Skip to content

[Doc] Clarify headless MuJoCo EGL dependencies#3879

Merged
vmoens merged 13 commits into
gh/vmoens/291/basefrom
gh/vmoens/291/head
Jun 19, 2026
Merged

[Doc] Clarify headless MuJoCo EGL dependencies#3879
vmoens merged 13 commits into
gh/vmoens/291/basefrom
gh/vmoens/291/head

Conversation

@vmoens

@vmoens vmoens commented Jun 17, 2026 •
edited
Loading

Copy link
Copy Markdown
Collaborator

@vmoens
Update
fa177dd 
[ghstack-poisoned]
@pytorch-bot

pytorch-bot Bot commented Jun 17, 2026 •
edited
Loading

Copy link
Copy Markdown

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/rl/3879

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:

❌ You can merge normally! (1 Unrelated Failure), 2 Unclassified Failures

As of commit 284223e with merge base 39f8c0f (image):

UNCLASSIFIED FAILURES - DrCI could not classify the following jobs because the workflow did not run on the merge base. The failures may be pre-existing on trunk or introduced by this PR:

BROKEN TRUNK - The following job 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.

This was referenced Jun 17, 2026
Merged
Merged
Merged
Merged
Merged
Merged
@github-actions github-actions Bot added the Documentation Improvements or additions to documentation label Jun 17, 2026
@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 Jun 17, 2026
@vmoens
Update
ad34a9e 
[ghstack-poisoned]
vmoens added a commit that referenced this pull request Jun 17, 2026
@vmoens
[Doc] Clarify headless MuJoCo EGL dependencies
ed55444 
ghstack-source-id: e1d2861
Pull-Request: #3879
@vmoens
Update
af26e94 
[ghstack-poisoned]
vmoens added a commit that referenced this pull request Jun 17, 2026
@vmoens
[Doc] Clarify headless MuJoCo EGL dependencies
541d210 
ghstack-source-id: d734501
Pull-Request: #3879
@github-actions

github-actions Bot commented Jun 17, 2026 •
edited
Loading

Copy link
Copy Markdown
Contributor

Benchmark Results: PR 4af53353 vs main fc9ef14b

Benchmark run: https://github.com/pytorch/rl/actions/runs/27813896457

Higher ops/sec is better. Tables are sorted by largest absolute change.

CPU

Compared 192 benchmarks. Regressions over 5%: 14. Improvements over 5%: 21.

Benchmark main ops PR ops Change
benchmarks/test_replaybuffer_benchmark.py::test_rb_populate[TensorDictReplayBuffer-ListStorage-RandomSampler-400] 36.71 195.84 +433.46%
benchmarks/test_replaybuffer_benchmark.py::test_rb_populate[TensorDictReplayBuffer-ListStorage-SamplerWithoutReplacement-400] 194.88 53.11 -72.75%
benchmarks/test_objectives_benchmarks.py::test_values[vec_td_lambda_return_estimate-True-False] 54.35 87.71 +61.39%
benchmarks/test_objectives_benchmarks.py::test_values[vec_td1_return_estimate-False-False] 55.33 87.40 +57.98%
benchmarks/test_objectives_benchmarks.py::test_values[vec_generalized_advantage_estimate-True-True] 56.25 88.09 +56.62%
benchmarks/test_objectives_benchmarks.py::test_dqn_speed[True-backward] 799.84 969.06 +21.16%
benchmarks/test_envs_benchmark.py::test_cat_frames_functional[4-same] 23.07 27.25 +18.09%
benchmarks/test_replaybuffer_benchmark.py::test_rb_sample[TensorDictReplayBuffer-LazyMemmapStorage-SamplerWithoutReplacement-10000] 2,913 3,422 +17.46%
benchmarks/test_replaybuffer_benchmark.py::test_rb_sample[TensorDictReplayBuffer-LazyTensorStorage-SamplerWithoutReplacement-10000] 3,122 3,660 +17.22%
benchmarks/test_replaybuffer_benchmark.py::test_rb_sample[TensorDictReplayBuffer-LazyMemmapStorage-sampler6-10000] 681.95 777.82 +14.06%
benchmarks/test_replaybuffer_benchmark.py::test_rb_sample[TensorDictReplayBuffer-LazyTensorStorage-sampler7-10000] 745.19 835.84 +12.17%
benchmarks/test_replaybuffer_benchmark.py::test_rb_iterate[TensorDictReplayBuffer-LazyTensorStorage-RandomSampler-10000] 3,279 2,888 -11.94%
benchmarks/test_replaybuffer_benchmark.py::test_rb_iterate[TensorDictReplayBuffer-LazyTensorStorage-SamplerWithoutReplacement-10000] 3,294 2,927 -11.13%
benchmarks/test_envs_benchmark.py::test_cat_frames_functional[16-same] 21.42 19.06 -10.98%
benchmarks/test_replaybuffer_benchmark.py::test_rb_sample[TensorDictPrioritizedReplayBuffer-LazyTensorStorage-None-10000] 1,904 2,098 +10.16%
benchmarks/test_replaybuffer_benchmark.py::test_rb_sample[TensorDictReplayBuffer-LazyMemmapStorage-RandomSampler-10000] 3,068 2,811 -8.40%
benchmarks/test_rnn_reset_backends_benchmark.py::test_rnn_rollout_with_intermediate_resets[b256-t128-i32-h512-scan-True-0-lstm] 2.8848 3.1223 +8.24%
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[False-False-True-False-False] 49,988 46,099 -7.78%
benchmarks/test_storage_write_benchmark.py::TestStorageWriteBenchmark::test_storage_write_contiguous[100-img_shape1-atari] 4,841 5,199 +7.40%
benchmarks/test_replaybuffer_benchmark.py::test_rb_populate[TensorDictReplayBuffer-LazyMemmapStorage-SamplerWithoutReplacement-400] 545.21 506.47 -7.11%
benchmarks/test_objectives_benchmarks.py::test_ddpg_speed[True-backward] 373.82 399.27 +6.81%
benchmarks/test_storage_write_benchmark.py::TestStorageWriteBenchmark::test_storage_write_contiguous[200-img_shape3-large_batch] 774.37 723.43 -6.58%
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[False-False-False-True-True] 18,000 16,897 -6.13%
benchmarks/test_objectives_benchmarks.py::test_cql_speed[False-None] 38.73 36.38 -6.06%
benchmarks/test_replaybuffer_benchmark.py::test_rb_sample[TensorDictPrioritizedReplayBuffer-LazyMemmapStorage-None-10000] 2,202 2,071 -5.98%
benchmarks/test_replaybuffer_benchmark.py::test_rb_populate[TensorDictPrioritizedReplayBuffer-LazyMemmapStorage-None-400] 430.11 454.76 +5.73%
benchmarks/test_rnn_reset_backends_benchmark.py::test_rnn_rollout_with_intermediate_resets[b256-t128-i32-h512-cudnn-True-0-lstm] 0.8790 0.9282 +5.60%
benchmarks/test_envs_benchmark.py::test_transformed 0.8649 0.9125 +5.50%
benchmarks/test_collectors_benchmark.py::test_single 8.3263 8.7775 +5.42%
benchmarks/test_objectives_benchmarks.py::test_ddpg_speed[reduce-overhead-None] 660.59 695.71 +5.32%
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[True-False-True-True-True] 21,072 19,953 -5.31%
benchmarks/test_storage_write_benchmark.py::TestStorageWriteBenchmark::test_storage_write_lazystack[50-img_shape0-small] 4,234 4,011 -5.27%
benchmarks/test_objectives_benchmarks.py::test_redq_speed[True-None] 211.65 222.74 +5.24%
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[True-False-False-False-False] 54,807 51,991 -5.14%
benchmarks/test_objectives_benchmarks.py::test_redq_deprec_speed[True-backward] 130.86 137.56 +5.12%
benchmarks/test_storage_write_benchmark.py::TestStorageWriteBenchmark::test_collector_stack_then_write[100-img_shape1-atari] 265.56 278.80 +4.98%
benchmarks/test_objectives_benchmarks.py::test_dqn_speed[reduce-overhead-None] 1,759 1,844 +4.83%
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[False-False-True-False-True] 30,630 29,158 -4.80%
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[False-False-True-True-True] 18,540 17,660 -4.75%
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[False-True-True-True-True] 20,480 19,513 -4.72%
benchmarks/test_envs_benchmark.py::test_cat_frames_functional[16-constant] 2,470 2,586 +4.68%
benchmarks/test_compressed_storage_benchmark.py::TestCompressedStorageBenchmark::test_tensor_to_bytestream_speed[pickle] 11,219 11,738 +4.62%
benchmarks/test_storage_write_benchmark.py::TestStorageWriteBenchmark::test_storage_write_contiguous[100-img_shape2-large_img] 576.96 551.04 -4.49%
benchmarks/test_storage_write_benchmark.py::TestStorageWriteBenchmark::test_storage_write_lazystack[100-img_shape2-large_img] 404.05 421.41 +4.30%
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[True-False-True-True-False] 34,721 33,300 -4.09%
benchmarks/test_replaybuffer_benchmark.py::test_rb_extend_sample[ReplayBuffer-LazyTensorStorage-RandomSampler-10000-10000-100-False] 51.42 53.46 +3.97%
benchmarks/test_replaybuffer_benchmark.py::test_rb_populate[TensorDictPrioritizedReplayBuffer-LazyTensorStorage-None-400] 875.17 840.85 -3.92%
benchmarks/test_storage_write_benchmark.py::TestStorageWriteBenchmark::test_collector_lazystack_then_write[100-img_shape2-large_img] 388.60 403.36 +3.80%
benchmarks/test_rnn_reset_backends_benchmark.py::test_rnn_rollout_with_intermediate_resets[b256-t128-i32-h512-scan-False-0-lstm] 1.9304 2.0015 +3.68%
benchmarks/test_objectives_benchmarks.py::test_gae_speed[vec_generalized_advantage_estimate-False-1-512] 2,167 2,245 +3.62%
benchmarks/test_storage_write_benchmark.py::TestStorageWriteBenchmark::test_collector_stack_then_write[50-img_shape0-small] 844.92 875.18 +3.58%
benchmarks/test_objectives_benchmarks.py::test_redq_deprec_speed[True-None] 264.22 273.60 +3.55%
benchmarks/test_storage_write_benchmark.py::TestStorageWriteBenchmark::test_collector_lazystack_then_write[200-img_shape3-large_batch] 302.77 313.40 +3.51%
benchmarks/test_non_tensor_env_benchmark.py::test_non_tensor_env_rollout_speed[1000-single-False] 1.5961 1.5404 -3.49%
benchmarks/test_objectives_benchmarks.py::test_gae_speed[generalized_advantage_estimate-False-1-512] 102.69 106.22 +3.44%
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[False-True-False-False-False] 50,044 48,328 -3.43%
benchmarks/test_storage_write_benchmark.py::TestStorageWriteBenchmark::test_storage_write_contiguous[50-img_shape0-small] 6,977 7,211 +3.35%
benchmarks/test_replaybuffer_benchmark.py::test_rb_sample[TensorDictReplayBuffer-LazyTensorStorage-RandomSampler-10000] 3,088 3,190 +3.33%
benchmarks/test_rnn_reset_backends_benchmark.py::test_rnn_rollout_with_intermediate_resets[b256-t128-i32-h512-cudnn-True-0-gru] 1.4777 1.4296 -3.26%
benchmarks/test_rnn_reset_backends_benchmark.py::test_rnn_rollout_with_intermediate_resets[b256-t128-i32-h512-cudnn-False-0-gru] 1.3535 1.3095 -3.25%
benchmarks/test_objectives_benchmarks.py::test_cql_speed[True-None] 82.88 85.52 +3.19%
benchmarks/test_replaybuffer_benchmark.py::test_rb_populate[TensorDictReplayBuffer-LazyTensorStorage-SamplerWithoutReplacement-400] 1,023 1,055 +3.15%
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[True-True-False-False-True] 38,486 37,281 -3.13%
benchmarks/test_replaybuffer_benchmark.py::test_rb_iterate[TensorDictPrioritizedReplayBuffer-LazyTensorStorage-None-10000] 2,138 2,072 -3.11%
benchmarks/test_objectives_benchmarks.py::test_iql_speed[reduce-overhead-None] 117.36 113.73 -3.09%
benchmarks/test_non_tensor_env_benchmark.py::test_non_tensor_env_rollout_speed[1000-serial-buffers-True] 0.5010 0.5161 +3.01%
benchmarks/test_objectives_benchmarks.py::test_cql_speed[False-backward] 28.39 27.55 -2.96%
benchmarks/test_objectives_benchmarks.py::test_sac_speed[True-None] 449.04 462.24 +2.94%
benchmarks/test_objectives_benchmarks.py::test_a2c_speed[True-backward] 116.08 119.45 +2.90%
benchmarks/test_objectives_benchmarks.py::test_redq_speed[False-backward] 55.42 53.88 -2.79%
benchmarks/test_replaybuffer_benchmark.py::test_rb_iterate[TensorDictReplayBuffer-LazyMemmapStorage-RandomSampler-10000] 2,855 2,778 -2.72%
benchmarks/test_objectives_benchmarks.py::test_values[td_lambda_return_estimate-True-False] 23.53 24.15 +2.64%
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[True-False-False-False-True] 34,465 33,568 -2.60%
benchmarks/test_storage_write_benchmark.py::TestStorageWriteBenchmark::test_collector_stack_then_write[200-img_shape3-large_batch] 137.47 140.99 +2.56%
benchmarks/test_envs_benchmark.py::test_serial 0.5676 0.5818 +2.49%
benchmarks/test_replaybuffer_benchmark.py::test_rb_sample[TensorDictPrioritizedReplayBuffer-ListStorage-None-4000] 160.65 164.65 +2.49%
benchmarks/test_objectives_benchmarks.py::test_a2c_speed[False-None] 177.19 172.78 -2.49%
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[False-True-True-False-True] 33,320 32,525 -2.39%
benchmarks/test_objectives_benchmarks.py::test_redq_deprec_speed[False-None] 86.17 88.19 +2.35%
benchmarks/test_objectives_benchmarks.py::test_ppo_speed[True-None] 257.73 251.80 -2.30%
benchmarks/test_objectives_benchmarks.py::test_redq_deprec_speed[False-backward] 59.86 61.22 +2.27%
benchmarks/test_objectives_benchmarks.py::test_ppo_speed[False-backward] 77.70 75.93 -2.27%
benchmarks/test_rnn_reset_backends_benchmark.py::test_rnn_rollout_with_intermediate_resets[b256-t128-i32-h512-scan-False-0-gru] 3.0650 2.9955 -2.27%
benchmarks/test_replaybuffer_benchmark.py::test_rb_extend_sample[ReplayBuffer-LazyTensorStorage-RandomSampler-100000-10000-100-False] 53.08 51.89 -2.23%
benchmarks/test_objectives_benchmarks.py::test_redq_deprec_speed[reduce-overhead-None] 281.92 275.68 -2.21%
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[True-True-True-False-False] 78,088 76,362 -2.21%
benchmarks/test_objectives_benchmarks.py::test_cql_speed[True-backward] 56.97 58.23 +2.21%
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[False-False-False-True-False] 27,320 26,718 -2.20%
benchmarks/test_collectors_benchmark.py::test_single_with_rb 8.4318 8.2498 -2.16%
benchmarks/test_objectives_benchmarks.py::test_ddpg_speed[True-None] 678.55 663.91 -2.16%
benchmarks/test_storage_write_benchmark.py::TestStorageWriteBenchmark::test_storage_write_lazystack[200-img_shape3-large_batch] 326.03 333.05 +2.15%
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[False-True-True-True-False] 34,514 33,788 -2.10%
benchmarks/test_replaybuffer_benchmark.py::test_rb_populate[TensorDictPrioritizedReplayBuffer-ListStorage-None-400] 191.76 187.74 -2.10%
benchmarks/test_replaybuffer_benchmark.py::TestPrioritizedReplayBufferBenchmark::test_sampler_sample_scale[1000000-cpu] 98.97 96.91 -2.08%
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[False-False-True-True-False] 28,085 27,505 -2.07%
benchmarks/test_non_tensor_env_benchmark.py::test_non_tensor_env_rollout_speed[1000-serial-no-buffers-True] 0.5929 0.5807 -2.06%
benchmarks/test_objectives_benchmarks.py::test_values[td1_return_estimate-False-False] 34.92 35.63 +2.02%
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[True-False-False-True-True] 20,061 19,657 -2.02%
benchmarks/test_envs_benchmark.py::test_parallel 0.9611 0.9419 -1.99%
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[True-True-True-True-True] 23,780 23,308 -1.98%
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[False-True-False-False-True] 30,867 30,276 -1.92%
benchmarks/test_objectives_benchmarks.py::test_ddpg_speed[False-None] 338.83 345.24 +1.89%
benchmarks/test_objectives_benchmarks.py::test_reinforce_speed[False-backward] 128.43 130.85 +1.88%
benchmarks/test_objectives_benchmarks.py::test_ddpg_speed[False-backward] 241.85 237.34 -1.86%
benchmarks/test_objectives_benchmarks.py::test_values[generalized_advantage_estimate-True-True] 93.56 95.23 +1.79%
benchmarks/test_replaybuffer_benchmark.py::test_rb_populate[TensorDictReplayBuffer-LazyMemmapStorage-RandomSampler-400] 537.07 546.36 +1.73%
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[False-True-False-True-True] 19,506 19,840 +1.71%
benchmarks/test_objectives_benchmarks.py::test_gae_speed[vec_generalized_advantage_estimate-True-32-512] 33.44 34.01 +1.70%
benchmarks/test_objectives_benchmarks.py::test_redq_speed[reduce-overhead-None] 222.83 226.56 +1.67%
benchmarks/test_objectives_benchmarks.py::test_dqn_speed[False-None] 685.31 696.37 +1.61%
benchmarks/test_envs_benchmark.py::test_simple 1.7626 1.7910 +1.61%
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[False-True-True-False-False] 57,487 56,585 -1.57%
benchmarks/test_storage_write_benchmark.py::TestStorageWriteBenchmark::test_collector_stack_then_write[100-img_shape2-large_img] 168.75 171.37 +1.55%
benchmarks/test_objectives_benchmarks.py::test_dqn_speed[False-backward] 507.11 514.93 +1.54%
benchmarks/test_replaybuffer_benchmark.py::test_rb_iterate[TensorDictReplayBuffer-ListStorage-SamplerWithoutReplacement-4000] 163.43 165.91 +1.52%
benchmarks/test_objectives_benchmarks.py::test_a2c_speed[False-backward] 81.91 80.67 -1.51%
benchmarks/test_storage_write_benchmark.py::TestCollectorIntegrationBenchmark::test_collector_without_rb[100-img_shape0-atari] 29.70 30.15 +1.49%
benchmarks/test_objectives_benchmarks.py::test_sac_speed[True-backward] 248.05 251.72 +1.48%
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[True-False-True-False-False] 64,235 63,296 -1.46%
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[True-True-False-True-True] 21,980 21,665 -1.43%
... ... ... Showing 120 of 192 comparisons, sorted by absolute change.

GPU

Compared 202 benchmarks. Regressions over 5%: 19. Improvements over 5%: 5.

Benchmark main ops PR ops Change
benchmarks/test_replaybuffer_benchmark.py::test_rb_populate[TensorDictReplayBuffer-ListStorage-SamplerWithoutReplacement-400] 43.17 191.02 +342.45%
benchmarks/test_replaybuffer_benchmark.py::test_rb_populate[TensorDictReplayBuffer-ListStorage-RandomSampler-400] 194.84 49.00 -74.85%
benchmarks/test_replaybuffer_benchmark.py::test_rb_sample[TensorDictReplayBuffer-LazyMemmapStorage-RandomSampler-10000] 2,628 1,157 -55.98%
benchmarks/test_replaybuffer_benchmark.py::test_rb_sample[TensorDictReplayBuffer-LazyTensorStorage-SamplerWithoutReplacement-10000] 3,136 3,698 +17.93%
benchmarks/test_replaybuffer_benchmark.py::test_rb_sample[TensorDictReplayBuffer-LazyTensorStorage-RandomSampler-10000] 3,116 3,665 +17.62%
benchmarks/test_replaybuffer_benchmark.py::test_rb_iterate[TensorDictPrioritizedReplayBuffer-LazyMemmapStorage-None-10000] 2,106 1,828 -13.20%
benchmarks/test_replaybuffer_benchmark.py::test_rb_populate[TensorDictPrioritizedReplayBuffer-LazyTensorStorage-None-400] 831.61 730.20 -12.19%
benchmarks/test_replaybuffer_benchmark.py::test_rb_sample[TensorDictPrioritizedReplayBuffer-LazyMemmapStorage-None-10000] 2,081 1,847 -11.27%
benchmarks/test_replaybuffer_benchmark.py::test_rb_populate[TensorDictReplayBuffer-LazyMemmapStorage-SamplerWithoutReplacement-400] 520.86 463.01 -11.11%
benchmarks/test_collectors_benchmark.py::test_single_with_rb_pixels 5.3276 4.7436 -10.96%
benchmarks/test_objectives_benchmarks.py::test_gae_speed[generalized_advantage_estimate-False-1-512] 48.55 44.16 -9.04%
benchmarks/test_compressed_storage_benchmark.py::TestCompressedStorageBenchmark::test_tensor_to_bytestream_speed[untyped_storage] 8.6719 7.9159 -8.72%
benchmarks/test_objectives_benchmarks.py::test_values[td1_return_estimate-False-False] 20.40 18.66 -8.55%
benchmarks/test_replaybuffer_benchmark.py::test_rb_sample[TensorDictPrioritizedReplayBuffer-LazyTensorStorage-None-10000] 2,112 1,933 -8.48%
benchmarks/test_replaybuffer_benchmark.py::test_rb_iterate[TensorDictPrioritizedReplayBuffer-LazyTensorStorage-None-10000] 2,223 2,042 -8.17%
benchmarks/test_objectives_benchmarks.py::test_values[td_lambda_return_estimate-True-False] 12.24 11.27 -7.94%
benchmarks/test_objectives_benchmarks.py::test_values[generalized_advantage_estimate-True-True] 48.62 44.94 -7.58%
benchmarks/test_objectives_benchmarks.py::test_dqn_speed[False-None] 643.38 603.61 -6.18%
benchmarks/test_compressed_storage_benchmark.py::TestCompressedStorageBenchmark::test_tensor_to_bytestream_speed[pickle] 12,349 11,634 -5.79%
benchmarks/test_envs_benchmark.py::test_cat_frames_functional[4-constant] 4,854 4,580 -5.65%
benchmarks/test_objectives_benchmarks.py::test_values[vec_generalized_advantage_estimate-True-True] 312.71 295.20 -5.60%
benchmarks/test_storage_write_benchmark.py::TestStorageWriteBenchmark::test_storage_write_lazystack[100-img_shape2-large_img] 399.98 422.06 +5.52%
benchmarks/test_storage_write_benchmark.py::TestStorageWriteBenchmark::test_collector_lazystack_then_write[100-img_shape2-large_img] 383.84 404.63 +5.42%
benchmarks/test_replaybuffer_benchmark.py::test_rb_iterate[TensorDictReplayBuffer-LazyTensorStorage-SamplerWithoutReplacement-10000] 3,678 3,482 -5.33%
benchmarks/test_objectives_benchmarks.py::test_gae_speed[vec_generalized_advantage_estimate-False-1-512] 1,351 1,286 -4.86%
benchmarks/test_objectives_benchmarks.py::test_gae_speed[vec_generalized_advantage_estimate-False-32-512] 1,319 1,258 -4.63%
benchmarks/test_objectives_benchmarks.py::test_sac_speed[True-backward] 313.45 327.89 +4.61%
benchmarks/test_envs_benchmark.py::test_simple 1.2489 1.1933 -4.45%
benchmarks/test_replaybuffer_benchmark.py::test_rb_populate[TensorDictPrioritizedReplayBuffer-LazyMemmapStorage-None-400] 457.28 477.21 +4.36%
benchmarks/test_envs_benchmark.py::test_cat_frames_functional[16-same] 5.2679 5.4878 +4.17%
benchmarks/test_replaybuffer_benchmark.py::TestPrioritizedReplayBufferBenchmark::test_sample_mixed_devices[1000000-cuda_storage_cpu_sampler] 87.12 90.62 +4.02%
benchmarks/test_replaybuffer_benchmark.py::test_rb_iterate[TensorDictReplayBuffer-LazyMemmapStorage-SamplerWithoutReplacement-10000] 3,229 3,102 -3.93%
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[False-False-True-False-True] 30,676 29,487 -3.87%
benchmarks/test_objectives_benchmarks.py::test_dqn_speed[True-None] 1,897 1,825 -3.81%
benchmarks/test_objectives_benchmarks.py::test_reinforce_speed[True-backward] 363.03 349.48 -3.73%
benchmarks/test_replaybuffer_benchmark.py::test_rb_populate[TensorDictReplayBuffer-LazyTensorStorage-SamplerWithoutReplacement-400] 997.66 960.90 -3.68%
benchmarks/test_objectives_benchmarks.py::test_dqn_speed[True-backward] 983.91 948.22 -3.63%
benchmarks/test_objectives_benchmarks.py::test_a2c_speed[True-None] 741.42 715.81 -3.45%
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[False-False-True-False-False] 49,038 47,534 -3.07%
benchmarks/test_storage_write_benchmark.py::TestStorageWriteBenchmark::test_storage_write_contiguous[200-img_shape3-large_batch] 741.60 718.94 -3.06%
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[True-False-False-False-True] 34,037 33,005 -3.03%
benchmarks/test_replaybuffer_benchmark.py::test_rb_populate[TensorDictPrioritizedReplayBuffer-ListStorage-None-400] 186.81 181.25 -2.97%
benchmarks/test_objectives_benchmarks.py::test_a2c_speed[False-None] 266.70 274.56 +2.95%
benchmarks/test_compressed_storage_benchmark.py::TestCompressedStorageBenchmark::test_tensor_to_bytestream_speed[numpy] 346,549 356,525 +2.88%
benchmarks/test_objectives_benchmarks.py::test_reinforce_speed[False-None] 380.91 391.46 +2.77%
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[True-False-False-False-False] 54,560 53,074 -2.72%
benchmarks/test_envs_benchmark.py::test_transformed 0.7140 0.6947 -2.70%
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[True-True-True-False-True] 41,706 40,617 -2.61%
benchmarks/test_objectives_benchmarks.py::test_ppo_speed[reduce-overhead-None] 788.02 808.21 +2.56%
benchmarks/test_objectives_benchmarks.py::test_dqn_speed[False-backward] 453.07 442.16 -2.41%
benchmarks/test_replaybuffer_benchmark.py::test_rb_iterate[TensorDictReplayBuffer-LazyTensorStorage-RandomSampler-10000] 3,101 3,175 +2.38%
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[True-True-False-False-True] 37,756 36,866 -2.36%
benchmarks/test_objectives_benchmarks.py::test_td3_speed[True-None] 712.88 696.08 -2.36%
benchmarks/test_rnn_reset_backends_benchmark.py::test_rnn_rollout_with_intermediate_resets[b256-t128-i32-h512-scan-True-0-lstm] 75.94 74.19 -2.32%
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[False-True-False-False-False] 50,028 48,882 -2.29%
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[True-True-False-True-True] 21,884 21,398 -2.22%
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[True-True-True-False-False] 75,900 74,228 -2.20%
benchmarks/test_objectives_benchmarks.py::test_ddpg_speed[True-None] 798.43 816.00 +2.20%
benchmarks/test_storage_write_benchmark.py::TestStorageWriteBenchmark::test_storage_write_contiguous[100-img_shape1-atari] 4,333 4,240 -2.15%
benchmarks/test_storage_write_benchmark.py::TestStorageWriteBenchmark::test_storage_write_lazystack[200-img_shape3-large_batch] 320.51 327.35 +2.13%
benchmarks/test_storage_write_benchmark.py::TestStorageWriteBenchmark::test_collector_stack_then_write[200-img_shape3-large_batch] 133.22 135.77 +1.91%
benchmarks/test_replaybuffer_benchmark.py::test_rb_populate[TensorDictReplayBuffer-LazyMemmapStorage-RandomSampler-400] 477.18 468.17 -1.89%
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[False-False-False-False-True] 28,003 27,476 -1.88%
benchmarks/test_storage_write_benchmark.py::TestStorageWriteBenchmark::test_collector_lazystack_then_write[200-img_shape3-large_batch] 299.52 305.07 +1.85%
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[False-False-True-True-True] 18,736 18,399 -1.80%
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[True-True-False-True-False] 36,844 37,505 +1.79%
benchmarks/test_objectives_benchmarks.py::test_dqn_speed[reduce-overhead-None] 1,895 1,862 -1.76%
benchmarks/test_storage_write_benchmark.py::TestCollectorIntegrationBenchmark::test_collector_with_rb[100-img_shape0-atari] 26.16 25.71 -1.74%
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[True-False-True-True-False] 34,026 34,617 +1.74%
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[False-True-True-False-False] 56,183 55,221 -1.71%
benchmarks/test_envs_benchmark.py::test_cat_frames_functional[4-same] 6.6784 6.5642 -1.71%
benchmarks/test_replaybuffer_benchmark.py::test_rb_sample[TensorDictReplayBuffer-ListStorage-RandomSampler-4000] 160.98 163.73 +1.71%
benchmarks/test_storage_write_benchmark.py::TestStorageWriteBenchmark::test_storage_write_lazystack[100-img_shape1-atari] 698.41 686.65 -1.68%
benchmarks/test_objectives_benchmarks.py::test_cql_speed[False-backward] 40.73 40.05 -1.66%
benchmarks/test_storage_write_benchmark.py::TestCollectorIntegrationBenchmark::test_collector_without_rb_cuda[200-img_shape1-large_batch] 8.8060 8.6622 -1.63%
benchmarks/test_objectives_benchmarks.py::test_redq_deprec_speed[False-backward] 70.22 71.36 +1.63%
benchmarks/test_objectives_benchmarks.py::test_td3_speed[False-backward] 84.16 82.81 -1.60%
benchmarks/test_objectives_benchmarks.py::test_ppo_speed[True-backward] 344.76 350.28 +1.60%
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[True-False-True-False-False] 63,598 62,588 -1.59%
benchmarks/test_storage_write_benchmark.py::TestStorageWriteBenchmark::test_storage_write_lazystack[50-img_shape0-small] 4,296 4,363 +1.58%
benchmarks/test_objectives_benchmarks.py::test_reinforce_speed[reduce-overhead-None] 123.17 121.24 -1.57%
benchmarks/test_non_tensor_env_benchmark.py::test_non_tensor_env_rollout_speed[1000-parallel-buffers-True] 0.5279 0.5196 -1.57%
benchmarks/test_objectives_benchmarks.py::test_reinforce_speed[True-None] 762.99 774.94 +1.57%
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[False-False-False-True-False] 26,889 27,298 +1.52%
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[True-True-True-True-True] 23,486 23,145 -1.45%
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[False-True-False-True-False] 30,797 31,214 +1.35%
benchmarks/test_storage_write_benchmark.py::TestCollectorIntegrationBenchmark::test_collector_with_rb_cuda[100-img_shape0-atari] 16.82 16.60 -1.32%
benchmarks/test_storage_write_benchmark.py::TestStorageWriteBenchmark::test_collector_stack_then_write[100-img_shape2-large_img] 168.53 170.73 +1.31%
benchmarks/test_replaybuffer_benchmark.py::test_rb_extend_sample[ReplayBuffer-LazyTensorStorage-RandomSampler-100000-10000-100-True] 22.56 22.85 +1.29%
benchmarks/test_non_tensor_env_benchmark.py::test_non_tensor_env_rollout_speed[1000-serial-no-buffers-False] 0.6656 0.6741 +1.28%
benchmarks/test_replaybuffer_benchmark.py::test_rb_iterate[TensorDictPrioritizedReplayBuffer-ListStorage-None-4000] 163.05 165.12 +1.27%
benchmarks/test_storage_write_benchmark.py::TestCollectorIntegrationBenchmark::test_collector_without_rb[200-img_shape1-large_batch] 15.28 15.09 -1.26%
benchmarks/test_objectives_benchmarks.py::test_a2c_speed[True-backward] 362.06 366.63 +1.26%
benchmarks/test_objectives_benchmarks.py::test_gae_speed[vec_generalized_advantage_estimate-True-32-512] 654.67 646.59 -1.23%
benchmarks/test_storage_write_benchmark.py::TestCollectorIntegrationBenchmark::test_collector_with_rb[200-img_shape1-large_batch] 13.32 13.16 -1.23%
benchmarks/test_storage_write_benchmark.py::TestStorageWriteBenchmark::test_storage_write_contiguous[100-img_shape2-large_img] 557.25 550.78 -1.16%
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[False-True-False-True-True] 19,206 19,428 +1.16%
benchmarks/test_storage_write_benchmark.py::TestStorageWriteBenchmark::test_collector_lazystack_then_write[100-img_shape1-atari] 633.13 640.29 +1.13%
benchmarks/test_objectives_benchmarks.py::test_redq_deprec_speed[True-None] 417.92 413.34 -1.10%
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[False-True-False-False-True] 29,827 29,508 -1.07%
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[True-True-False-False-False] 63,019 62,406 -0.97%
benchmarks/test_rnn_reset_backends_benchmark.py::test_rnn_rollout_with_intermediate_resets[b256-t128-i32-h512-scan-True-0-gru] 48.45 48.91 +0.96%
benchmarks/test_objectives_benchmarks.py::test_sac_speed[True-None] 614.25 608.40 -0.95%
benchmarks/test_replaybuffer_benchmark.py::TestPrioritizedReplayBufferBenchmark::test_sample_mixed_devices[1000000-cuda_storage_cuda_samp... 1,498 1,484 -0.95%
benchmarks/test_replaybuffer_benchmark.py::test_rb_iterate[TensorDictReplayBuffer-LazyMemmapStorage-RandomSampler-10000] 2,941 2,968 +0.93%
benchmarks/test_objectives_benchmarks.py::test_redq_deprec_speed[True-backward] 268.33 270.69 +0.88%
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[True-False-True-False-True] 36,686 36,364 -0.88%
benchmarks/test_storage_write_benchmark.py::TestCollectorIntegrationBenchmark::test_collector_without_rb_cuda[100-img_shape0-atari] 17.64 17.49 -0.87%
benchmarks/test_compressed_storage_benchmark.py::TestCompressedStorageBenchmark::test_tensor_to_bytestream_speed[safetensors] 23,575 23,779 +0.87%
benchmarks/test_objectives_benchmarks.py::test_iql_speed[True-backward] 237.54 239.59 +0.86%
benchmarks/test_objectives_benchmarks.py::test_cql_speed[False-None] 53.75 53.29 -0.85%
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[False-False-False-True-True] 17,862 17,710 -0.85%
benchmarks/test_objectives_benchmarks.py::test_reinforce_speed[False-backward] 266.33 268.59 +0.85%
benchmarks/test_storage_write_benchmark.py::TestStorageWriteBenchmark::test_collector_lazystack_then_write[50-img_shape0-small] 3,460 3,431 -0.85%
benchmarks/test_non_tensor_env_benchmark.py::test_non_tensor_env_rollout_speed[1000-single-False] 1.5800 1.5667 -0.84%
benchmarks/test_objectives_benchmarks.py::test_iql_speed[reduce-overhead-None] 103.81 102.94 -0.84%
benchmarks/test_objectives_benchmarks.py::test_redq_deprec_speed[False-None] 97.91 97.10 -0.83%
benchmarks/test_replaybuffer_benchmark.py::test_rb_sample[TensorDictReplayBuffer-LazyTensorStorage-sampler7-10000] 839.16 832.23 -0.83%
benchmarks/test_objectives_benchmarks.py::test_cql_speed[True-None] 365.06 367.97 +0.80%
benchmarks/test_non_tensor_env_benchmark.py::test_non_tensor_env_rollout_speed[1000-serial-no-buffers-True] 0.5806 0.5851 +0.78%
... ... ... Showing 120 of 202 comparisons, sorted by absolute change.

@vmoens
Update
e07c236 
[ghstack-poisoned]
vmoens added a commit that referenced this pull request Jun 17, 2026
@vmoens
[Doc] Clarify headless MuJoCo EGL dependencies
5572c83 
ghstack-source-id: c149816
Pull-Request: #3879
@vmoens
Update
1f21b6c 
[ghstack-poisoned]
vmoens added a commit that referenced this pull request Jun 18, 2026
@vmoens
[Doc] Clarify headless MuJoCo EGL dependencies
7bda608 
ghstack-source-id: f424e96
Pull-Request: #3879
@vmoens
Update
8cb7990 
[ghstack-poisoned]
vmoens added a commit that referenced this pull request Jun 18, 2026
@vmoens
[Doc] Clarify headless MuJoCo EGL dependencies
60fc32c 
ghstack-source-id: 1a8268b
Pull-Request: #3879
@vmoens
Update
b6ea786 
[ghstack-poisoned]
vmoens added a commit that referenced this pull request Jun 18, 2026
@vmoens
[Doc] Clarify headless MuJoCo EGL dependencies
d063fa2 
ghstack-source-id: 4a45109
Pull-Request: #3879
@vmoens
Update
2679d06 
[ghstack-poisoned]
vmoens added a commit that referenced this pull request Jun 18, 2026
@vmoens
[Doc] Clarify headless MuJoCo EGL dependencies
8521193 
ghstack-source-id: 7034b04
Pull-Request: #3879
@vmoens
Update
0586029 
[ghstack-poisoned]
vmoens added a commit that referenced this pull request Jun 18, 2026
@vmoens
[Doc] Clarify headless MuJoCo EGL dependencies
f3b1454 
ghstack-source-id: 1464d2d
Pull-Request: #3879
@vmoens
Update
7cfaa93 
[ghstack-poisoned]
vmoens added a commit that referenced this pull request Jun 18, 2026
@vmoens
[Doc] Clarify headless MuJoCo EGL dependencies
165e4e9 
ghstack-source-id: 595ab37
Pull-Request: #3879
@vmoens
Update
f99885a 
[ghstack-poisoned]
vmoens added a commit that referenced this pull request Jun 18, 2026
@vmoens
[Doc] Clarify headless MuJoCo EGL dependencies
912f79e 
ghstack-source-id: 27bd822
Pull-Request: #3879
vmoens added a commit that referenced this pull request Jun 18, 2026
@vmoens
[Doc] Clarify headless MuJoCo EGL dependencies
75b8167 
ghstack-source-id: 27bd822
Pull-Request: #3879
@vmoens
Update
4af5335 
[ghstack-poisoned]
vmoens added a commit that referenced this pull request Jun 19, 2026
@vmoens
[Doc] Clarify headless MuJoCo EGL dependencies
0f72cd5 
ghstack-source-id: 1c7e0d4
Pull-Request: #3879
@vmoens
Update
284223e 
[ghstack-poisoned]
vmoens added a commit that referenced this pull request Jun 19, 2026
@vmoens
[Doc] Clarify headless MuJoCo EGL dependencies
ef5861a 
ghstack-source-id: 6f802b7
Pull-Request: #3879
vmoens added a commit that referenced this pull request Jun 19, 2026
@vmoens
[Doc] Clarify headless MuJoCo EGL dependencies
ec8d72a 
ghstack-source-id: 6f802b7
Pull-Request: #3879
@vmoens vmoens mentioned this pull request Jun 19, 2026
Merged
vmoens added a commit that referenced this pull request Jun 19, 2026
@vmoens
[Doc] Clarify headless MuJoCo EGL dependencies
92a3f7f 
ghstack-source-id: 6f802b7
Pull-Request: #3879
@vmoens vmoens merged commit 284223e into gh/vmoens/291/base Jun 19, 2026
109 of 112 checks passed
@vmoens vmoens deleted the gh/vmoens/291/head branch June 19, 2026 20:34
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

CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed. Documentation Improvements or additions to documentation

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

1 participant

@vmoens