\color{#d91a1a}-1.50\%$
test_step_mdp_speed[False-True-False-False-True] 68.6210μs 34.8063μs 28.7305 KOps/s 29.3517 KOps/s $\color{#d91a1a}-2.12\%$
test_step_mdp_speed[False-True-False-False-False] 57.4010μs 21.2425μs 47.0755 KOps/s 46.6778 KOps/s $\color{#35bf28}+0.85\%$
test_step_mdp_speed[False-False-True-True-True] 93.9020μs 57.4683μs 17.4009 KOps/s 17.6388 KOps/s $\color{#d91a1a}-1.35\%$
test_step_mdp_speed[False-False-True-True-False] 93.4520μs 36.9942μs 27.0312 KOps/s 27.5741 KOps/s $\color{#d91a1a}-1.97\%$
test_step_mdp_speed[False-False-True-False-True] 65.0610μs 35.4956μs 28.1725 KOps/s 28.8567 KOps/s $\color{#d91a1a}-2.37\%$
test_step_mdp_speed[False-False-True-False-False] 50.5210μs 21.3942μs 46.7416 KOps/s 47.2231 KOps/s $\color{#d91a1a}-1.02\%$
test_step_mdp_speed[False-False-False-True-True] 85.2920μs 59.4046μs 16.8337 KOps/s 16.7052 KOps/s $\color{#35bf28}+0.77\%$
test_step_mdp_speed[False-False-False-True-False] 76.2420μs 39.0319μs 25.6201 KOps/s 25.6127 KOps/s $\color{#35bf28}+0.03\%$
test_step_mdp_speed[False-False-False-False-True] 69.0320μs 37.1675μs 26.9052 KOps/s 26.8813 KOps/s $\color{#35bf28}+0.09\%$
test_step_mdp_speed[False-False-False-False-False] 61.4110μs 24.0514μs 41.5776 KOps/s 41.5550 KOps/s $\color{#35bf28}+0.05\%$
test_non_tensor_env_rollout_speed[1000-single-True] 0.8635s 0.7665s 1.3047 Ops/s 1.3035 Ops/s $\color{#35bf28}+0.09\%$
test_non_tensor_env_rollout_speed[1000-single-False] 0.7220s 0.6286s 1.5909 Ops/s 1.5717 Ops/s $\color{#35bf28}+1.22\%$
test_non_tensor_env_rollout_speed[1000-serial-no-buffers-True] 1.7447s 1.6695s 0.5990 Ops/s 0.5970 Ops/s $\color{#35bf28}+0.33\%$
test_non_tensor_env_rollout_speed[1000-serial-no-buffers-False] 1.5264s 1.4545s 0.6875 Ops/s 0.6854 Ops/s $\color{#35bf28}+0.30\%$
test_non_tensor_env_rollout_speed[1000-serial-buffers-True] 1.9974s 1.9195s 0.5210 Ops/s 0.5198 Ops/s $\color{#35bf28}+0.22\%$
test_non_tensor_env_rollout_speed[1000-serial-buffers-False] 1.7840s 1.6989s 0.5886 Ops/s 0.5847 Ops/s $\color{#35bf28}+0.67\%$
test_non_tensor_env_rollout_speed[1000-parallel-no-buffers-True] 4.7520s 4.6393s 0.2155 Ops/s 0.2153 Ops/s $\color{#35bf28}+0.10\%$
test_non_tensor_env_rollout_speed[1000-parallel-no-buffers-False] 4.4739s 4.4060s 0.2270 Ops/s 0.2236 Ops/s $\color{#35bf28}+1.48\%$
test_non_tensor_env_rollout_speed[1000-parallel-buffers-True] 2.0429s 1.9581s 0.5107 Ops/s 0.5112 Ops/s $\color{#d91a1a}-0.10\%$
test_non_tensor_env_rollout_speed[1000-parallel-buffers-False] 1.7849s 1.6716s 0.5982 Ops/s 0.6010 Ops/s $\color{#d91a1a}-0.46\%$
test_values[generalized_advantage_estimate-True-True] 20.6260ms 20.1104ms 49.7255 Ops/s 48.8872 Ops/s $\color{#35bf28}+1.71\%$
test_values[vec_generalized_advantage_estimate-True-True] 0.1452s 3.8208ms 261.7236 Ops/s 281.9253 Ops/s $\textbf{\color{#d91a1a}-7.17\%}$
test_values[td0_return_estimate-False-False] 0.1056ms 82.2260μs 12.1616 KOps/s 12.1330 KOps/s $\color{#35bf28}+0.24\%$
test_values[td1_return_estimate-False-False] 49.0253ms 48.3676ms 20.6750 Ops/s 20.1516 Ops/s $\color{#35bf28}+2.60\%$
test_values[vec_td1_return_estimate-False-False] 1.3360ms 1.0829ms 923.4239 Ops/s 922.5701 Ops/s $\color{#35bf28}+0.09\%$
test_values[td_lambda_return_estimate-True-False] 82.1886ms 79.2898ms 12.6120 Ops/s 11.8437 Ops/s $\textbf{\color{#35bf28}+6.49\%}$
test_values[vec_td_lambda_return_estimate-True-False] 1.2807ms 1.0777ms 927.9032 Ops/s 918.5116 Ops/s $\color{#35bf28}+1.02\%$
test_gae_speed[generalized_advantage_estimate-False-1-512] 21.3950ms 20.5212ms 48.7300 Ops/s 44.8499 Ops/s $\textbf{\color{#35bf28}+8.65\%}$
test_gae_speed[vec_generalized_advantage_estimate-True-1-512] 1.0299ms 0.7535ms 1.3271 KOps/s 1.3156 KOps/s $\color{#35bf28}+0.87\%$
test_gae_speed[vec_generalized_advantage_estimate-False-1-512] 0.7861ms 0.6767ms 1.4778 KOps/s 1.3994 KOps/s $\textbf{\color{#35bf28}+5.60\%}$
test_gae_speed[vec_generalized_advantage_estimate-True-32-512] 1.5367ms 1.4842ms 673.7713 Ops/s 669.3209 Ops/s $\color{#35bf28}+0.66\%$
test_gae_speed[vec_generalized_advantage_estimate-False-32-512] 0.7843ms 0.7134ms 1.4018 KOps/s 1.4123 KOps/s $\color{#d91a1a}-0.74\%$
test_dqn_speed[False-None] 1.6960ms 1.5457ms 646.9558 Ops/s 643.9441 Ops/s $\color{#35bf28}+0.47\%$
test_dqn_speed[False-backward] 2.2736ms 2.1682ms 461.2077 Ops/s 454.3641 Ops/s $\color{#35bf28}+1.51\%$
test_dqn_speed[True-None] 0.6523ms 0.5926ms 1.6873 KOps/s 1.6523 KOps/s $\color{#35bf28}+2.12\%$
test_dqn_speed[True-backward] 1.4152ms 1.2275ms 814.6888 Ops/s 812.4151 Ops/s $\color{#35bf28}+0.28\%$
test_dqn_speed[reduce-overhead-None] 0.6483ms 0.5922ms 1.6886 KOps/s 1.5766 KOps/s $\textbf{\color{#35bf28}+7.10\%}$
test_ddpg_speed[False-None] 3.2581ms 2.8935ms 345.5994 Ops/s 344.0234 Ops/s $\color{#35bf28}+0.46\%$
test_ddpg_speed[False-backward] 4.6230ms 4.2974ms 232.6993 Ops/s 230.7705 Ops/s $\color{#35bf28}+0.84\%$
test_ddpg_speed[True-None] 1.4757ms 1.3470ms 742.4122 Ops/s 734.3799 Ops/s $\color{#35bf28}+1.09\%$
test_ddpg_speed[True-backward] 2.6578ms 2.5827ms 387.1912 Ops/s 384.8939 Ops/s $\color{#35bf28}+0.60\%$
test_ddpg_speed[reduce-overhead-None] 1.5904ms 1.3973ms 715.6612 Ops/s 720.3885 Ops/s $\color{#d91a1a}-0.66\%$
test_sac_speed[False-None] 8.9600ms 8.5650ms 116.7540 Ops/s 117.6178 Ops/s $\color{#d91a1a}-0.73\%$
test_sac_speed[False-backward] 12.2300ms 11.4418ms 87.3988 Ops/s 84.5695 Ops/s $\color{#35bf28}+3.35\%$
test_sac_speed[True-None] 1.9565ms 1.8782ms 532.4286 Ops/s 530.5221 Ops/s $\color{#35bf28}+0.36\%$
test_sac_speed[True-backward] 3.6361ms 3.5238ms 283.7845 Ops/s 276.7921 Ops/s $\color{#35bf28}+2.53\%$
test_sac_speed[reduce-overhead-None] 19.0638ms 10.6515ms 93.8839 Ops/s 95.1869 Ops/s $\color{#d91a1a}-1.37\%$
test_redq_deprec_speed[False-None] 9.9007ms 9.3500ms 106.9514 Ops/s 105.2000 Ops/s $\color{#35bf28}+1.66\%$
test_redq_deprec_speed[False-backward] 13.0198ms 12.4996ms 80.0023 Ops/s 78.6291 Ops/s $\color{#35bf28}+1.75\%$
test_redq_deprec_speed[True-None] 3.0690ms 2.5833ms 387.0990 Ops/s 383.4437 Ops/s $\color{#35bf28}+0.95\%$
test_redq_deprec_speed[True-backward] 4.8631ms 4.4129ms 226.6102 Ops/s 222.1520 Ops/s $\color{#35bf28}+2.01\%$
test_redq_deprec_speed[reduce-overhead-None] 15.3867ms 9.5453ms 104.7637 Ops/s 88.0339 Ops/s $\textbf{\color{#35bf28}+19.00\%}$
test_td3_speed[False-None] 8.2985ms 8.1997ms 121.9556 Ops/s 119.3850 Ops/s $\color{#35bf28}+2.15\%$
test_td3_speed[False-backward] 11.1665ms 10.8065ms 92.5365 Ops/s 90.0765 Ops/s $\color{#35bf28}+2.73\%$
test_td3_speed[True-None] 1.7474ms 1.7045ms 586.6789 Ops/s 581.9559 Ops/s $\color{#35bf28}+0.81\%$
test_td3_speed[True-backward] 3.5502ms 3.4093ms 293.3132 Ops/s 290.1936 Ops/s $\color{#35bf28}+1.08\%$
test_td3_speed[reduce-overhead-None] 46.5780ms 23.4088ms 42.7190 Ops/s 41.3965 Ops/s $\color{#35bf28}+3.19\%$
test_cql_speed[False-None] 17.7738ms 17.3754ms 57.5526 Ops/s 57.0823 Ops/s $\color{#35bf28}+0.82\%$
test_cql_speed[False-backward] 23.5329ms 23.0226ms 43.4356 Ops/s 42.9453 Ops/s $\color{#35bf28}+1.14\%$
test_cql_speed[True-None] 3.5411ms 3.4303ms 291.5160 Ops/s 285.0441 Ops/s $\color{#35bf28}+2.27\%$
test_cql_speed[True-backward] 6.7570ms 5.8065ms 172.2217 Ops/s 170.2681 Ops/s $\color{#35bf28}+1.15\%$
test_cql_speed[reduce-overhead-None] 18.8563ms 11.8127ms 84.6549 Ops/s 85.5624 Ops/s $\color{#d91a1a}-1.06\%$
test_a2c_speed[False-None] 3.3874ms 3.2693ms 305.8733 Ops/s 307.4546 Ops/s $\color{#d91a1a}-0.51\%$
test_a2c_speed[False-backward] 6.9351ms 6.4909ms 154.0614 Ops/s 153.1208 Ops/s $\color{#35bf28}+0.61\%$
test_a2c_speed[True-None] 1.4343ms 1.3600ms 735.2869 Ops/s 727.5087 Ops/s $\color{#35bf28}+1.07\%$
test_a2c_speed[True-backward] 3.2456ms 3.2122ms 311.3138 Ops/s 310.9292 Ops/s $\color{#35bf28}+0.12\%$
test_a2c_speed[reduce-overhead-None] 1.1146ms 0.9575ms 1.0444 KOps/s 1.0354 KOps/s $\color{#35bf28}+0.87\%$
test_ppo_speed[False-None] 3.9199ms 3.8295ms 261.1338 Ops/s 258.0243 Ops/s $\color{#35bf28}+1.21\%$
test_ppo_speed[False-backward] 7.6554ms 7.2386ms 138.1478 Ops/s 137.9178 Ops/s $\color{#35bf28}+0.17\%$
test_ppo_speed[True-None] 1.4976ms 1.4144ms 707.0257 Ops/s 689.4620 Ops/s $\color{#35bf28}+2.55\%$
test_ppo_speed[True-backward] 3.4140ms 3.3739ms 296.3968 Ops/s 291.3008 Ops/s $\color{#35bf28}+1.75\%$
test_ppo_speed[reduce-overhead-None] 1.0780ms 1.0202ms 980.1942 Ops/s 948.0852 Ops/s $\color{#35bf28}+3.39\%$
test_reinforce_speed[False-None] 2.4819ms 2.3240ms 430.2957 Ops/s 434.9046 Ops/s $\color{#d91a1a}-1.06\%$
test_reinforce_speed[False-backward] 3.8740ms 3.4755ms 287.7268 Ops/s 288.4319 Ops/s $\color{#d91a1a}-0.24\%$
test_reinforce_speed[True-None] 1.4316ms 1.2816ms 780.2454 Ops/s 747.1476 Ops/s $\color{#35bf28}+4.43\%$
test_reinforce_speed[True-backward] 3.2142ms 3.1695ms 315.5064 Ops/s 313.1567 Ops/s $\color{#35bf28}+0.75\%$
test_reinforce_speed[reduce-overhead-None] 0.4605s 10.0579ms 99.4247 Ops/s 95.9491 Ops/s $\color{#35bf28}+3.62\%$
test_iql_speed[False-None] 10.0237ms 9.4630ms 105.6750 Ops/s 104.3044 Ops/s $\color{#35bf28}+1.31\%$
test_iql_speed[False-backward] 13.9832ms 13.5326ms 73.8957 Ops/s 72.8815 Ops/s $\color{#35bf28}+1.39\%$
test_iql_speed[True-None] 2.6802ms 2.2664ms 441.2262 Ops/s 435.5747 Ops/s $\color{#35bf28}+1.30\%$
test_iql_speed[True-backward] 5.5470ms 5.0923ms 196.3737 Ops/s 192.8012 Ops/s $\color{#35bf28}+1.85\%$
test_iql_speed[reduce-overhead-None] 0.7046s 12.5103ms 79.9339 Ops/s 100.1059 Ops/s $\textbf{\color{#d91a1a}-20.15\%}$
test_rb_sample[TensorDictReplayBuffer-ListStorage-RandomSampler-4000] 6.5275ms 6.0623ms 164.9543 Ops/s 164.0083 Ops/s $\color{#35bf28}+0.58\%$
test_rb_sample[TensorDictReplayBuffer-LazyMemmapStorage-RandomSampler-10000] 0.7153ms 0.3124ms 3.2008 KOps/s 2.8016 KOps/s $\textbf{\color{#35bf28}+14.25\%}$
test_rb_sample[TensorDictReplayBuffer-LazyTensorStorage-RandomSampler-10000] 0.7280ms 0.2645ms 3.7809 KOps/s 3.4282 KOps/s $\textbf{\color{#35bf28}+10.29\%}$
test_rb_sample[TensorDictReplayBuffer-ListStorage-SamplerWithoutReplacement-4000] 6.5509ms 5.8293ms 171.5470 Ops/s 172.8220 Ops/s $\color{#d91a1a}-0.74\%$
test_rb_sample[TensorDictReplayBuffer-LazyMemmapStorage-SamplerWithoutReplacement-10000] 2.1850ms 0.3693ms 2.7075 KOps/s 3.0136 KOps/s $\textbf{\color{#d91a1a}-10.16\%}$
test_rb_sample[TensorDictReplayBuffer-LazyTensorStorage-SamplerWithoutReplacement-10000] 0.4741ms 0.2624ms 3.8113 KOps/s 3.0101 KOps/s $\textbf{\color{#35bf28}+26.61\%}$
test_rb_sample[TensorDictReplayBuffer-LazyMemmapStorage-sampler6-10000] 1.6720ms 1.4401ms 694.4106 Ops/s 693.3350 Ops/s $\color{#35bf28}+0.16\%$
test_rb_sample[TensorDictReplayBuffer-LazyTensorStorage-sampler7-10000] 1.4287ms 1.2208ms 819.1660 Ops/s 745.9775 Ops/s $\textbf{\color{#35bf28}+9.81\%}$
test_rb_sample[TensorDictPrioritizedReplayBuffer-ListStorage-None-4000] 6.1915ms 5.9761ms 167.3326 Ops/s 166.9310 Ops/s $\color{#35bf28}+0.24\%$
test_rb_sample[TensorDictPrioritizedReplayBuffer-LazyMemmapStorage-None-10000] 1.0413ms 0.4857ms 2.0589 KOps/s 2.0492 KOps/s $\color{#35bf28}+0.47\%$
test_rb_sample[TensorDictPrioritizedReplayBuffer-LazyTensorStorage-None-10000] 0.6281ms 0.4099ms 2.4398 KOps/s 1.9905 KOps/s $\textbf{\color{#35bf28}+22.58\%}$
test_rb_iterate[TensorDictReplayBuffer-ListStorage-RandomSampler-4000] 6.0131ms 5.8494ms 170.9583 Ops/s 170.8901 Ops/s $\color{#35bf28}+0.04\%$
test_rb_iterate[TensorDictReplayBuffer-LazyMemmapStorage-RandomSampler-10000] 1.6373ms 0.2858ms 3.4989 KOps/s 3.1386 KOps/s $\textbf{\color{#35bf28}+11.48\%}$
test_rb_iterate[TensorDictReplayBuffer-LazyTensorStorage-RandomSampler-10000] 0.4577ms 0.2657ms 3.7641 KOps/s 3.0159 KOps/s $\textbf{\color{#35bf28}+24.81\%}$
test_rb_iterate[TensorDictReplayBuffer-ListStorage-SamplerWithoutReplacement-4000] 6.3242ms 5.7658ms 173.4359 Ops/s 172.8319 Ops/s $\color{#35bf28}+0.35\%$
test_rb_iterate[TensorDictReplayBuffer-LazyMemmapStorage-SamplerWithoutReplacement-10000] 2.3230ms 0.3971ms 2.5184 KOps/s 3.0593 KOps/s $\textbf{\color{#d91a1a}-17.68\%}$
test_rb_iterate[TensorDictReplayBuffer-LazyTensorStorage-SamplerWithoutReplacement-10000] 0.5860ms 0.3750ms 2.6670 KOps/s 2.8082 KOps/s $\textbf{\color{#d91a1a}-5.03\%}$
test_rb_iterate[TensorDictPrioritizedReplayBuffer-ListStorage-None-4000] 6.3827ms 5.9683ms 167.5506 Ops/s 168.1954 Ops/s $\color{#d91a1a}-0.38\%$
test_rb_iterate[TensorDictPrioritizedReplayBuffer-LazyMemmapStorage-None-10000] 2.2907ms 0.4395ms 2.2751 KOps/s 2.0527 KOps/s $\textbf{\color{#35bf28}+10.83\%}$
test_rb_iterate[TensorDictPrioritizedReplayBuffer-LazyTensorStorage-None-10000] 0.6115ms 0.4166ms 2.4004 KOps/s 2.3784 KOps/s $\color{#35bf28}+0.93\%$
test_rb_populate[TensorDictReplayBuffer-ListStorage-RandomSampler-400] 6.8729ms 5.0549ms 197.8270 Ops/s 48.4793 Ops/s $\textbf{\color{#35bf28}+308.06\%}$
test_rb_populate[TensorDictReplayBuffer-LazyMemmapStorage-RandomSampler-400] 4.4402ms 2.1676ms 461.3373 Ops/s 596.0226 Ops/s $\textbf{\color{#d91a1a}-22.60\%}$
test_rb_populate[TensorDictReplayBuffer-LazyTensorStorage-RandomSampler-400] 1.1026ms 0.9485ms 1.0543 KOps/s 802.5600 Ops/s $\textbf{\color{#35bf28}+31.37\%}$
test_rb_populate[TensorDictReplayBuffer-ListStorage-SamplerWithoutReplacement-400] 0.5919s 16.8699ms 59.2771 Ops/s 195.5395 Ops/s $\textbf{\color{#d91a1a}-69.69\%}$
test_rb_populate[TensorDictReplayBuffer-LazyMemmapStorage-SamplerWithoutReplacement-400] 5.3517ms 1.9165ms 521.7792 Ops/s 480.0797 Ops/s $\textbf{\color{#35bf28}+8.69\%}$
test_rb_populate[TensorDictReplayBuffer-LazyTensorStorage-SamplerWithoutReplacement-400] 2.3717ms 1.1368ms 879.6435 Ops/s 875.3112 Ops/s $\color{#35bf28}+0.49\%$
test_rb_populate[TensorDictPrioritizedReplayBuffer-ListStorage-None-400] 9.5222ms 5.3447ms 187.1001 Ops/s 188.1440 Ops/s $\color{#d91a1a}-0.55\%$
test_rb_populate[TensorDictPrioritizedReplayBuffer-LazyMemmapStorage-None-400] 4.1536ms 1.9777ms 505.6349 Ops/s 516.2901 Ops/s $\color{#d91a1a}-2.06\%$
test_rb_populate[TensorDictPrioritizedReplayBuffer-LazyTensorStorage-None-400] 1.3584ms 1.1092ms 901.5405 Ops/s 879.1265 Ops/s $\color{#35bf28}+2.55\%$
test_rb_extend_sample[ReplayBuffer-LazyTensorStorage-RandomSampler-10000-10000-100-True] 39.0690ms 34.3803ms 29.0864 Ops/s 20.8107 Ops/s $\textbf{\color{#35bf28}+39.77\%}$
test_rb_extend_sample[ReplayBuffer-LazyTensorStorage-RandomSampler-10000-10000-100-False] 19.7165ms 17.9113ms 55.8307 Ops/s 56.8935 Ops/s $\color{#d91a1a}-1.87\%$
test_rb_extend_sample[ReplayBuffer-LazyTensorStorage-RandomSampler-100000-10000-100-True] 39.1366ms 35.3836ms 28.2617 Ops/s 28.4289 Ops/s $\color{#d91a1a}-0.59\%$
test_rb_extend_sample[ReplayBuffer-LazyTensorStorage-RandomSampler-100000-10000-100-False] 19.4006ms 17.8848ms 55.9133 Ops/s 55.5836 Ops/s $\color{#35bf28}+0.59\%$
test_rb_extend_sample[ReplayBuffer-LazyTensorStorage-RandomSampler-1000000-10000-100-True] 38.1900ms 37.0219ms 27.0110 Ops/s 26.9441 Ops/s $\color{#35bf28}+0.25\%$
test_rb_extend_sample[ReplayBuffer-LazyTensorStorage-RandomSampler-1000000-10000-100-False] 21.0465ms 19.3377ms 51.7126 Ops/s 51.5135 Ops/s $\color{#35bf28}+0.39\%$

@github-actions

github-actions Bot commented Jan 19, 2026 •
edited
Loading

Copy link
Copy Markdown
Contributor

$\color{#D29922}\textsf{\Large&amp;#x26A0;\kern{0.2cm}\normalsize Warning}$ Result of CPU Benchmark Tests

Total Benchmarks: 153. Improved: $\large\color{#35bf28}13$. Worsened: $\large\color{#d91a1a}9$.

Expand to view detailed results
Name Max Mean Ops Ops on Repo HEAD Change
test_tensor_to_bytestream_speed[pickle] 81.2837μs 80.3086μs 12.4520 KOps/s 12.2235 KOps/s $\color{#35bf28}+1.87\%$
test_tensor_to_bytestream_speed[torch.save] 0.1454ms 0.1408ms 7.1029 KOps/s 7.1398 KOps/s $\color{#d91a1a}-0.52\%$
test_tensor_to_bytestream_speed[untyped_storage] 0.1097s 0.1094s 9.1366 Ops/s 9.1938 Ops/s $\color{#d91a1a}-0.62\%$
test_tensor_to_bytestream_speed[numpy] 2.5325μs 2.5187μs 397.0233 KOps/s 394.7655 KOps/s $\color{#35bf28}+0.57\%$
test_tensor_to_bytestream_speed[safetensors] 38.0290μs 37.7856μs 26.4651 KOps/s 26.4116 KOps/s $\color{#35bf28}+0.20\%$
test_simple 0.6569s 0.5653s 1.7690 Ops/s 1.7502 Ops/s $\color{#35bf28}+1.07\%$
test_transformed 1.2285s 1.1365s 0.8799 Ops/s 0.8764 Ops/s $\color{#35bf28}+0.40\%$
test_serial 1.7783s 1.6776s 0.5961 Ops/s 0.5972 Ops/s $\color{#d91a1a}-0.19\%$
test_parallel 1.1937s 1.1066s 0.9036 Ops/s 0.9139 Ops/s $\color{#d91a1a}-1.13\%$
test_step_mdp_speed[True-True-True-True-True] 0.2440ms 44.4256μs 22.5095 KOps/s 22.3583 KOps/s $\color{#35bf28}+0.68\%$
test_step_mdp_speed[True-True-True-True-False] 54.9810μs 25.1661μs 39.7360 KOps/s 40.3270 KOps/s $\color{#d91a1a}-1.47\%$
test_step_mdp_speed[True-True-True-False-True] 68.8120μs 25.3310μs 39.4774 KOps/s 39.7981 KOps/s $\color{#d91a1a}-0.81\%$
test_step_mdp_speed[True-True-True-False-False] 52.0210μs 13.9471μs 71.6997 KOps/s 72.4350 KOps/s $\color{#d91a1a}-1.02\%$
test_step_mdp_speed[True-True-False-True-True] 81.5820μs 48.9275μs 20.4384 KOps/s 20.8355 KOps/s $\color{#d91a1a}-1.91\%$
test_step_mdp_speed[True-True-False-True-False] 63.9110μs 28.3508μs 35.2724 KOps/s 36.4006 KOps/s $\color{#d91a1a}-3.10\%$
test_step_mdp_speed[True-True-False-False-True] 61.5420μs 28.4650μs 35.1308 KOps/s 35.0318 KOps/s $\color{#35bf28}+0.28\%$
test_step_mdp_speed[True-True-False-False-False] 50.7810μs 16.8849μs 59.2243 KOps/s 61.0160 KOps/s $\color{#d91a1a}-2.94\%$
test_step_mdp_speed[True-False-True-True-True] 86.3620μs 52.2129μs 19.1524 KOps/s 19.4455 KOps/s $\color{#d91a1a}-1.51\%$
test_step_mdp_speed[True-False-True-True-False] 59.8110μs 31.2073μs 32.0438 KOps/s 32.7223 KOps/s $\color{#d91a1a}-2.07\%$
test_step_mdp_speed[True-False-True-False-True] 65.4110μs 28.3330μs 35.2945 KOps/s 36.2221 KOps/s $\color{#d91a1a}-2.56\%$
test_step_mdp_speed[True-False-True-False-False] 45.7310μs 16.7985μs 59.5292 KOps/s 60.5662 KOps/s $\color{#d91a1a}-1.71\%$
test_step_mdp_speed[True-False-False-True-True] 87.3210μs 54.4751μs 18.3570 KOps/s 18.9159 KOps/s $\color{#d91a1a}-2.95\%$
test_step_mdp_speed[True-False-False-True-False] 97.4220μs 33.5571μs 29.7999 KOps/s 30.4078 KOps/s $\color{#d91a1a}-2.00\%$
test_step_mdp_speed[True-False-False-False-True] 63.6110μs 30.9660μs 32.2934 KOps/s 33.0048 KOps/s $\color{#d91a1a}-2.16\%$
test_step_mdp_speed[True-False-False-False-False] 51.5710μs 19.4116μs 51.5157 KOps/s 51.6480 KOps/s $\color{#d91a1a}-0.26\%$
test_step_mdp_speed[False-True-True-True-True] 82.7010μs 51.5439μs 19.4009 KOps/s 20.2534 KOps/s $\color{#d91a1a}-4.21\%$
test_step_mdp_speed[False-True-True-True-False] 62.1410μs 30.9742μs 32.2849 KOps/s 33.1368 KOps/s $\color{#d91a1a}-2.57\%$
test_step_mdp_speed[False-True-True-False-True] 59.8720μs 32.6327μs 30.6441 KOps/s 31.6170 KOps/s $\color{#d91a1a}-3.08\%$
test_step_mdp_speed[False-True-True-False-False] 45.2710μs 18.3579μs 54.4723 KOps/s 55.0278 KOps/s $\color{#d91a1a}-1.01\%$
test_step_mdp_speed[False-True-False-True-True] 2.6295ms 53.8792μs 18.5600 KOps/s 18.9266 KOps/s $\color{#d91a1a}-1.94\%$
test_step_mdp_speed[False-True-False-True-False] 60.6910μs 33.5091μs 29.8426 KOps/s 29.9931 KOps/s $\color{#d91a1a}-0.50\%$
test_step_mdp_speed[False-True-False-False-True] 64.4510μs 34.3080μs 29.1477 KOps/s 29.9417 KOps/s $\color{#d91a1a}-2.65\%$
test_step_mdp_speed[False-True-False-False-False] 50.2110μs 21.0270μs 47.5579 KOps/s 47.5232 KOps/s $\color{#35bf28}+0.07\%$
test_step_mdp_speed[False-False-True-True-True] 91.7220μs 56.7470μs 17.6221 KOps/s 17.7535 KOps/s $\color{#d91a1a}-0.74\%$
test_step_mdp_speed[False-False-True-True-False] 75.5520μs 36.5570μs 27.3545 KOps/s 27.7955 KOps/s $\color{#d91a1a}-1.59\%$
test_step_mdp_speed[False-False-True-False-True] 96.6320μs 34.6361μs 28.8716 KOps/s 29.3402 KOps/s $\color{#d91a1a}-1.60\%$
test_step_mdp_speed[False-False-True-False-False] 52.3610μs 21.0140μs 47.5873 KOps/s 48.3116 KOps/s $\color{#d91a1a}-1.50\%$
test_step_mdp_speed[False-False-False-True-True] 87.4320μs 58.2485μs 17.1678 KOps/s 17.1528 KOps/s $\color{#35bf28}+0.09\%$
test_step_mdp_speed[False-False-False-True-False] 72.8510μs 38.8865μs 25.7158 KOps/s 26.1324 KOps/s $\color{#d91a1a}-1.59\%$
test_step_mdp_speed[False-False-False-False-True] 66.3010μs 36.7290μs 27.2264 KOps/s 28.1411 KOps/s $\color{#d91a1a}-3.25\%$
test_step_mdp_speed[False-False-False-False-False] 76.7010μs 23.5199μs 42.5173 KOps/s 42.6216 KOps/s $\color{#d91a1a}-0.24\%$
test_non_tensor_env_rollout_speed[1000-single-True] 0.8606s 0.7863s 1.2718 Ops/s 1.3069 Ops/s $\color{#d91a1a}-2.68\%$
test_non_tensor_env_rollout_speed[1000-single-False] 0.7292s 0.6303s 1.5865 Ops/s 1.5836 Ops/s $\color{#35bf28}+0.18\%$
test_non_tensor_env_rollout_speed[1000-serial-no-buffers-True] 1.7473s 1.6736s 0.5975 Ops/s 0.6006 Ops/s $\color{#d91a1a}-0.52\%$
test_non_tensor_env_rollout_speed[1000-serial-no-buffers-False] 1.5186s 1.4427s 0.6932 Ops/s 0.6907 Ops/s $\color{#35bf28}+0.35\%$
test_non_tensor_env_rollout_speed[1000-serial-buffers-True] 1.9871s 1.9127s 0.5228 Ops/s 0.5240 Ops/s $\color{#d91a1a}-0.23\%$
test_non_tensor_env_rollout_speed[1000-serial-buffers-False] 1.7668s 1.6925s 0.5909 Ops/s 0.5837 Ops/s $\color{#35bf28}+1.22\%$
test_non_tensor_env_rollout_speed[1000-parallel-no-buffers-True] 4.7728s 4.6703s 0.2141 Ops/s 0.2166 Ops/s $\color{#d91a1a}-1.13\%$
test_non_tensor_env_rollout_speed[1000-parallel-no-buffers-False] 4.5349s 4.4289s 0.2258 Ops/s 0.2299 Ops/s $\color{#d91a1a}-1.78\%$
test_non_tensor_env_rollout_speed[1000-parallel-buffers-True] 2.1040s 1.9719s 0.5071 Ops/s 0.4993 Ops/s $\color{#35bf28}+1.57\%$
test_non_tensor_env_rollout_speed[1000-parallel-buffers-False] 1.7505s 1.6701s 0.5988 Ops/s 0.6051 Ops/s $\color{#d91a1a}-1.05\%$
test_values[generalized_advantage_estimate-True-True] 10.5713ms 10.0502ms 99.5000 Ops/s 100.0113 Ops/s $\color{#d91a1a}-0.51\%$
test_values[vec_generalized_advantage_estimate-True-True] 21.4201ms 18.1454ms 55.1103 Ops/s 55.8371 Ops/s $\color{#d91a1a}-1.30\%$
test_values[td0_return_estimate-False-False] 0.1898ms 0.1254ms 7.9766 KOps/s 8.1666 KOps/s $\color{#d91a1a}-2.33\%$
test_values[td1_return_estimate-False-False] 26.9582ms 26.6463ms 37.5287 Ops/s 36.8714 Ops/s $\color{#35bf28}+1.78\%$
test_values[vec_td1_return_estimate-False-False] 23.0219ms 18.1571ms 55.0750 Ops/s 55.8111 Ops/s $\color{#d91a1a}-1.32\%$
test_values[td_lambda_return_estimate-True-False] 43.5461ms 39.6500ms 25.2207 Ops/s 24.9586 Ops/s $\color{#35bf28}+1.05\%$
test_values[vec_td_lambda_return_estimate-True-False] 21.8538ms 18.1472ms 55.1050 Ops/s 55.7294 Ops/s $\color{#d91a1a}-1.12\%$
test_gae_speed[generalized_advantage_estimate-False-1-512] 9.1333ms 8.8492ms 113.0044 Ops/s 112.6471 Ops/s $\color{#35bf28}+0.32\%$
test_gae_speed[vec_generalized_advantage_estimate-True-1-512] 1.6362ms 1.4711ms 679.7726 Ops/s 654.1901 Ops/s $\color{#35bf28}+3.91\%$
test_gae_speed[vec_generalized_advantage_estimate-False-1-512] 0.4611ms 0.4030ms 2.4817 KOps/s 2.4053 KOps/s $\color{#35bf28}+3.17\%$
test_gae_speed[vec_generalized_advantage_estimate-True-32-512] 34.9690ms 34.0484ms 29.3700 Ops/s 32.1969 Ops/s $\textbf{\color{#d91a1a}-8.78\%}$
test_gae_speed[vec_generalized_advantage_estimate-False-32-512] 2.1503ms 1.7442ms 573.3184 Ops/s 576.5063 Ops/s $\color{#d91a1a}-0.55\%$
test_dqn_speed[False-None] 1.8964ms 1.3896ms 719.6431 Ops/s 713.4368 Ops/s $\color{#35bf28}+0.87\%$
test_dqn_speed[False-backward] 1.9652ms 1.8860ms 530.2094 Ops/s 523.5156 Ops/s $\color{#35bf28}+1.28\%$
test_dqn_speed[True-None] 0.9220ms 0.5311ms 1.8827 KOps/s 1.8887 KOps/s $\color{#d91a1a}-0.31\%$
test_dqn_speed[True-backward] 1.0074ms 0.9738ms 1.0269 KOps/s 841.9399 Ops/s $\textbf{\color{#35bf28}+21.96\%}$
test_dqn_speed[reduce-overhead-None] 0.6738ms 0.5183ms 1.9293 KOps/s 1.8615 KOps/s $\color{#35bf28}+3.64\%$
test_ddpg_speed[False-None] 3.0225ms 2.8147ms 355.2762 Ops/s 350.8765 Ops/s $\color{#35bf28}+1.25\%$
test_ddpg_speed[False-backward] 4.1221ms 4.0273ms 248.3024 Ops/s 242.3724 Ops/s $\color{#35bf28}+2.45\%$
test_ddpg_speed[True-None] 1.8150ms 1.3779ms 725.7664 Ops/s 715.2393 Ops/s $\color{#35bf28}+1.47\%$
test_ddpg_speed[True-backward] 2.3614ms 2.3267ms 429.7928 Ops/s 380.9187 Ops/s $\textbf{\color{#35bf28}+12.83\%}$
test_ddpg_speed[reduce-overhead-None] 1.7745ms 1.3596ms 735.4870 Ops/s 719.8732 Ops/s $\color{#35bf28}+2.17\%$
test_sac_speed[False-None] 8.5541ms 7.9333ms 126.0514 Ops/s 126.2819 Ops/s $\color{#d91a1a}-0.18\%$
test_sac_speed[False-backward] 11.6154ms 11.0994ms 90.0950 Ops/s 85.4982 Ops/s $\textbf{\color{#35bf28}+5.38\%}$
test_sac_speed[True-None] 2.3195ms 2.1303ms 469.4253 Ops/s 453.2902 Ops/s $\color{#35bf28}+3.56\%$
test_sac_speed[True-backward] 4.1619ms 4.0665ms 245.9116 Ops/s 237.3993 Ops/s $\color{#35bf28}+3.59\%$
test_sac_speed[reduce-overhead-None] 2.4527ms 2.1340ms 468.6043 Ops/s 451.8093 Ops/s $\color{#35bf28}+3.72\%$
test_redq_speed[False-None] 15.2262ms 10.6008ms 94.3322 Ops/s 96.3717 Ops/s $\color{#d91a1a}-2.12\%$
test_redq_speed[False-backward] 22.2968ms 18.1311ms 55.1540 Ops/s 56.1372 Ops/s $\color{#d91a1a}-1.75\%$
test_redq_speed[True-None] 4.7168ms 4.4877ms 222.8315 Ops/s 233.1582 Ops/s $\color{#d91a1a}-4.43\%$
test_redq_speed[True-backward] 10.1935ms 9.7723ms 102.3304 Ops/s 101.4160 Ops/s $\color{#35bf28}+0.90\%$
test_redq_speed[reduce-overhead-None] 4.9771ms 4.4773ms 223.3471 Ops/s 230.0482 Ops/s $\color{#d91a1a}-2.91\%$
test_redq_deprec_speed[False-None] 11.5538ms 10.9096ms 91.6627 Ops/s 90.0617 Ops/s $\color{#35bf28}+1.78\%$
test_redq_deprec_speed[False-backward] 15.9931ms 15.5907ms 64.1406 Ops/s 63.2214 Ops/s $\color{#35bf28}+1.45\%$
test_redq_deprec_speed[True-None] 4.1910ms 3.6744ms 272.1521 Ops/s 260.9864 Ops/s $\color{#35bf28}+4.28\%$
test_redq_deprec_speed[True-backward] 7.9680ms 7.6053ms 131.4874 Ops/s 129.1778 Ops/s $\color{#35bf28}+1.79\%$
test_redq_deprec_speed[reduce-overhead-None] 4.0928ms 3.6291ms 275.5489 Ops/s 276.2684 Ops/s $\color{#d91a1a}-0.26\%$
test_td3_speed[False-None] 8.1090ms 7.9536ms 125.7290 Ops/s 126.1278 Ops/s $\color{#d91a1a}-0.32\%$
test_td3_speed[False-backward] 11.3879ms 10.9150ms 91.6167 Ops/s 93.1242 Ops/s $\color{#d91a1a}-1.62\%$
test_td3_speed[True-None] 1.9031ms 1.8311ms 546.1222 Ops/s 539.9373 Ops/s $\color{#35bf28}+1.15\%$
test_td3_speed[True-backward] 3.8133ms 3.6589ms 273.3080 Ops/s 269.3162 Ops/s $\color{#35bf28}+1.48\%$
test_td3_speed[reduce-overhead-None] 1.8359ms 1.7930ms 557.7202 Ops/s 546.4028 Ops/s $\color{#35bf28}+2.07\%$
test_cql_speed[False-None] 26.6604ms 25.8673ms 38.6588 Ops/s 38.4550 Ops/s $\color{#35bf28}+0.53\%$
test_cql_speed[False-backward] 39.2131ms 35.7627ms 27.9621 Ops/s 28.0138 Ops/s $\color{#d91a1a}-0.18\%$
test_cql_speed[True-None] 13.2284ms 12.4663ms 80.2160 Ops/s 79.7510 Ops/s $\color{#35bf28}+0.58\%$
test_cql_speed[True-backward] 18.6615ms 18.3044ms 54.6317 Ops/s 54.3004 Ops/s $\color{#35bf28}+0.61\%$
test_cql_speed[reduce-overhead-None] 12.7464ms 12.4291ms 80.4566 Ops/s 76.8999 Ops/s $\color{#35bf28}+4.63\%$
test_a2c_speed[False-None] 5.6759ms 5.4015ms 185.1321 Ops/s 185.3374 Ops/s $\color{#d91a1a}-0.11\%$
test_a2c_speed[False-backward] 12.3833ms 11.8588ms 84.3254 Ops/s 84.1757 Ops/s $\color{#35bf28}+0.18\%$
test_a2c_speed[True-None] 4.0201ms 3.7522ms 266.5098 Ops/s 264.9207 Ops/s $\color{#35bf28}+0.60\%$
test_a2c_speed[True-backward] 8.7274ms 8.4849ms 117.8562 Ops/s 116.2058 Ops/s $\color{#35bf28}+1.42\%$
test_a2c_speed[reduce-overhead-None] 3.9361ms 3.7048ms 269.9207 Ops/s 265.9719 Ops/s $\color{#35bf28}+1.48\%$
test_ppo_speed[False-None] 6.1498ms 5.9393ms 168.3706 Ops/s 170.4368 Ops/s $\color{#d91a1a}-1.21\%$
test_ppo_speed[False-backward] 12.7649ms 12.3999ms 80.6456 Ops/s 79.4454 Ops/s $\color{#35bf28}+1.51\%$
test_ppo_speed[True-None] 3.8176ms 3.6139ms 276.7097 Ops/s 272.1810 Ops/s $\color{#35bf28}+1.66\%$
test_ppo_speed[True-backward] 8.5377ms 8.2363ms 121.4134 Ops/s 107.2899 Ops/s $\textbf{\color{#35bf28}+13.16\%}$
test_ppo_speed[reduce-overhead-None] 3.8115ms 3.5805ms 279.2928 Ops/s 272.9519 Ops/s $\color{#35bf28}+2.32\%$
test_reinforce_speed[False-None] 4.8236ms 4.5603ms 219.2816 Ops/s 216.1279 Ops/s $\color{#35bf28}+1.46\%$
test_reinforce_speed[False-backward] 7.6377ms 7.3182ms 136.6462 Ops/s 135.2530 Ops/s $\color{#35bf28}+1.03\%$
test_reinforce_speed[True-None] 3.0137ms 2.8357ms 352.6476 Ops/s 337.7983 Ops/s $\color{#35bf28}+4.40\%$
test_reinforce_speed[True-backward] 8.1858ms 7.7625ms 128.8242 Ops/s 128.9854 Ops/s $\color{#d91a1a}-0.12\%$
test_reinforce_speed[reduce-overhead-None] 3.3963ms 2.8325ms 353.0468 Ops/s 328.9056 Ops/s $\textbf{\color{#35bf28}+7.34\%}$
test_iql_speed[False-None] 25.1874ms 20.0896ms 49.7769 Ops/s 51.1400 Ops/s $\color{#d91a1a}-2.67\%$
test_iql_speed[False-backward] 31.3604ms 30.2000ms 33.1126 Ops/s 33.6314 Ops/s $\color{#d91a1a}-1.54\%$
test_iql_speed[True-None] 8.8537ms 8.4938ms 117.7336 Ops/s 116.9393 Ops/s $\color{#35bf28}+0.68\%$
test_iql_speed[True-backward] 17.0237ms 16.7065ms 59.8569 Ops/s 58.6250 Ops/s $\color{#35bf28}+2.10\%$
test_iql_speed[reduce-overhead-None] 9.2607ms 8.5739ms 116.6331 Ops/s 116.0056 Ops/s $\color{#35bf28}+0.54\%$
test_rb_sample[TensorDictReplayBuffer-ListStorage-RandomSampler-4000] 7.7221ms 6.0777ms 164.5361 Ops/s 164.5553 Ops/s $\color{#d91a1a}-0.01\%$
test_rb_sample[TensorDictReplayBuffer-LazyMemmapStorage-RandomSampler-10000] 0.5626ms 0.3029ms 3.3012 KOps/s 3.4919 KOps/s $\textbf{\color{#d91a1a}-5.46\%}$
test_rb_sample[TensorDictReplayBuffer-LazyTensorStorage-RandomSampler-10000] 0.5293ms 0.2930ms 3.4128 KOps/s 3.7446 KOps/s $\textbf{\color{#d91a1a}-8.86\%}$
test_rb_sample[TensorDictReplayBuffer-ListStorage-SamplerWithoutReplacement-4000] 6.0076ms 5.7727ms 173.2292 Ops/s 172.0634 Ops/s $\color{#35bf28}+0.68\%$
test_rb_sample[TensorDictReplayBuffer-LazyMemmapStorage-SamplerWithoutReplacement-10000] 1.7621ms 0.2816ms 3.5512 KOps/s 2.9193 KOps/s $\textbf{\color{#35bf28}+21.65\%}$
test_rb_sample[TensorDictReplayBuffer-LazyTensorStorage-SamplerWithoutReplacement-10000] 0.4960ms 0.3229ms 3.0972 KOps/s 3.1591 KOps/s $\color{#d91a1a}-1.96\%$
test_rb_sample[TensorDictReplayBuffer-LazyMemmapStorage-sampler6-10000] 1.6494ms 1.3714ms 729.1904 Ops/s 699.7866 Ops/s $\color{#35bf28}+4.20\%$
test_rb_sample[TensorDictReplayBuffer-LazyTensorStorage-sampler7-10000] 1.5466ms 1.2942ms 772.6653 Ops/s 741.0066 Ops/s $\color{#35bf28}+4.27\%$
test_rb_sample[TensorDictPrioritizedReplayBuffer-ListStorage-None-4000] 10.1449ms 6.0886ms 164.2415 Ops/s 166.7862 Ops/s $\color{#d91a1a}-1.53\%$
test_rb_sample[TensorDictPrioritizedReplayBuffer-LazyMemmapStorage-None-10000] 1.0417ms 0.5041ms 1.9839 KOps/s 1.8809 KOps/s $\textbf{\color{#35bf28}+5.48\%}$
test_rb_sample[TensorDictPrioritizedReplayBuffer-LazyTensorStorage-None-10000] 0.6853ms 0.4867ms 2.0545 KOps/s 1.9322 KOps/s $\textbf{\color{#35bf28}+6.33\%}$
test_rb_iterate[TensorDictReplayBuffer-ListStorage-RandomSampler-4000] 5.9758ms 5.8061ms 172.2316 Ops/s 170.7930 Ops/s $\color{#35bf28}+0.84\%$
test_rb_iterate[TensorDictReplayBuffer-LazyMemmapStorage-RandomSampler-10000] 1.8976ms 0.3600ms 2.7776 KOps/s 2.8849 KOps/s $\color{#d91a1a}-3.72\%$
test_rb_iterate[TensorDictReplayBuffer-LazyTensorStorage-RandomSampler-10000] 0.5111ms 0.3124ms 3.2011 KOps/s 2.8783 KOps/s $\textbf{\color{#35bf28}+11.21\%}$
test_rb_iterate[TensorDictReplayBuffer-ListStorage-SamplerWithoutReplacement-4000] 6.0047ms 5.7290ms 174.5503 Ops/s 171.8112 Ops/s $\color{#35bf28}+1.59\%$
test_rb_iterate[TensorDictReplayBuffer-LazyMemmapStorage-SamplerWithoutReplacement-10000] 1.5940ms 0.2868ms 3.4867 KOps/s 3.2093 KOps/s $\textbf{\color{#35bf28}+8.64\%}$
test_rb_iterate[TensorDictReplayBuffer-LazyTensorStorage-SamplerWithoutReplacement-10000] 0.4682ms 0.2605ms 3.8393 KOps/s 3.7931 KOps/s $\color{#35bf28}+1.22\%$
test_rb_iterate[TensorDictPrioritizedReplayBuffer-ListStorage-None-4000] 6.4333ms 5.8875ms 169.8504 Ops/s 166.7241 Ops/s $\color{#35bf28}+1.88\%$
test_rb_iterate[TensorDictPrioritizedReplayBuffer-LazyMemmapStorage-None-10000] 1.1016ms 0.4891ms 2.0447 KOps/s 2.2985 KOps/s $\textbf{\color{#d91a1a}-11.04\%}$
test_rb_iterate[TensorDictPrioritizedReplayBuffer-LazyTensorStorage-None-10000] 0.6846ms 0.4621ms 2.1639 KOps/s 2.4400 KOps/s $\textbf{\color{#d91a1a}-11.31\%}$
test_rb_populate[TensorDictReplayBuffer-ListStorage-RandomSampler-400] 0.5817s 16.5442ms 60.4442 Ops/s 198.9706 Ops/s $\textbf{\color{#d91a1a}-69.62\%}$
test_rb_populate[TensorDictReplayBuffer-LazyMemmapStorage-RandomSampler-400] 11.2131ms 1.9867ms 503.3471 Ops/s 528.7773 Ops/s $\color{#d91a1a}-4.81\%$
test_rb_populate[TensorDictReplayBuffer-LazyTensorStorage-RandomSampler-400] 1.1316ms 0.8776ms 1.1395 KOps/s 801.6284 Ops/s $\textbf{\color{#35bf28}+42.14\%}$
test_rb_populate[TensorDictReplayBuffer-ListStorage-SamplerWithoutReplacement-400] 6.6430ms 4.9508ms 201.9858 Ops/s 199.6685 Ops/s $\color{#35bf28}+1.16\%$
test_rb_populate[TensorDictReplayBuffer-LazyMemmapStorage-SamplerWithoutReplacement-400] 12.8741ms 1.9742ms 506.5275 Ops/s 564.4229 Ops/s $\textbf{\color{#d91a1a}-10.26\%}$
test_rb_populate[TensorDictReplayBuffer-LazyTensorStorage-SamplerWithoutReplacement-400] 3.3411ms 1.1368ms 879.6549 Ops/s 1.1147 KOps/s $\textbf{\color{#d91a1a}-21.09\%}$
test_rb_populate[TensorDictPrioritizedReplayBuffer-ListStorage-None-400] 0.5342s 15.8242ms 63.1944 Ops/s 58.1275 Ops/s $\textbf{\color{#35bf28}+8.72\%}$
test_rb_populate[TensorDictPrioritizedReplayBuffer-LazyMemmapStorage-None-400] 7.7225ms 1.9910ms 502.2675 Ops/s 476.4822 Ops/s $\textbf{\color{#35bf28}+5.41\%}$
test_rb_populate[TensorDictPrioritizedReplayBuffer-LazyTensorStorage-None-400] 11.5173ms 1.4142ms 707.1329 Ops/s 911.9366 Ops/s $\textbf{\color{#d91a1a}-22.46\%}$
test_rb_extend_sample[ReplayBuffer-LazyTensorStorage-RandomSampler-10000-10000-100-True] 36.4853ms 34.0743ms 29.3476 Ops/s 28.2238 Ops/s $\color{#35bf28}+3.98\%$
test_rb_extend_sample[ReplayBuffer-LazyTensorStorage-RandomSampler-10000-10000-100-False] 19.2975ms 17.6242ms 56.7401 Ops/s 56.1080 Ops/s $\color{#35bf28}+1.13\%$
test_rb_extend_sample[ReplayBuffer-LazyTensorStorage-RandomSampler-100000-10000-100-True] 38.5354ms 35.5258ms 28.1486 Ops/s 28.1298 Ops/s $\color{#35bf28}+0.07\%$
test_rb_extend_sample[ReplayBuffer-LazyTensorStorage-RandomSampler-100000-10000-100-False] 20.1737ms 18.3446ms 54.5120 Ops/s 56.0796 Ops/s $\color{#d91a1a}-2.80\%$
test_rb_extend_sample[ReplayBuffer-LazyTensorStorage-RandomSampler-1000000-10000-100-True] 38.5097ms 37.0044ms 27.0238 Ops/s 26.6795 Ops/s $\color{#35bf28}+1.29\%$
test_rb_extend_sample[ReplayBuffer-LazyTensorStorage-RandomSampler-1000000-10000-100-False] 20.9225ms 19.4764ms 51.3443 Ops/s 51.5441 Ops/s $\color{#d91a1a}-0.39\%$

vmoens added 7 commits January 19, 2026 18:06
@vmoens
[BugFix] Temporarily disable Habitat CI tests
72fa818 
Habitat tests are disabled until habitat-sim officially supports
Python 3.10+ via conda. Current issues:
- tensordict requires Python >= 3.10
- habitat-sim conda packages only support Python < 3.10
- Building from source has CMake compatibility issues
- Dataset downloads via git-lfs prune take hours

Track progress at: facebookresearch/habitat-sim#2289
@vmoens
[BugFix] Add --no-prune flag to speed up Habitat dataset download
db63e60 
The git-lfs prune operation was taking an extremely long time (hours).
Using --no-prune skips this unnecessary step during CI.
@vmoens
[BugFix] Re-enable Habitat CI with --no-prune fix
28c0e83 
Re-enabling the Habitat tests now that the --no-prune flag is added
to the dataset download, which should make downloads complete in
minutes instead of hours.
@vmoens
Fix slow git-lfs downloads by using smudge-disabled clone
da7bfcf 
- Configure git-lfs with higher timeouts (600s activity, 60s dial/tls)
- Enable 8 concurrent transfers for faster parallel downloads
- Use HTTP/1.1 for better compatibility with Hugging Face servers
- Replace datasets_download utility with direct git clone + git lfs pull
- Use smudge-disabled clone to avoid slow sequential LFS fetches
- Download only required datasets: replica_cad, ycb objects, hab_fetch robot

This should reduce dataset download time from 1+ hour to under 1 minute.
@vmoens
Add verbose progress output to dataset downloads for debugging
f99836d
@vmoens
Fix: remove unsupported --progress flag from git lfs pull
7479eea
@vmoens
Use datasets_download utility for rearrange_task_assets
f29da05 
The manual dataset download was missing episode data files
(rearrange_easy.json.gz). Using the datasets_download utility with
rearrange_task_assets UID ensures all required files are downloaded
including scene datasets, objects, robots, and episode data.

The git-lfs config improvements (higher timeouts, 8 concurrent
transfers, HTTP/1.1) should help with download performance.
@vmoens vmoens merged commit ae50733 into main Jan 20, 2026
108 of 113 checks passed
@vmoens vmoens deleted the fix-habitat-claude2 branch January 20, 2026 08:58
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

No reviews

Assignees

No one assigned

Labels

bug Something isn't working CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

1 participant

@vmoens