Skip to content

[CI] Add torch_geometric integration tests#3552

Merged
vmoens merged 2 commits intomainfrom
ci/torch-geometric-tests
Apr 12, 2026
Merged

[CI] Add torch_geometric integration tests#3552
vmoens merged 2 commits intomainfrom
ci/torch-geometric-tests

Conversation

@vmoens
Copy link
Copy Markdown
Collaborator

@vmoens vmoens commented Mar 14, 2026

Summary

  • Adds a TestTorchGeometric test class to test/test_libs.py with 5 tests covering deepcopy compatibility, collector integration, and TensorDictModule wrapping with torch_geometric layers (addresses [BUG] torch geometric layers not working in policy network #2679)
  • Adds CI job unittests-torch_geometric to test-linux-libs.yml with dedicated scripts, triggered by Integrations / Integrations/torch_geometric labels
  • Adds auto-labeling rules for the new Integrations/torch_geometric label in .github/labeler.yml

Note: the original deepcopy bug from #2679 was fixed upstream in torch_geometric 2.7.0 (they removed the custom __deepcopy__ from Linear). These tests serve as a regression guard to catch any future incompatibility on either side.

Test plan

  • TestTorchGeometric tests pass locally (4 passed, 1 skipped for no CUDA)
  • CI job runs successfully on GPU runner (exercises the CUDA cross-device collector test)

Made with Cursor

Add CI workflow and tests to verify torchrl compatibility with
torch_geometric (addresses #2679).

Tests cover:
- deepcopy of modules containing torch_geometric layers
- The collector's meta-device deepcopy pattern
- Collector integration with torch_geometric-based policies
- TensorDictModule wrapping of torch_geometric modules

Made-with: Cursor
@vmoens vmoens added the CI Has to do with CI setup (e.g. wheels & builds, tests...) label Mar 14, 2026
@pytorch-bot
Copy link
Copy Markdown

pytorch-bot bot commented Mar 14, 2026

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/rl/3552

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

❗ 1 Active SEVs

There are 1 currently active SEVs. If your PR is affected, please view them below:

❌ 1 New Failure, 11 Unrelated Failures

As of commit 6a20940 with merge base a4301ee (image):

NEW FAILURE - The following job has failed:

BROKEN TRUNK - The following jobs failed but were present on the merge base:

👉 Rebase onto the `viable/strict` branch to avoid these failures

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 Mar 14, 2026
@github-actions github-actions bot added the Environments Adds or modifies an environment wrapper label Mar 14, 2026
@github-actions
Copy link
Copy Markdown
Contributor

github-actions bot commented Mar 14, 2026

$\color{#D29922}\textsf{\Large⚠\kern{0.2cm}\normalsize Warning}$ Result of GPU Benchmark Tests

Total Benchmarks: 172. Improved: $\large\color{#35bf28}13$. Worsened: $\large\color{#d91a1a}5$.

Expand to view detailed results
Name Max Mean Ops Ops on Repo HEAD Change
test_tensor_to_bytestream_speed[pickle] 81.1120μs 80.2458μs 12.4617 KOps/s 12.6027 KOps/s $\color{#d91a1a}-1.12\%$
test_tensor_to_bytestream_speed[torch.save] 0.1431ms 0.1389ms 7.1995 KOps/s 7.2775 KOps/s $\color{#d91a1a}-1.07\%$
test_tensor_to_bytestream_speed[untyped_storage] 0.1039s 0.1036s 9.6493 Ops/s 9.7038 Ops/s $\color{#d91a1a}-0.56\%$
test_tensor_to_bytestream_speed[numpy] 2.4081μs 2.3996μs 416.7350 KOps/s 415.0828 KOps/s $\color{#35bf28}+0.40\%$
test_tensor_to_bytestream_speed[safetensors] 38.0537μs 37.4392μs 26.7100 KOps/s 26.6197 KOps/s $\color{#35bf28}+0.34\%$
test_simple 0.7803s 0.7799s 1.2823 Ops/s 1.2421 Ops/s $\color{#35bf28}+3.23\%$
test_transformed 1.3715s 1.3674s 0.7313 Ops/s 0.7199 Ops/s $\color{#35bf28}+1.59\%$
test_serial 2.2926s 2.2905s 0.4366 Ops/s 0.4307 Ops/s $\color{#35bf28}+1.36\%$
test_parallel 1.9064s 1.8054s 0.5539 Ops/s 0.5480 Ops/s $\color{#35bf28}+1.09\%$
test_step_mdp_speed[True-True-True-True-True] 0.2291ms 41.4278μs 24.1384 KOps/s 24.2642 KOps/s $\color{#d91a1a}-0.52\%$
test_step_mdp_speed[True-True-True-True-False] 51.7100μs 22.3921μs 44.6585 KOps/s 44.9218 KOps/s $\color{#d91a1a}-0.59\%$
test_step_mdp_speed[True-True-True-False-True] 75.3210μs 23.8289μs 41.9658 KOps/s 44.0052 KOps/s $\color{#d91a1a}-4.63\%$
test_step_mdp_speed[True-True-True-False-False] 47.7600μs 12.3698μs 80.8418 KOps/s 80.3550 KOps/s $\color{#35bf28}+0.61\%$
test_step_mdp_speed[True-True-False-True-True] 75.1410μs 44.8043μs 22.3193 KOps/s 23.0520 KOps/s $\color{#d91a1a}-3.18\%$
test_step_mdp_speed[True-True-False-True-False] 48.8000μs 24.9041μs 40.1540 KOps/s 40.6461 KOps/s $\color{#d91a1a}-1.21\%$
test_step_mdp_speed[True-True-False-False-True] 61.1310μs 25.6266μs 39.0219 KOps/s 39.0336 KOps/s $\color{#d91a1a}-0.03\%$
test_step_mdp_speed[True-True-False-False-False] 40.7310μs 14.9497μs 66.8909 KOps/s 67.2187 KOps/s $\color{#d91a1a}-0.49\%$
test_step_mdp_speed[True-False-True-True-True] 0.1382ms 45.5689μs 21.9448 KOps/s 22.1611 KOps/s $\color{#d91a1a}-0.98\%$
test_step_mdp_speed[True-False-True-True-False] 61.7310μs 27.1307μs 36.8586 KOps/s 36.4972 KOps/s $\color{#35bf28}+0.99\%$
test_step_mdp_speed[True-False-True-False-True] 55.9410μs 25.4526μs 39.2887 KOps/s 39.0849 KOps/s $\color{#35bf28}+0.52\%$
test_step_mdp_speed[True-False-True-False-False] 43.4610μs 14.9183μs 67.0318 KOps/s 66.3202 KOps/s $\color{#35bf28}+1.07\%$
test_step_mdp_speed[True-False-False-True-True] 80.5410μs 48.0585μs 20.8080 KOps/s 20.2117 KOps/s $\color{#35bf28}+2.95\%$
test_step_mdp_speed[True-False-False-True-False] 57.7210μs 29.4826μs 33.9183 KOps/s 33.8173 KOps/s $\color{#35bf28}+0.30\%$
test_step_mdp_speed[True-False-False-False-True] 75.3210μs 27.5704μs 36.2708 KOps/s 35.3311 KOps/s $\color{#35bf28}+2.66\%$
test_step_mdp_speed[True-False-False-False-False] 44.6010μs 17.1662μs 58.2540 KOps/s 56.6667 KOps/s $\color{#35bf28}+2.80\%$
test_step_mdp_speed[False-True-True-True-True] 78.9910μs 45.6474μs 21.9070 KOps/s 21.5970 KOps/s $\color{#35bf28}+1.44\%$
test_step_mdp_speed[False-True-True-True-False] 63.1200μs 27.0719μs 36.9387 KOps/s 36.8039 KOps/s $\color{#35bf28}+0.37\%$
test_step_mdp_speed[False-True-True-False-True] 2.4257ms 29.5461μs 33.8454 KOps/s 34.3909 KOps/s $\color{#d91a1a}-1.59\%$
test_step_mdp_speed[False-True-True-False-False] 64.0710μs 16.8981μs 59.1784 KOps/s 60.6920 KOps/s $\color{#d91a1a}-2.49\%$
test_step_mdp_speed[False-True-False-True-True] 0.1127ms 48.2971μs 20.7052 KOps/s 21.0648 KOps/s $\color{#d91a1a}-1.71\%$
test_step_mdp_speed[False-True-False-True-False] 57.6610μs 29.1396μs 34.3176 KOps/s 33.4877 KOps/s $\color{#35bf28}+2.48\%$
test_step_mdp_speed[False-True-False-False-True] 73.0410μs 30.8423μs 32.4230 KOps/s 31.7986 KOps/s $\color{#35bf28}+1.96\%$
test_step_mdp_speed[False-True-False-False-False] 49.4310μs 18.7287μs 53.3941 KOps/s 52.3012 KOps/s $\color{#35bf28}+2.09\%$
test_step_mdp_speed[False-False-True-True-True] 86.8010μs 50.7167μs 19.7174 KOps/s 19.4899 KOps/s $\color{#35bf28}+1.17\%$
test_step_mdp_speed[False-False-True-True-False] 63.6810μs 32.0465μs 31.2047 KOps/s 31.0381 KOps/s $\color{#35bf28}+0.54\%$
test_step_mdp_speed[False-False-True-False-True] 63.8510μs 31.3459μs 31.9021 KOps/s 31.5037 KOps/s $\color{#35bf28}+1.26\%$
test_step_mdp_speed[False-False-True-False-False] 49.7410μs 18.3729μs 54.4281 KOps/s 53.2621 KOps/s $\color{#35bf28}+2.19\%$
test_step_mdp_speed[False-False-False-True-True] 84.7920μs 51.5131μs 19.4125 KOps/s 19.0368 KOps/s $\color{#35bf28}+1.97\%$
test_step_mdp_speed[False-False-False-True-False] 69.1010μs 34.4486μs 29.0288 KOps/s 28.7211 KOps/s $\color{#35bf28}+1.07\%$
test_step_mdp_speed[False-False-False-False-True] 99.7120μs 33.5164μs 29.8361 KOps/s 29.5790 KOps/s $\color{#35bf28}+0.87\%$
test_step_mdp_speed[False-False-False-False-False] 55.7610μs 21.7351μs 46.0085 KOps/s 46.6732 KOps/s $\color{#d91a1a}-1.42\%$
test_non_tensor_env_rollout_speed[1000-single-True] 0.7133s 0.7039s 1.4207 Ops/s 1.3670 Ops/s $\color{#35bf28}+3.93\%$
test_non_tensor_env_rollout_speed[1000-single-False] 0.6955s 0.5962s 1.6772 Ops/s 1.6794 Ops/s $\color{#d91a1a}-0.13\%$
test_non_tensor_env_rollout_speed[1000-serial-no-buffers-True] 1.6961s 1.6091s 0.6215 Ops/s 0.6196 Ops/s $\color{#35bf28}+0.30\%$
test_non_tensor_env_rollout_speed[1000-serial-no-buffers-False] 1.4958s 1.4052s 0.7116 Ops/s 0.7142 Ops/s $\color{#d91a1a}-0.36\%$
test_non_tensor_env_rollout_speed[1000-serial-buffers-True] 1.9511s 1.8659s 0.5359 Ops/s 0.5382 Ops/s $\color{#d91a1a}-0.41\%$
test_non_tensor_env_rollout_speed[1000-serial-buffers-False] 1.7349s 1.6482s 0.6067 Ops/s 0.6093 Ops/s $\color{#d91a1a}-0.43\%$
test_non_tensor_env_rollout_speed[1000-parallel-no-buffers-True] 4.5882s 4.5195s 0.2213 Ops/s 0.2197 Ops/s $\color{#35bf28}+0.71\%$
test_non_tensor_env_rollout_speed[1000-parallel-no-buffers-False] 4.5463s 4.3915s 0.2277 Ops/s 0.2266 Ops/s $\color{#35bf28}+0.51\%$
test_non_tensor_env_rollout_speed[1000-parallel-buffers-True] 1.9396s 1.8738s 0.5337 Ops/s 0.5452 Ops/s $\color{#d91a1a}-2.11\%$
test_non_tensor_env_rollout_speed[1000-parallel-buffers-False] 1.6408s 1.5584s 0.6417 Ops/s 0.6362 Ops/s $\color{#35bf28}+0.86\%$
test_values[generalized_advantage_estimate-True-True] 21.4964ms 20.8919ms 47.8654 Ops/s 49.1656 Ops/s $\color{#d91a1a}-2.64\%$
test_values[vec_generalized_advantage_estimate-True-True] 0.1262s 3.4488ms 289.9521 Ops/s 274.5974 Ops/s $\textbf{\color{#35bf28}+5.59\%}$
test_values[td0_return_estimate-False-False] 0.1062ms 84.1121μs 11.8889 KOps/s 11.8962 KOps/s $\color{#d91a1a}-0.06\%$
test_values[td1_return_estimate-False-False] 50.6444ms 49.5001ms 20.2020 Ops/s 20.6972 Ops/s $\color{#d91a1a}-2.39\%$
test_values[vec_td1_return_estimate-False-False] 1.3471ms 1.1005ms 908.6959 Ops/s 909.7921 Ops/s $\color{#d91a1a}-0.12\%$
test_values[td_lambda_return_estimate-True-False] 82.9348ms 81.8778ms 12.2133 Ops/s 12.5821 Ops/s $\color{#d91a1a}-2.93\%$
test_values[vec_td_lambda_return_estimate-True-False] 1.2893ms 1.0945ms 913.6627 Ops/s 911.0260 Ops/s $\color{#35bf28}+0.29\%$
test_gae_speed[generalized_advantage_estimate-False-1-512] 21.5592ms 21.1034ms 47.3856 Ops/s 48.7485 Ops/s $\color{#d91a1a}-2.80\%$
test_gae_speed[vec_generalized_advantage_estimate-True-1-512] 1.0325ms 0.7781ms 1.2852 KOps/s 1.3101 KOps/s $\color{#d91a1a}-1.90\%$
test_gae_speed[vec_generalized_advantage_estimate-False-1-512] 0.7661ms 0.6981ms 1.4325 KOps/s 1.4626 KOps/s $\color{#d91a1a}-2.06\%$
test_gae_speed[vec_generalized_advantage_estimate-True-32-512] 1.5819ms 1.4934ms 669.6060 Ops/s 669.9431 Ops/s $\color{#d91a1a}-0.05\%$
test_gae_speed[vec_generalized_advantage_estimate-False-32-512] 0.7793ms 0.7203ms 1.3883 KOps/s 1.4272 KOps/s $\color{#d91a1a}-2.73\%$
test_dqn_speed[False-None] 1.6990ms 1.5965ms 626.3640 Ops/s 628.0624 Ops/s $\color{#d91a1a}-0.27\%$
test_dqn_speed[False-backward] 2.4752ms 2.2460ms 445.2377 Ops/s 447.9881 Ops/s $\color{#d91a1a}-0.61\%$
test_dqn_speed[True-None] 0.6835ms 0.6020ms 1.6611 KOps/s 1.5985 KOps/s $\color{#35bf28}+3.92\%$
test_dqn_speed[True-backward] 1.3374ms 1.2778ms 782.5974 Ops/s 785.8600 Ops/s $\color{#d91a1a}-0.42\%$
test_dqn_speed[reduce-overhead-None] 0.7384ms 0.6094ms 1.6410 KOps/s 1.5682 KOps/s $\color{#35bf28}+4.65\%$
test_ddpg_speed[False-None] 3.4057ms 2.9745ms 336.1905 Ops/s 330.2699 Ops/s $\color{#35bf28}+1.79\%$
test_ddpg_speed[False-backward] 4.8325ms 4.4137ms 226.5661 Ops/s 223.5347 Ops/s $\color{#35bf28}+1.36\%$
test_ddpg_speed[True-None] 1.7276ms 1.3931ms 717.8276 Ops/s 718.7848 Ops/s $\color{#d91a1a}-0.13\%$
test_ddpg_speed[True-backward] 2.6595ms 2.5748ms 388.3733 Ops/s 382.2563 Ops/s $\color{#35bf28}+1.60\%$
test_ddpg_speed[reduce-overhead-None] 1.4801ms 1.3864ms 721.2964 Ops/s 721.3542 Ops/s $-0.01\%$
test_sac_speed[False-None] 8.8509ms 8.4634ms 118.1565 Ops/s 117.7497 Ops/s $\color{#35bf28}+0.35\%$
test_sac_speed[False-backward] 12.1975ms 11.7612ms 85.0255 Ops/s 86.7909 Ops/s $\color{#d91a1a}-2.03\%$
test_sac_speed[True-None] 2.1165ms 1.9733ms 506.7608 Ops/s 514.3854 Ops/s $\color{#d91a1a}-1.48\%$
test_sac_speed[True-backward] 4.1176ms 3.7810ms 264.4800 Ops/s 276.2952 Ops/s $\color{#d91a1a}-4.28\%$
test_sac_speed[reduce-overhead-None] 16.3190ms 10.0553ms 99.4503 Ops/s 99.8528 Ops/s $\color{#d91a1a}-0.40\%$
test_redq_deprec_speed[False-None] 10.2484ms 9.5001ms 105.2622 Ops/s 103.5040 Ops/s $\color{#35bf28}+1.70\%$
test_redq_deprec_speed[False-backward] 13.3921ms 12.8783ms 77.6503 Ops/s 79.0355 Ops/s $\color{#d91a1a}-1.75\%$
test_redq_deprec_speed[True-None] 2.8940ms 2.7333ms 365.8631 Ops/s 358.9927 Ops/s $\color{#35bf28}+1.91\%$
test_redq_deprec_speed[True-backward] 4.8947ms 4.4504ms 224.6965 Ops/s 221.5300 Ops/s $\color{#35bf28}+1.43\%$
test_redq_deprec_speed[reduce-overhead-None] 14.3171ms 9.5449ms 104.7675 Ops/s 103.7330 Ops/s $\color{#35bf28}+1.00\%$
test_td3_speed[False-None] 8.5924ms 8.3902ms 119.1867 Ops/s 119.2537 Ops/s $\color{#d91a1a}-0.06\%$
test_td3_speed[False-backward] 11.6851ms 10.9914ms 90.9798 Ops/s 90.6846 Ops/s $\color{#35bf28}+0.33\%$
test_td3_speed[True-None] 1.7338ms 1.7107ms 584.5425 Ops/s 565.7174 Ops/s $\color{#35bf28}+3.33\%$
test_td3_speed[True-backward] 3.7125ms 3.2793ms 304.9466 Ops/s 299.0697 Ops/s $\color{#35bf28}+1.97\%$
test_td3_speed[reduce-overhead-None] 49.1874ms 25.3240ms 39.4883 Ops/s 38.2545 Ops/s $\color{#35bf28}+3.23\%$
test_cql_speed[False-None] 17.8997ms 17.4989ms 57.1466 Ops/s 56.5875 Ops/s $\color{#35bf28}+0.99\%$
test_cql_speed[False-backward] 23.5447ms 23.1617ms 43.1748 Ops/s 43.2404 Ops/s $\color{#d91a1a}-0.15\%$
test_cql_speed[True-None] 3.7251ms 3.4705ms 288.1394 Ops/s 277.3763 Ops/s $\color{#35bf28}+3.88\%$
test_cql_speed[True-backward] 6.1590ms 5.8052ms 172.2594 Ops/s 174.0285 Ops/s $\color{#d91a1a}-1.02\%$
test_cql_speed[reduce-overhead-None] 17.4489ms 11.9948ms 83.3696 Ops/s 82.6919 Ops/s $\color{#35bf28}+0.82\%$
test_a2c_speed[False-None] 3.5041ms 3.3206ms 301.1548 Ops/s 298.8631 Ops/s $\color{#35bf28}+0.77\%$
test_a2c_speed[False-backward] 6.9332ms 6.4833ms 154.2433 Ops/s 152.2781 Ops/s $\color{#35bf28}+1.29\%$
test_a2c_speed[True-None] 1.5821ms 1.4745ms 678.1923 Ops/s 679.8552 Ops/s $\color{#d91a1a}-0.24\%$
test_a2c_speed[True-backward] 3.3395ms 3.2687ms 305.9345 Ops/s 300.4206 Ops/s $\color{#35bf28}+1.84\%$
test_a2c_speed[reduce-overhead-None] 1.1357ms 1.0813ms 924.8203 Ops/s 913.2838 Ops/s $\color{#35bf28}+1.26\%$
test_ppo_speed[False-None] 4.0346ms 3.9462ms 253.4058 Ops/s 243.6165 Ops/s $\color{#35bf28}+4.02\%$
test_ppo_speed[False-backward] 7.8131ms 7.1725ms 139.4213 Ops/s 134.0711 Ops/s $\color{#35bf28}+3.99\%$
test_ppo_speed[True-None] 1.6509ms 1.5631ms 639.7467 Ops/s 628.1225 Ops/s $\color{#35bf28}+1.85\%$
test_ppo_speed[True-backward] 3.3738ms 3.2950ms 303.4922 Ops/s 285.4864 Ops/s $\textbf{\color{#35bf28}+6.31\%}$
test_ppo_speed[reduce-overhead-None] 1.2339ms 1.1475ms 871.4848 Ops/s 858.2470 Ops/s $\color{#35bf28}+1.54\%$
test_reinforce_speed[False-None] 2.7755ms 2.4383ms 410.1223 Ops/s 423.5951 Ops/s $\color{#d91a1a}-3.18\%$
test_reinforce_speed[False-backward] 3.6405ms 3.5132ms 284.6431 Ops/s 282.8383 Ops/s $\color{#35bf28}+0.64\%$
test_reinforce_speed[True-None] 1.5810ms 1.4452ms 691.9401 Ops/s 705.1643 Ops/s $\color{#d91a1a}-1.88\%$
test_reinforce_speed[True-backward] 3.3145ms 3.2735ms 305.4805 Ops/s 301.4821 Ops/s $\color{#35bf28}+1.33\%$
test_reinforce_speed[reduce-overhead-None] 15.1208ms 8.7436ms 114.3689 Ops/s 115.1725 Ops/s $\color{#d91a1a}-0.70\%$
test_iql_speed[False-None] 10.0748ms 9.6403ms 103.7312 Ops/s 102.5506 Ops/s $\color{#35bf28}+1.15\%$
test_iql_speed[False-backward] 13.9198ms 13.4883ms 74.1386 Ops/s 74.2159 Ops/s $\color{#d91a1a}-0.10\%$
test_iql_speed[True-None] 2.4859ms 2.3238ms 430.3270 Ops/s 427.4255 Ops/s $\color{#35bf28}+0.68\%$
test_iql_speed[True-backward] 5.3596ms 4.9007ms 204.0514 Ops/s 195.6265 Ops/s $\color{#35bf28}+4.31\%$
test_iql_speed[reduce-overhead-None] 16.5059ms 10.0553ms 99.4502 Ops/s 99.5219 Ops/s $\color{#d91a1a}-0.07\%$
test_rb_sample[TensorDictReplayBuffer-ListStorage-RandomSampler-4000] 7.4618ms 5.8328ms 171.4447 Ops/s 170.6868 Ops/s $\color{#35bf28}+0.44\%$
test_rb_sample[TensorDictReplayBuffer-LazyMemmapStorage-RandomSampler-10000] 0.5476ms 0.3335ms 2.9987 KOps/s 2.9475 KOps/s $\color{#35bf28}+1.74\%$
test_rb_sample[TensorDictReplayBuffer-LazyTensorStorage-RandomSampler-10000] 0.7197ms 0.3247ms 3.0798 KOps/s 3.2165 KOps/s $\color{#d91a1a}-4.25\%$
test_rb_sample[TensorDictReplayBuffer-ListStorage-SamplerWithoutReplacement-4000] 6.0228ms 5.7328ms 174.4348 Ops/s 176.4170 Ops/s $\color{#d91a1a}-1.12\%$
test_rb_sample[TensorDictReplayBuffer-LazyMemmapStorage-SamplerWithoutReplacement-10000] 0.7013ms 0.3089ms 3.2376 KOps/s 3.0265 KOps/s $\textbf{\color{#35bf28}+6.97\%}$
test_rb_sample[TensorDictReplayBuffer-LazyTensorStorage-SamplerWithoutReplacement-10000] 0.5238ms 0.2851ms 3.5071 KOps/s 3.3978 KOps/s $\color{#35bf28}+3.22\%$
test_rb_sample[TensorDictReplayBuffer-LazyMemmapStorage-sampler6-10000] 1.4303ms 1.2519ms 798.7663 Ops/s 713.5320 Ops/s $\textbf{\color{#35bf28}+11.95\%}$
test_rb_sample[TensorDictReplayBuffer-LazyTensorStorage-sampler7-10000] 1.3897ms 1.1753ms 850.8485 Ops/s 758.1438 Ops/s $\textbf{\color{#35bf28}+12.23\%}$
test_rb_sample[TensorDictPrioritizedReplayBuffer-ListStorage-None-4000] 10.0303ms 6.0388ms 165.5959 Ops/s 173.4138 Ops/s $\color{#d91a1a}-4.51\%$
test_rb_sample[TensorDictPrioritizedReplayBuffer-LazyMemmapStorage-None-10000] 0.8842ms 0.4869ms 2.0538 KOps/s 2.0691 KOps/s $\color{#d91a1a}-0.74\%$
test_rb_sample[TensorDictPrioritizedReplayBuffer-LazyTensorStorage-None-10000] 0.6755ms 0.4431ms 2.2569 KOps/s 2.1139 KOps/s $\textbf{\color{#35bf28}+6.76\%}$
test_rb_iterate[TensorDictReplayBuffer-ListStorage-RandomSampler-4000] 5.8586ms 5.7123ms 175.0622 Ops/s 181.2391 Ops/s $\color{#d91a1a}-3.41\%$
test_rb_iterate[TensorDictReplayBuffer-LazyMemmapStorage-RandomSampler-10000] 1.2007ms 0.2821ms 3.5442 KOps/s 2.5978 KOps/s $\textbf{\color{#35bf28}+36.43\%}$
test_rb_iterate[TensorDictReplayBuffer-LazyTensorStorage-RandomSampler-10000] 0.4400ms 0.2625ms 3.8100 KOps/s 2.8335 KOps/s $\textbf{\color{#35bf28}+34.47\%}$
test_rb_iterate[TensorDictReplayBuffer-ListStorage-SamplerWithoutReplacement-4000] 5.9307ms 5.6574ms 176.7607 Ops/s 180.8049 Ops/s $\color{#d91a1a}-2.24\%$
test_rb_iterate[TensorDictReplayBuffer-LazyMemmapStorage-SamplerWithoutReplacement-10000] 1.9724ms 0.2817ms 3.5498 KOps/s 508.5591 Ops/s $\textbf{\color{#35bf28}+598.01\%}$
test_rb_iterate[TensorDictReplayBuffer-LazyTensorStorage-SamplerWithoutReplacement-10000] 0.5789ms 0.2627ms 3.8068 KOps/s 2.7876 KOps/s $\textbf{\color{#35bf28}+36.56\%}$
test_rb_iterate[TensorDictPrioritizedReplayBuffer-ListStorage-None-4000] 8.0242ms 5.8549ms 170.7975 Ops/s 168.9060 Ops/s $\color{#35bf28}+1.12\%$
test_rb_iterate[TensorDictPrioritizedReplayBuffer-LazyMemmapStorage-None-10000] 2.2409ms 0.4280ms 2.3363 KOps/s 1.9096 KOps/s $\textbf{\color{#35bf28}+22.34\%}$
test_rb_iterate[TensorDictPrioritizedReplayBuffer-LazyTensorStorage-None-10000] 0.6259ms 0.4114ms 2.4306 KOps/s 1.9799 KOps/s $\textbf{\color{#35bf28}+22.76\%}$
test_rb_populate[TensorDictReplayBuffer-ListStorage-RandomSampler-400] 0.7272s 19.4932ms 51.2999 Ops/s 198.1377 Ops/s $\textbf{\color{#d91a1a}-74.11\%}$
test_rb_populate[TensorDictReplayBuffer-LazyMemmapStorage-RandomSampler-400] 11.6285ms 1.9522ms 512.2431 Ops/s 542.9032 Ops/s $\textbf{\color{#d91a1a}-5.65\%}$
test_rb_populate[TensorDictReplayBuffer-LazyTensorStorage-RandomSampler-400] 8.0032ms 1.2713ms 786.6007 Ops/s 1.0459 KOps/s $\textbf{\color{#d91a1a}-24.79\%}$
test_rb_populate[TensorDictReplayBuffer-ListStorage-SamplerWithoutReplacement-400] 6.5051ms 5.0118ms 199.5282 Ops/s 195.8967 Ops/s $\color{#35bf28}+1.85\%$
test_rb_populate[TensorDictReplayBuffer-LazyMemmapStorage-SamplerWithoutReplacement-400] 3.8927ms 1.8353ms 544.8799 Ops/s 526.2787 Ops/s $\color{#35bf28}+3.53\%$
test_rb_populate[TensorDictReplayBuffer-LazyTensorStorage-SamplerWithoutReplacement-400] 5.7795ms 1.2638ms 791.2708 Ops/s 1.0916 KOps/s $\textbf{\color{#d91a1a}-27.51\%}$
test_rb_populate[TensorDictPrioritizedReplayBuffer-ListStorage-None-400] 6.8322ms 5.2037ms 192.1708 Ops/s 45.1678 Ops/s $\textbf{\color{#35bf28}+325.46\%}$
test_rb_populate[TensorDictPrioritizedReplayBuffer-LazyMemmapStorage-None-400] 13.3309ms 2.2555ms 443.3520 Ops/s 474.8756 Ops/s $\textbf{\color{#d91a1a}-6.64\%}$
test_rb_populate[TensorDictPrioritizedReplayBuffer-LazyTensorStorage-None-400] 1.4418ms 1.1174ms 894.9266 Ops/s 912.2069 Ops/s $\color{#d91a1a}-1.89\%$
test_rb_extend_sample[ReplayBuffer-LazyTensorStorage-RandomSampler-10000-10000-100-True] 43.6024ms 38.7138ms 25.8306 Ops/s 25.5193 Ops/s $\color{#35bf28}+1.22\%$
test_rb_extend_sample[ReplayBuffer-LazyTensorStorage-RandomSampler-10000-10000-100-False] 19.5959ms 18.2047ms 54.9309 Ops/s 55.3518 Ops/s $\color{#d91a1a}-0.76\%$
test_rb_extend_sample[ReplayBuffer-LazyTensorStorage-RandomSampler-100000-10000-100-True] 43.6303ms 40.6280ms 24.6135 Ops/s 24.9023 Ops/s $\color{#d91a1a}-1.16\%$
test_rb_extend_sample[ReplayBuffer-LazyTensorStorage-RandomSampler-100000-10000-100-False] 20.3024ms 18.8237ms 53.1244 Ops/s 54.0861 Ops/s $\color{#d91a1a}-1.78\%$
test_rb_extend_sample[ReplayBuffer-LazyTensorStorage-RandomSampler-1000000-10000-100-True] 43.3265ms 41.6468ms 24.0115 Ops/s 23.7839 Ops/s $\color{#35bf28}+0.96\%$
test_rb_extend_sample[ReplayBuffer-LazyTensorStorage-RandomSampler-1000000-10000-100-False] 21.0239ms 19.8885ms 50.2804 Ops/s 49.7920 Ops/s $\color{#35bf28}+0.98\%$
test_storage_write_lazystack[50-img_shape0-small] 0.8609ms 0.2214ms 4.5157 KOps/s 4.3540 KOps/s $\color{#35bf28}+3.71\%$
test_storage_write_lazystack[100-img_shape1-atari] 1.5807ms 1.3841ms 722.4965 Ops/s 718.7693 Ops/s $\color{#35bf28}+0.52\%$
test_storage_write_lazystack[100-img_shape2-large_img] 2.6057ms 2.2793ms 438.7373 Ops/s 424.0092 Ops/s $\color{#35bf28}+3.47\%$
test_storage_write_lazystack[200-img_shape3-large_batch] 3.1092ms 2.9119ms 343.4164 Ops/s 342.4873 Ops/s $\color{#35bf28}+0.27\%$
test_storage_write_contiguous[50-img_shape0-small] 0.3486ms 0.1609ms 6.2159 KOps/s 6.0563 KOps/s $\color{#35bf28}+2.63\%$
test_storage_write_contiguous[100-img_shape1-atari] 0.3954ms 0.2393ms 4.1795 KOps/s 4.0625 KOps/s $\color{#35bf28}+2.88\%$
test_storage_write_contiguous[100-img_shape2-large_img] 1.9266ms 1.8244ms 548.1177 Ops/s 562.3000 Ops/s $\color{#d91a1a}-2.52\%$
test_storage_write_contiguous[200-img_shape3-large_batch] 1.5438ms 1.3621ms 734.1772 Ops/s 722.9947 Ops/s $\color{#35bf28}+1.55\%$
test_collector_stack_then_write[50-img_shape0-small] 1.3756ms 1.1434ms 874.5671 Ops/s 885.1517 Ops/s $\color{#d91a1a}-1.20\%$
test_collector_stack_then_write[100-img_shape1-atari] 3.7321ms 3.5688ms 280.2051 Ops/s 278.8452 Ops/s $\color{#35bf28}+0.49\%$
test_collector_stack_then_write[100-img_shape2-large_img] 11.0877ms 5.7743ms 173.1802 Ops/s 175.9036 Ops/s $\color{#d91a1a}-1.55\%$
test_collector_stack_then_write[200-img_shape3-large_batch] 15.0110ms 7.0533ms 141.7771 Ops/s 143.5746 Ops/s $\color{#d91a1a}-1.25\%$
test_collector_lazystack_then_write[50-img_shape0-small] 0.4325ms 0.2818ms 3.5482 KOps/s 3.5025 KOps/s $\color{#35bf28}+1.30\%$
test_collector_lazystack_then_write[100-img_shape1-atari] 1.6473ms 1.5161ms 659.5819 Ops/s 662.4545 Ops/s $\color{#d91a1a}-0.43\%$
test_collector_lazystack_then_write[100-img_shape2-large_img] 2.8547ms 2.4359ms 410.5330 Ops/s 401.5269 Ops/s $\color{#35bf28}+2.24\%$
test_collector_lazystack_then_write[200-img_shape3-large_batch] 3.4505ms 3.1358ms 318.8981 Ops/s 318.0490 Ops/s $\color{#35bf28}+0.27\%$
test_collector_without_rb[100-img_shape0-atari] 33.7789ms 32.9316ms 30.3660 Ops/s 29.9814 Ops/s $\color{#35bf28}+1.28\%$
test_collector_without_rb[200-img_shape1-large_batch] 65.8317ms 64.4962ms 15.5048 Ops/s 15.6387 Ops/s $\color{#d91a1a}-0.86\%$
test_collector_with_rb[100-img_shape0-atari] 38.4607ms 37.4329ms 26.7145 Ops/s 27.0134 Ops/s $\color{#d91a1a}-1.11\%$
test_collector_with_rb[200-img_shape1-large_batch] 74.1273ms 72.8617ms 13.7246 Ops/s 13.8805 Ops/s $\color{#d91a1a}-1.12\%$
test_collector_without_rb_cuda[100-img_shape0-atari] 58.1211ms 56.2500ms 17.7778 Ops/s 17.8348 Ops/s $\color{#d91a1a}-0.32\%$
test_collector_without_rb_cuda[200-img_shape1-large_batch] 0.1158s 0.1135s 8.8101 Ops/s 8.9337 Ops/s $\color{#d91a1a}-1.38\%$
test_collector_with_rb_cuda[100-img_shape0-atari] 59.9476ms 58.2011ms 17.1818 Ops/s 17.3997 Ops/s $\color{#d91a1a}-1.25\%$
test_collector_with_rb_cuda[200-img_shape1-large_batch] 0.1195s 0.1171s 8.5364 Ops/s 8.7161 Ops/s $\color{#d91a1a}-2.06\%$

Resolve conflict: test/test_libs.py was deleted by main (split into
test/libs/) while our branch added TestTorchGeometric. Moved the new
tests to test/libs/test_torch_geometric.py and updated the CI script.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
@github-actions
Copy link
Copy Markdown
Contributor

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

Total Benchmarks: 172. Improved: $\large\color{#35bf28}15$. 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] 83.1501μs 82.1421μs 12.1740 KOps/s 12.5157 KOps/s $\color{#d91a1a}-2.73\%$
test_tensor_to_bytestream_speed[torch.save] 0.1420ms 0.1415ms 7.0666 KOps/s 7.1555 KOps/s $\color{#d91a1a}-1.24\%$
test_tensor_to_bytestream_speed[untyped_storage] 0.1054s 0.1050s 9.5225 Ops/s 9.5650 Ops/s $\color{#d91a1a}-0.44\%$
test_tensor_to_bytestream_speed[numpy] 2.4902μs 2.4870μs 402.0889 KOps/s 396.4185 KOps/s $\color{#35bf28}+1.43\%$
test_tensor_to_bytestream_speed[safetensors] 39.1061μs 38.7776μs 25.7881 KOps/s 25.9725 KOps/s $\color{#d91a1a}-0.71\%$
test_simple 0.5594s 0.5484s 1.8234 Ops/s 1.7444 Ops/s $\color{#35bf28}+4.53\%$
test_transformed 1.1179s 1.0983s 0.9105 Ops/s 0.8967 Ops/s $\color{#35bf28}+1.54\%$
test_serial 1.7079s 1.6948s 0.5900 Ops/s 0.5675 Ops/s $\color{#35bf28}+3.97\%$
test_parallel 1.0550s 1.0233s 0.9772 Ops/s 0.9792 Ops/s $\color{#d91a1a}-0.20\%$
test_step_mdp_speed[True-True-True-True-True] 0.3140ms 41.7858μs 23.9316 KOps/s 23.7558 KOps/s $\color{#35bf28}+0.74\%$
test_step_mdp_speed[True-True-True-True-False] 49.9530μs 22.9354μs 43.6007 KOps/s 43.2958 KOps/s $\color{#35bf28}+0.70\%$
test_step_mdp_speed[True-True-True-False-True] 52.7730μs 23.7317μs 42.1377 KOps/s 42.0033 KOps/s $\color{#35bf28}+0.32\%$
test_step_mdp_speed[True-True-True-False-False] 38.9820μs 12.9313μs 77.3316 KOps/s 77.9343 KOps/s $\color{#d91a1a}-0.77\%$
test_step_mdp_speed[True-True-False-True-True] 0.1035ms 44.9325μs 22.2556 KOps/s 22.4060 KOps/s $\color{#d91a1a}-0.67\%$
test_step_mdp_speed[True-True-False-True-False] 67.9140μs 25.4299μs 39.3238 KOps/s 39.0654 KOps/s $\color{#35bf28}+0.66\%$
test_step_mdp_speed[True-True-False-False-True] 98.8170μs 26.0449μs 38.3952 KOps/s 39.3188 KOps/s $\color{#d91a1a}-2.35\%$
test_step_mdp_speed[True-True-False-False-False] 42.6530μs 15.2143μs 65.7278 KOps/s 64.9060 KOps/s $\color{#35bf28}+1.27\%$
test_step_mdp_speed[True-False-True-True-True] 96.1360μs 47.3617μs 21.1141 KOps/s 20.8062 KOps/s $\color{#35bf28}+1.48\%$
test_step_mdp_speed[True-False-True-True-False] 73.8440μs 28.3376μs 35.2888 KOps/s 34.9210 KOps/s $\color{#35bf28}+1.05\%$
test_step_mdp_speed[True-False-True-False-True] 54.4140μs 26.2517μs 38.0928 KOps/s 38.2018 KOps/s $\color{#d91a1a}-0.29\%$
test_step_mdp_speed[True-False-True-False-False] 90.4460μs 15.2148μs 65.7254 KOps/s 63.9389 KOps/s $\color{#35bf28}+2.79\%$
test_step_mdp_speed[True-False-False-True-True] 0.5038ms 48.8848μs 20.4563 KOps/s 20.0293 KOps/s $\color{#35bf28}+2.13\%$
test_step_mdp_speed[True-False-False-True-False] 0.4576ms 30.0077μs 33.3248 KOps/s 32.3029 KOps/s $\color{#35bf28}+3.16\%$
test_step_mdp_speed[True-False-False-False-True] 59.4630μs 28.3080μs 35.3256 KOps/s 34.5984 KOps/s $\color{#35bf28}+2.10\%$
test_step_mdp_speed[True-False-False-False-False] 45.1220μs 17.7180μs 56.4398 KOps/s 55.4126 KOps/s $\color{#35bf28}+1.85\%$
test_step_mdp_speed[False-True-True-True-True] 0.4909ms 46.4122μs 21.5460 KOps/s 20.9936 KOps/s $\color{#35bf28}+2.63\%$
test_step_mdp_speed[False-True-True-True-False] 0.4722ms 27.7278μs 36.0648 KOps/s 35.0903 KOps/s $\color{#35bf28}+2.78\%$
test_step_mdp_speed[False-True-True-False-True] 2.5242ms 30.7512μs 32.5191 KOps/s 33.5324 KOps/s $\color{#d91a1a}-3.02\%$
test_step_mdp_speed[False-True-True-False-False] 0.4681ms 17.3438μs 57.6574 KOps/s 57.2793 KOps/s $\color{#35bf28}+0.66\%$
test_step_mdp_speed[False-True-False-True-True] 0.4941ms 49.9604μs 20.0158 KOps/s 20.5764 KOps/s $\color{#d91a1a}-2.72\%$
test_step_mdp_speed[False-True-False-True-False] 0.4672ms 30.2984μs 33.0050 KOps/s 32.8108 KOps/s $\color{#35bf28}+0.59\%$
test_step_mdp_speed[False-True-False-False-True] 67.5440μs 32.4621μs 30.8052 KOps/s 31.4202 KOps/s $\color{#d91a1a}-1.96\%$
test_step_mdp_speed[False-True-False-False-False] 0.4897ms 19.6581μs 50.8696 KOps/s 51.0320 KOps/s $\color{#d91a1a}-0.32\%$
test_step_mdp_speed[False-False-True-True-True] 0.4969ms 52.4169μs 19.0778 KOps/s 19.1201 KOps/s $\color{#d91a1a}-0.22\%$
test_step_mdp_speed[False-False-True-True-False] 0.4630ms 32.8015μs 30.4864 KOps/s 29.9038 KOps/s $\color{#35bf28}+1.95\%$
test_step_mdp_speed[False-False-True-False-True] 60.8930μs 31.9898μs 31.2600 KOps/s 30.9097 KOps/s $\color{#35bf28}+1.13\%$
test_step_mdp_speed[False-False-True-False-False] 0.4496ms 19.3425μs 51.6996 KOps/s 51.4110 KOps/s $\color{#35bf28}+0.56\%$
test_step_mdp_speed[False-False-False-True-True] 0.4862ms 53.0968μs 18.8335 KOps/s 18.4852 KOps/s $\color{#35bf28}+1.88\%$
test_step_mdp_speed[False-False-False-True-False] 0.5020ms 35.2228μs 28.3907 KOps/s 27.8887 KOps/s $\color{#35bf28}+1.80\%$
test_step_mdp_speed[False-False-False-False-True] 60.7440μs 33.4068μs 29.9340 KOps/s 29.4040 KOps/s $\color{#35bf28}+1.80\%$
test_step_mdp_speed[False-False-False-False-False] 0.4612ms 21.7532μs 45.9703 KOps/s 45.5050 KOps/s $\color{#35bf28}+1.02\%$
test_non_tensor_env_rollout_speed[1000-single-True] 0.8490s 0.7453s 1.3417 Ops/s 1.3232 Ops/s $\color{#35bf28}+1.40\%$
test_non_tensor_env_rollout_speed[1000-single-False] 0.7084s 0.6077s 1.6455 Ops/s 1.6164 Ops/s $\color{#35bf28}+1.80\%$
test_non_tensor_env_rollout_speed[1000-serial-no-buffers-True] 1.7509s 1.6536s 0.6047 Ops/s 0.5990 Ops/s $\color{#35bf28}+0.96\%$
test_non_tensor_env_rollout_speed[1000-serial-no-buffers-False] 1.5110s 1.4273s 0.7006 Ops/s 0.6931 Ops/s $\color{#35bf28}+1.09\%$
test_non_tensor_env_rollout_speed[1000-serial-buffers-True] 1.9692s 1.8918s 0.5286 Ops/s 0.5195 Ops/s $\color{#35bf28}+1.76\%$
test_non_tensor_env_rollout_speed[1000-serial-buffers-False] 1.7639s 1.6871s 0.5927 Ops/s 0.5906 Ops/s $\color{#35bf28}+0.35\%$
test_non_tensor_env_rollout_speed[1000-parallel-no-buffers-True] 4.7187s 4.6409s 0.2155 Ops/s 0.2180 Ops/s $\color{#d91a1a}-1.17\%$
test_non_tensor_env_rollout_speed[1000-parallel-no-buffers-False] 4.4739s 4.3806s 0.2283 Ops/s 0.2274 Ops/s $\color{#35bf28}+0.39\%$
test_non_tensor_env_rollout_speed[1000-parallel-buffers-True] 1.9821s 1.9053s 0.5249 Ops/s 0.5221 Ops/s $\color{#35bf28}+0.52\%$
test_non_tensor_env_rollout_speed[1000-parallel-buffers-False] 1.6921s 1.5980s 0.6258 Ops/s 0.6333 Ops/s $\color{#d91a1a}-1.18\%$
test_values[generalized_advantage_estimate-True-True] 10.3146ms 9.9741ms 100.2600 Ops/s 100.3207 Ops/s $\color{#d91a1a}-0.06\%$
test_values[vec_generalized_advantage_estimate-True-True] 15.3334ms 11.4467ms 87.3610 Ops/s 57.4036 Ops/s $\textbf{\color{#35bf28}+52.19\%}$
test_values[td0_return_estimate-False-False] 0.2327ms 0.1261ms 7.9318 KOps/s 8.3668 KOps/s $\textbf{\color{#d91a1a}-5.20\%}$
test_values[td1_return_estimate-False-False] 27.6023ms 27.1334ms 36.8549 Ops/s 36.3289 Ops/s $\color{#35bf28}+1.45\%$
test_values[vec_td1_return_estimate-False-False] 17.4018ms 11.7342ms 85.2206 Ops/s 57.0902 Ops/s $\textbf{\color{#35bf28}+49.27\%}$
test_values[td_lambda_return_estimate-True-False] 40.7159ms 40.1169ms 24.9271 Ops/s 24.4804 Ops/s $\color{#35bf28}+1.83\%$
test_values[vec_td_lambda_return_estimate-True-False] 11.9295ms 11.4873ms 87.0527 Ops/s 56.7882 Ops/s $\textbf{\color{#35bf28}+53.29\%}$
test_gae_speed[generalized_advantage_estimate-False-1-512] 8.9127ms 8.7318ms 114.5236 Ops/s 114.2643 Ops/s $\color{#35bf28}+0.23\%$
test_gae_speed[vec_generalized_advantage_estimate-True-1-512] 1.8187ms 1.5444ms 647.4972 Ops/s 644.2505 Ops/s $\color{#35bf28}+0.50\%$
test_gae_speed[vec_generalized_advantage_estimate-False-1-512] 0.4893ms 0.4254ms 2.3505 KOps/s 2.3792 KOps/s $\color{#d91a1a}-1.20\%$
test_gae_speed[vec_generalized_advantage_estimate-True-32-512] 36.0329ms 31.7722ms 31.4740 Ops/s 28.7947 Ops/s $\textbf{\color{#35bf28}+9.30\%}$
test_gae_speed[vec_generalized_advantage_estimate-False-32-512] 2.0824ms 1.7367ms 575.8099 Ops/s 569.4938 Ops/s $\color{#35bf28}+1.11\%$
test_dqn_speed[False-None] 1.5416ms 1.4200ms 704.2389 Ops/s 696.5192 Ops/s $\color{#35bf28}+1.11\%$
test_dqn_speed[False-backward] 1.9947ms 1.9367ms 516.3459 Ops/s 490.0585 Ops/s $\textbf{\color{#35bf28}+5.36\%}$
test_dqn_speed[True-None] 0.7503ms 0.5813ms 1.7203 KOps/s 1.6755 KOps/s $\color{#35bf28}+2.67\%$
test_dqn_speed[True-backward] 1.1275ms 1.0601ms 943.3290 Ops/s 944.1678 Ops/s $\color{#d91a1a}-0.09\%$
test_dqn_speed[reduce-overhead-None] 0.7312ms 0.5615ms 1.7809 KOps/s 1.7512 KOps/s $\color{#35bf28}+1.69\%$
test_ddpg_speed[False-None] 3.2836ms 2.8832ms 346.8351 Ops/s 347.2953 Ops/s $\color{#d91a1a}-0.13\%$
test_ddpg_speed[False-backward] 4.2684ms 4.1266ms 242.3291 Ops/s 243.5587 Ops/s $\color{#d91a1a}-0.50\%$
test_ddpg_speed[True-None] 1.7001ms 1.4687ms 680.8582 Ops/s 667.8000 Ops/s $\color{#35bf28}+1.96\%$
test_ddpg_speed[True-backward] 2.5510ms 2.4850ms 402.4096 Ops/s 364.8289 Ops/s $\textbf{\color{#35bf28}+10.30\%}$
test_ddpg_speed[reduce-overhead-None] 1.6369ms 1.4638ms 683.1620 Ops/s 671.9330 Ops/s $\color{#35bf28}+1.67\%$
test_sac_speed[False-None] 8.7599ms 8.1404ms 122.8434 Ops/s 122.9770 Ops/s $\color{#d91a1a}-0.11\%$
test_sac_speed[False-backward] 12.1182ms 11.4301ms 87.4881 Ops/s 87.6551 Ops/s $\color{#d91a1a}-0.19\%$
test_sac_speed[True-None] 2.4855ms 2.2776ms 439.0674 Ops/s 432.1747 Ops/s $\color{#35bf28}+1.59\%$
test_sac_speed[True-backward] 4.5017ms 4.3123ms 231.8928 Ops/s 228.5952 Ops/s $\color{#35bf28}+1.44\%$
test_sac_speed[reduce-overhead-None] 2.5710ms 2.2577ms 442.9378 Ops/s 441.6195 Ops/s $\color{#35bf28}+0.30\%$
test_redq_speed[False-None] 15.2482ms 10.7998ms 92.5940 Ops/s 91.0374 Ops/s $\color{#35bf28}+1.71\%$
test_redq_speed[False-backward] 19.6755ms 18.7589ms 53.3080 Ops/s 54.7515 Ops/s $\color{#d91a1a}-2.64\%$
test_redq_speed[True-None] 5.1118ms 4.7983ms 208.4078 Ops/s 207.7825 Ops/s $\color{#35bf28}+0.30\%$
test_redq_speed[reduce-overhead-None] 5.0747ms 4.6913ms 213.1624 Ops/s 210.8196 Ops/s $\color{#35bf28}+1.11\%$
test_redq_deprec_speed[False-None] 12.1010ms 11.4618ms 87.2466 Ops/s 88.7416 Ops/s $\color{#d91a1a}-1.68\%$
test_redq_deprec_speed[False-backward] 17.2136ms 16.5468ms 60.4346 Ops/s 62.2282 Ops/s $\color{#d91a1a}-2.88\%$
test_redq_deprec_speed[True-None] 4.1788ms 3.8014ms 263.0636 Ops/s 272.9205 Ops/s $\color{#d91a1a}-3.61\%$
test_redq_deprec_speed[True-backward] 10.0339ms 8.0444ms 124.3108 Ops/s 129.6331 Ops/s $\color{#d91a1a}-4.11\%$
test_redq_deprec_speed[reduce-overhead-None] 5.0298ms 3.7385ms 267.4902 Ops/s 246.2127 Ops/s $\textbf{\color{#35bf28}+8.64\%}$
test_td3_speed[False-None] 8.2771ms 8.1638ms 122.4922 Ops/s 121.6526 Ops/s $\color{#35bf28}+0.69\%$
test_td3_speed[False-backward] 11.5814ms 11.0678ms 90.3520 Ops/s 89.7403 Ops/s $\color{#35bf28}+0.68\%$
test_td3_speed[True-None] 1.9528ms 1.8665ms 535.7556 Ops/s 527.2732 Ops/s $\color{#35bf28}+1.61\%$
test_td3_speed[True-backward] 3.8759ms 3.7661ms 265.5290 Ops/s 274.0768 Ops/s $\color{#d91a1a}-3.12\%$
test_td3_speed[reduce-overhead-None] 1.9472ms 1.8277ms 547.1332 Ops/s 534.0988 Ops/s $\color{#35bf28}+2.44\%$
test_cql_speed[False-None] 31.4472ms 27.2037ms 36.7598 Ops/s 37.7102 Ops/s $\color{#d91a1a}-2.52\%$
test_cql_speed[False-backward] 37.4414ms 36.3463ms 27.5131 Ops/s 27.3308 Ops/s $\color{#35bf28}+0.67\%$
test_cql_speed[True-None] 13.5808ms 12.9182ms 77.4100 Ops/s 74.3010 Ops/s $\color{#35bf28}+4.18\%$
test_cql_speed[True-backward] 19.4492ms 18.6850ms 53.5188 Ops/s 54.0112 Ops/s $\color{#d91a1a}-0.91\%$
test_cql_speed[reduce-overhead-None] 13.3914ms 12.7806ms 78.2436 Ops/s 76.8554 Ops/s $\color{#35bf28}+1.81\%$
test_a2c_speed[False-None] 5.9332ms 5.4362ms 183.9520 Ops/s 179.8928 Ops/s $\color{#35bf28}+2.26\%$
test_a2c_speed[False-backward] 12.5611ms 12.0516ms 82.9766 Ops/s 82.3660 Ops/s $\color{#35bf28}+0.74\%$
test_a2c_speed[True-None] 4.0985ms 3.8728ms 258.2110 Ops/s 252.0333 Ops/s $\color{#35bf28}+2.45\%$
test_a2c_speed[True-backward] 9.3854ms 8.9744ms 111.4276 Ops/s 106.4067 Ops/s $\color{#35bf28}+4.72\%$
test_a2c_speed[reduce-overhead-None] 4.0649ms 3.8757ms 258.0146 Ops/s 248.1010 Ops/s $\color{#35bf28}+4.00\%$
test_ppo_speed[False-None] 6.3931ms 6.0328ms 165.7596 Ops/s 162.0138 Ops/s $\color{#35bf28}+2.31\%$
test_ppo_speed[False-backward] 13.5221ms 12.9137ms 77.4374 Ops/s 76.0715 Ops/s $\color{#35bf28}+1.80\%$
test_ppo_speed[True-None] 10.8770ms 3.9307ms 254.4096 Ops/s 253.0512 Ops/s $\color{#35bf28}+0.54\%$
test_ppo_speed[True-backward] 9.3140ms 8.8983ms 112.3806 Ops/s 110.8933 Ops/s $\color{#35bf28}+1.34\%$
test_ppo_speed[reduce-overhead-None] 4.0796ms 3.8783ms 257.8441 Ops/s 258.7446 Ops/s $\color{#d91a1a}-0.35\%$
test_reinforce_speed[False-None] 5.0975ms 4.6841ms 213.4894 Ops/s 211.3709 Ops/s $\color{#35bf28}+1.00\%$
test_reinforce_speed[False-backward] 8.3387ms 7.7497ms 129.0377 Ops/s 131.1726 Ops/s $\color{#d91a1a}-1.63\%$
test_reinforce_speed[True-None] 3.3644ms 3.1076ms 321.7967 Ops/s 325.3790 Ops/s $\color{#d91a1a}-1.10\%$
test_reinforce_speed[True-backward] 8.7475ms 8.1435ms 122.7968 Ops/s 118.8151 Ops/s $\color{#35bf28}+3.35\%$
test_reinforce_speed[reduce-overhead-None] 3.3999ms 3.0985ms 322.7376 Ops/s 317.6588 Ops/s $\color{#35bf28}+1.60\%$
test_iql_speed[False-None] 26.7106ms 21.5270ms 46.4533 Ops/s 47.9256 Ops/s $\color{#d91a1a}-3.07\%$
test_iql_speed[False-backward] 38.3944ms 31.9156ms 31.3327 Ops/s 31.6815 Ops/s $\color{#d91a1a}-1.10\%$
test_iql_speed[True-None] 9.1583ms 8.7457ms 114.3420 Ops/s 112.9438 Ops/s $\color{#35bf28}+1.24\%$
test_iql_speed[True-backward] 18.3696ms 17.3256ms 57.7181 Ops/s 54.6003 Ops/s $\textbf{\color{#35bf28}+5.71\%}$
test_iql_speed[reduce-overhead-None] 9.4425ms 8.7693ms 114.0348 Ops/s 113.0069 Ops/s $\color{#35bf28}+0.91\%$
test_rb_sample[TensorDictReplayBuffer-ListStorage-RandomSampler-4000] 6.4492ms 6.0988ms 163.9658 Ops/s 163.2565 Ops/s $\color{#35bf28}+0.43\%$
test_rb_sample[TensorDictReplayBuffer-LazyMemmapStorage-RandomSampler-10000] 3.1914ms 0.3370ms 2.9675 KOps/s 3.2650 KOps/s $\textbf{\color{#d91a1a}-9.11\%}$
test_rb_sample[TensorDictReplayBuffer-LazyTensorStorage-RandomSampler-10000] 0.6943ms 0.3549ms 2.8178 KOps/s 3.5707 KOps/s $\textbf{\color{#d91a1a}-21.09\%}$
test_rb_sample[TensorDictReplayBuffer-ListStorage-SamplerWithoutReplacement-4000] 6.1322ms 5.8822ms 170.0054 Ops/s 169.9011 Ops/s $\color{#35bf28}+0.06\%$
test_rb_sample[TensorDictReplayBuffer-LazyMemmapStorage-SamplerWithoutReplacement-10000] 0.7332ms 0.3486ms 2.8689 KOps/s 3.3849 KOps/s $\textbf{\color{#d91a1a}-15.24\%}$
test_rb_sample[TensorDictReplayBuffer-LazyTensorStorage-SamplerWithoutReplacement-10000] 0.5840ms 0.3430ms 2.9157 KOps/s 3.1818 KOps/s $\textbf{\color{#d91a1a}-8.36\%}$
test_rb_sample[TensorDictReplayBuffer-LazyMemmapStorage-sampler6-10000] 1.5386ms 1.2871ms 776.9455 Ops/s 709.8201 Ops/s $\textbf{\color{#35bf28}+9.46\%}$
test_rb_sample[TensorDictReplayBuffer-LazyTensorStorage-sampler7-10000] 1.4256ms 1.2142ms 823.6122 Ops/s 749.0712 Ops/s $\textbf{\color{#35bf28}+9.95\%}$
test_rb_sample[TensorDictPrioritizedReplayBuffer-ListStorage-None-4000] 9.5109ms 6.0922ms 164.1451 Ops/s 166.3714 Ops/s $\color{#d91a1a}-1.34\%$
test_rb_sample[TensorDictPrioritizedReplayBuffer-LazyMemmapStorage-None-10000] 1.6004ms 0.4837ms 2.0675 KOps/s 2.0896 KOps/s $\color{#d91a1a}-1.06\%$
test_rb_sample[TensorDictPrioritizedReplayBuffer-LazyTensorStorage-None-10000] 0.7307ms 0.4802ms 2.0825 KOps/s 2.1469 KOps/s $\color{#d91a1a}-3.00\%$
test_rb_iterate[TensorDictReplayBuffer-ListStorage-RandomSampler-4000] 6.1606ms 5.8795ms 170.0837 Ops/s 170.6910 Ops/s $\color{#d91a1a}-0.36\%$
test_rb_iterate[TensorDictReplayBuffer-LazyMemmapStorage-RandomSampler-10000] 0.7868ms 0.3211ms 3.1140 KOps/s 2.8438 KOps/s $\textbf{\color{#35bf28}+9.50\%}$
test_rb_iterate[TensorDictReplayBuffer-LazyTensorStorage-RandomSampler-10000] 0.6270ms 0.3620ms 2.7627 KOps/s 3.0424 KOps/s $\textbf{\color{#d91a1a}-9.19\%}$
test_rb_iterate[TensorDictReplayBuffer-ListStorage-SamplerWithoutReplacement-4000] 6.0869ms 5.8019ms 172.3574 Ops/s 171.0432 Ops/s $\color{#35bf28}+0.77\%$
test_rb_iterate[TensorDictReplayBuffer-LazyMemmapStorage-SamplerWithoutReplacement-10000] 0.8610ms 0.3648ms 2.7414 KOps/s 2.9319 KOps/s $\textbf{\color{#d91a1a}-6.50\%}$
test_rb_iterate[TensorDictReplayBuffer-LazyTensorStorage-SamplerWithoutReplacement-10000] 0.5664ms 0.3360ms 2.9765 KOps/s 2.9300 KOps/s $\color{#35bf28}+1.59\%$
test_rb_iterate[TensorDictPrioritizedReplayBuffer-ListStorage-None-4000] 6.1144ms 5.9953ms 166.7970 Ops/s 167.1006 Ops/s $\color{#d91a1a}-0.18\%$
test_rb_iterate[TensorDictPrioritizedReplayBuffer-LazyMemmapStorage-None-10000] 2.1843ms 0.4908ms 2.0375 KOps/s 2.0912 KOps/s $\color{#d91a1a}-2.57\%$
test_rb_iterate[TensorDictPrioritizedReplayBuffer-LazyTensorStorage-None-10000] 0.6608ms 0.4611ms 2.1688 KOps/s 2.0783 KOps/s $\color{#35bf28}+4.35\%$
test_rb_populate[TensorDictReplayBuffer-ListStorage-RandomSampler-400] 6.5165ms 5.0162ms 199.3534 Ops/s 194.9912 Ops/s $\color{#35bf28}+2.24\%$
test_rb_populate[TensorDictReplayBuffer-LazyMemmapStorage-RandomSampler-400] 9.4998ms 2.1868ms 457.2855 Ops/s 496.0490 Ops/s $\textbf{\color{#d91a1a}-7.81\%}$
test_rb_populate[TensorDictReplayBuffer-LazyTensorStorage-RandomSampler-400] 3.1665ms 0.9256ms 1.0804 KOps/s 1.0872 KOps/s $\color{#d91a1a}-0.63\%$
test_rb_populate[TensorDictReplayBuffer-ListStorage-SamplerWithoutReplacement-400] 0.6730s 18.4781ms 54.1180 Ops/s 49.1776 Ops/s $\textbf{\color{#35bf28}+10.05\%}$
test_rb_populate[TensorDictReplayBuffer-LazyMemmapStorage-SamplerWithoutReplacement-400] 3.9259ms 1.8602ms 537.5736 Ops/s 536.1700 Ops/s $\color{#35bf28}+0.26\%$
test_rb_populate[TensorDictReplayBuffer-LazyTensorStorage-SamplerWithoutReplacement-400] 7.2513ms 1.2121ms 825.0484 Ops/s 923.0924 Ops/s $\textbf{\color{#d91a1a}-10.62\%}$
test_rb_populate[TensorDictPrioritizedReplayBuffer-ListStorage-None-400] 6.7766ms 5.2560ms 190.2586 Ops/s 185.7243 Ops/s $\color{#35bf28}+2.44\%$
test_rb_populate[TensorDictPrioritizedReplayBuffer-LazyMemmapStorage-None-400] 13.1351ms 2.0812ms 480.4910 Ops/s 480.7290 Ops/s $\color{#d91a1a}-0.05\%$
test_rb_populate[TensorDictPrioritizedReplayBuffer-LazyTensorStorage-None-400] 1.4189ms 1.0970ms 911.5634 Ops/s 805.9124 Ops/s $\textbf{\color{#35bf28}+13.11\%}$
test_rb_extend_sample[ReplayBuffer-LazyTensorStorage-RandomSampler-10000-10000-100-True] 43.6915ms 39.6638ms 25.2119 Ops/s 25.1833 Ops/s $\color{#35bf28}+0.11\%$
test_rb_extend_sample[ReplayBuffer-LazyTensorStorage-RandomSampler-10000-10000-100-False] 20.4114ms 18.8633ms 53.0131 Ops/s 53.2720 Ops/s $\color{#d91a1a}-0.49\%$
test_rb_extend_sample[ReplayBuffer-LazyTensorStorage-RandomSampler-100000-10000-100-True] 45.1084ms 40.8803ms 24.4617 Ops/s 24.5157 Ops/s $\color{#d91a1a}-0.22\%$
test_rb_extend_sample[ReplayBuffer-LazyTensorStorage-RandomSampler-100000-10000-100-False] 20.9900ms 19.1699ms 52.1652 Ops/s 31.3775 Ops/s $\textbf{\color{#35bf28}+66.25\%}$
test_rb_extend_sample[ReplayBuffer-LazyTensorStorage-RandomSampler-1000000-10000-100-True] 44.2019ms 42.5133ms 23.5221 Ops/s 23.3279 Ops/s $\color{#35bf28}+0.83\%$
test_rb_extend_sample[ReplayBuffer-LazyTensorStorage-RandomSampler-1000000-10000-100-False] 21.8003ms 20.5583ms 48.6421 Ops/s 48.8434 Ops/s $\color{#d91a1a}-0.41\%$
test_storage_write_lazystack[50-img_shape0-small] 0.8795ms 0.2326ms 4.2994 KOps/s 4.2614 KOps/s $\color{#35bf28}+0.89\%$
test_storage_write_lazystack[100-img_shape1-atari] 1.7665ms 1.4835ms 674.0760 Ops/s 683.8767 Ops/s $\color{#d91a1a}-1.43\%$
test_storage_write_lazystack[100-img_shape2-large_img] 2.7487ms 2.4040ms 415.9765 Ops/s 401.1844 Ops/s $\color{#35bf28}+3.69\%$
test_storage_write_lazystack[200-img_shape3-large_batch] 3.6826ms 3.0574ms 327.0794 Ops/s 328.6726 Ops/s $\color{#d91a1a}-0.48\%$
test_storage_write_contiguous[50-img_shape0-small] 0.2220ms 0.1380ms 7.2475 KOps/s 7.2879 KOps/s $\color{#d91a1a}-0.55\%$
test_storage_write_contiguous[100-img_shape1-atari] 0.3557ms 0.1787ms 5.5960 KOps/s 5.1498 KOps/s $\textbf{\color{#35bf28}+8.66\%}$
test_storage_write_contiguous[100-img_shape2-large_img] 2.2255ms 1.8477ms 541.2132 Ops/s 545.7254 Ops/s $\color{#d91a1a}-0.83\%$
test_storage_write_contiguous[200-img_shape3-large_batch] 1.6328ms 1.3512ms 740.0725 Ops/s 741.0902 Ops/s $\color{#d91a1a}-0.14\%$
test_collector_stack_then_write[50-img_shape0-small] 1.2153ms 1.1310ms 884.1750 Ops/s 876.0358 Ops/s $\color{#35bf28}+0.93\%$
test_collector_stack_then_write[100-img_shape1-atari] 6.2887ms 3.7443ms 267.0735 Ops/s 276.0824 Ops/s $\color{#d91a1a}-3.26\%$
test_collector_stack_then_write[100-img_shape2-large_img] 11.4668ms 5.8304ms 171.5151 Ops/s 170.7034 Ops/s $\color{#35bf28}+0.48\%$
test_collector_stack_then_write[200-img_shape3-large_batch] 13.9141ms 7.5630ms 132.2233 Ops/s 136.8827 Ops/s $\color{#d91a1a}-3.40\%$
test_collector_lazystack_then_write[50-img_shape0-small] 0.4418ms 0.2780ms 3.5969 KOps/s 3.4982 KOps/s $\color{#35bf28}+2.82\%$
test_collector_lazystack_then_write[100-img_shape1-atari] 2.0436ms 1.6539ms 604.6414 Ops/s 629.3624 Ops/s $\color{#d91a1a}-3.93\%$
test_collector_lazystack_then_write[100-img_shape2-large_img] 3.3719ms 2.6210ms 381.5396 Ops/s 381.7910 Ops/s $\color{#d91a1a}-0.07\%$
test_collector_lazystack_then_write[200-img_shape3-large_batch] 3.7677ms 3.2628ms 306.4882 Ops/s 304.9370 Ops/s $\color{#35bf28}+0.51\%$
test_collector_without_rb[100-img_shape0-atari] 34.1851ms 33.1886ms 30.1308 Ops/s 29.3914 Ops/s $\color{#35bf28}+2.52\%$
test_collector_without_rb[200-img_shape1-large_batch] 67.7079ms 65.2053ms 15.3362 Ops/s 15.1721 Ops/s $\color{#35bf28}+1.08\%$
test_collector_with_rb[100-img_shape0-atari] 39.5132ms 38.1265ms 26.2285 Ops/s 25.9866 Ops/s $\color{#35bf28}+0.93\%$
test_collector_with_rb[200-img_shape1-large_batch] 75.2981ms 73.7796ms 13.5539 Ops/s 13.2434 Ops/s $\color{#35bf28}+2.34\%$

@vmoens vmoens merged commit 5b5d3f0 into main Apr 12, 2026
97 of 109 checks passed
@vmoens vmoens deleted the ci/torch-geometric-tests branch April 12, 2026 17:03
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

CI Has to do with CI setup (e.g. wheels & builds, tests...) CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed. Environments Adds or modifies an environment wrapper

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant