[Feature] Add max-inflight guard for remote policy clients#3897
Draft
vmoens wants to merge 1 commit into
Draft
Conversation
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/rl/3897
Note: Links to docs will display an error until the docs builds have been completed. ❗ 1 Active SEVsThere are 1 currently active SEVs. If your PR is affected, please view them below: ❌ 12 New Failures, 1 Cancelled JobAs of commit bf71148 with merge base d7ef78b ( NEW FAILURES - The following jobs have failed:
CANCELLED JOB - The following job was cancelled. Please retry:
This comment was automatically generated by Dr. CI and updates every 15 minutes. |
This was referenced Jun 21, 2026
github-actions
Bot
added
Feature
meta-cla
Bot
added
the
CLA Signed
Contributor
Benchmark Results: PR
|
| Benchmark | main ops | PR ops | Change |
|---|---|---|---|
benchmarks/test_replaybuffer_benchmark.py::test_rb_populate[TensorDictReplayBuffer-ListStorage-RandomSampler-400] |
36.61 | 201.26 | +449.72% |
benchmarks/test_replaybuffer_benchmark.py::test_rb_iterate[TensorDictPrioritizedReplayBuffer-LazyMemmapStorage-None-10000] |
1,981 | 461.13 | -76.73% |
benchmarks/test_replaybuffer_benchmark.py::test_rb_populate[TensorDictPrioritizedReplayBuffer-ListStorage-None-400] |
192.81 | 56.76 | -70.56% |
benchmarks/test_objectives_benchmarks.py::test_ppo_speed[True-backward] |
91.62 | 115.08 | +25.61% |
benchmarks/test_replaybuffer_benchmark.py::test_rb_sample[TensorDictReplayBuffer-LazyTensorStorage-SamplerWithoutReplacement-10000] |
2,971 | 3,659 | +23.17% |
benchmarks/test_objectives_benchmarks.py::test_dqn_speed[True-backward] |
826.24 | 987.87 | +19.56% |
benchmarks/test_replaybuffer_benchmark.py::test_rb_iterate[TensorDictPrioritizedReplayBuffer-LazyTensorStorage-None-10000] |
1,980 | 2,357 | +19.02% |
benchmarks/test_replaybuffer_benchmark.py::test_rb_sample[TensorDictReplayBuffer-LazyMemmapStorage-SamplerWithoutReplacement-10000] |
2,909 | 3,430 | +17.90% |
benchmarks/test_replaybuffer_benchmark.py::test_rb_iterate[TensorDictReplayBuffer-LazyTensorStorage-RandomSampler-10000] |
3,679 | 3,077 | -16.38% |
benchmarks/test_replaybuffer_benchmark.py::test_rb_populate[TensorDictPrioritizedReplayBuffer-LazyMemmapStorage-None-400] |
435.14 | 504.05 | +15.84% |
benchmarks/test_replaybuffer_benchmark.py::test_rb_sample[TensorDictReplayBuffer-LazyTensorStorage-sampler7-10000] |
722.72 | 827.73 | +14.53% |
benchmarks/test_objectives_benchmarks.py::test_gae_speed[vec_generalized_advantage_estimate-True-32-512] |
33.26 | 28.58 | -14.08% |
benchmarks/test_replaybuffer_benchmark.py::test_rb_sample[TensorDictReplayBuffer-LazyMemmapStorage-sampler6-10000] |
678.58 | 765.12 | +12.75% |
benchmarks/test_replaybuffer_benchmark.py::test_rb_iterate[TensorDictReplayBuffer-LazyMemmapStorage-RandomSampler-10000] |
3,346 | 2,941 | -12.11% |
benchmarks/test_objectives_benchmarks.py::test_ddpg_speed[True-backward] |
377.42 | 419.18 | +11.06% |
benchmarks/test_replaybuffer_benchmark.py::test_rb_iterate[TensorDictReplayBuffer-LazyMemmapStorage-SamplerWithoutReplacement-10000] |
3,116 | 2,843 | -8.76% |
benchmarks/test_objectives_benchmarks.py::test_sac_speed[True-backward] |
227.55 | 247.29 | +8.67% |
benchmarks/test_envs_benchmark.py::test_simple |
1.7155 | 1.8394 | +7.23% |
benchmarks/test_replaybuffer_benchmark.py::test_rb_populate[TensorDictReplayBuffer-LazyTensorStorage-RandomSampler-400] |
1,108 | 1,032 | -6.86% |
benchmarks/test_envs_benchmark.py::test_parallel |
0.9915 | 0.9312 | -6.08% |
benchmarks/test_storage_write_benchmark.py::TestStorageWriteBenchmark::test_storage_write_lazystack[100-img_shape2-large_img] |
395.91 | 419.98 | +6.08% |
benchmarks/test_replaybuffer_benchmark.py::test_rb_sample[TensorDictPrioritizedReplayBuffer-LazyTensorStorage-None-10000] |
2,251 | 2,382 | +5.84% |
benchmarks/test_objectives_benchmarks.py::test_ppo_speed[reduce-overhead-None] |
254.66 | 268.84 | +5.57% |
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[True-False-True-True-True] |
21,856 | 20,642 | -5.56% |
benchmarks/test_compressed_storage_benchmark.py::TestCompressedStorageBenchmark::test_tensor_to_bytestream_speed[safetensors] |
23,335 | 24,589 | +5.38% |
benchmarks/test_objectives_benchmarks.py::test_redq_speed[reduce-overhead-None] |
217.63 | 229.08 | +5.26% |
benchmarks/test_storage_write_benchmark.py::TestStorageWriteBenchmark::test_collector_lazystack_then_write[100-img_shape2-large_img] |
385.97 | 405.45 | +5.05% |
benchmarks/test_non_tensor_env_benchmark.py::test_non_tensor_env_rollout_speed[1000-single-True] |
1.4067 | 1.3437 | -4.48% |
benchmarks/test_replaybuffer_benchmark.py::test_rb_sample[TensorDictPrioritizedReplayBuffer-LazyMemmapStorage-None-10000] |
2,189 | 2,285 | +4.40% |
benchmarks/test_compressed_storage_benchmark.py::TestCompressedStorageBenchmark::test_tensor_to_bytestream_speed[pickle] |
12,032 | 12,554 | +4.34% |
benchmarks/test_replaybuffer_benchmark.py::test_rb_iterate[TensorDictReplayBuffer-LazyTensorStorage-SamplerWithoutReplacement-10000] |
3,247 | 3,108 | -4.30% |
benchmarks/test_replaybuffer_benchmark.py::test_rb_extend_sample[ReplayBuffer-LazyTensorStorage-RandomSampler-10000-10000-100-True] |
25.08 | 26.16 | +4.30% |
benchmarks/test_storage_write_benchmark.py::TestStorageWriteBenchmark::test_collector_stack_then_write[100-img_shape1-atari] |
273.71 | 284.73 | +4.02% |
benchmarks/test_envs_benchmark.py::test_transformed |
0.8961 | 0.9313 | +3.93% |
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[True-False-False-False-True] |
36,053 | 34,657 | -3.87% |
benchmarks/test_objectives_benchmarks.py::test_dqn_speed[reduce-overhead-None] |
1,800 | 1,869 | +3.82% |
benchmarks/test_rnn_reset_backends_benchmark.py::test_rnn_rollout_with_intermediate_resets[b256-t128-i32-h512-cudnn-False-0-gru] |
1.3170 | 1.3654 | +3.68% |
benchmarks/test_objectives_benchmarks.py::test_values[td1_return_estimate-False-False] |
37.57 | 36.20 | -3.65% |
benchmarks/test_objectives_benchmarks.py::test_reinforce_speed[False-backward] |
129.66 | 134.36 | +3.63% |
benchmarks/test_objectives_benchmarks.py::test_td3_speed[False-backward] |
92.75 | 89.58 | -3.42% |
benchmarks/test_replaybuffer_benchmark.py::test_rb_sample[TensorDictReplayBuffer-LazyTensorStorage-RandomSampler-10000] |
3,063 | 2,960 | -3.37% |
benchmarks/test_replaybuffer_benchmark.py::test_rb_sample[TensorDictReplayBuffer-LazyMemmapStorage-RandomSampler-10000] |
2,945 | 2,849 | -3.27% |
benchmarks/test_collectors_benchmark.py::test_sync |
16.65 | 17.18 | +3.21% |
benchmarks/test_envs_benchmark.py::test_serial |
0.5737 | 0.5920 | +3.17% |
benchmarks/test_replaybuffer_benchmark.py::test_rb_populate[TensorDictReplayBuffer-LazyMemmapStorage-RandomSampler-400] |
541.40 | 558.53 | +3.16% |
benchmarks/test_collectors_benchmark.py::test_single |
8.8167 | 9.0936 | +3.14% |
benchmarks/test_replaybuffer_benchmark.py::TestPrioritizedReplayBufferBenchmark::test_sample_mixed_devices[1000000-memmap_cpu_storage_cpu... |
80.84 | 83.29 | +3.04% |
benchmarks/test_storage_write_benchmark.py::TestStorageWriteBenchmark::test_storage_write_contiguous[200-img_shape3-large_batch] |
752.06 | 774.58 | +3.00% |
benchmarks/test_objectives_benchmarks.py::test_reinforce_speed[False-None] |
207.67 | 213.60 | +2.86% |
benchmarks/test_compressed_storage_benchmark.py::TestCompressedStorageBenchmark::test_tensor_to_bytestream_speed[untyped_storage] |
8.9122 | 8.6582 | -2.85% |
benchmarks/test_storage_write_benchmark.py::TestStorageWriteBenchmark::test_collector_stack_then_write[200-img_shape3-large_batch] |
139.60 | 143.54 | +2.83% |
benchmarks/test_objectives_benchmarks.py::test_redq_speed[True-None] |
224.99 | 231.25 | +2.78% |
benchmarks/test_objectives_benchmarks.py::test_a2c_speed[False-backward] |
84.28 | 81.94 | -2.77% |
benchmarks/test_rnn_reset_backends_benchmark.py::test_rnn_rollout_with_intermediate_resets[b256-t128-i32-h512-cudnn-False-0-lstm] |
0.8517 | 0.8751 | +2.75% |
benchmarks/test_storage_write_benchmark.py::TestStorageWriteBenchmark::test_storage_write_lazystack[100-img_shape1-atari] |
675.88 | 694.35 | +2.73% |
benchmarks/test_replaybuffer_benchmark.py::test_rb_extend_sample[ReplayBuffer-LazyTensorStorage-RandomSampler-1000000-10000-100-True] |
23.51 | 24.15 | +2.73% |
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[True-True-False-False-True] |
39,964 | 38,887 | -2.69% |
benchmarks/test_storage_write_benchmark.py::TestStorageWriteBenchmark::test_collector_lazystack_then_write[200-img_shape3-large_batch] |
308.34 | 316.54 | +2.66% |
benchmarks/test_replaybuffer_benchmark.py::test_rb_populate[TensorDictReplayBuffer-ListStorage-SamplerWithoutReplacement-400] |
197.08 | 202.29 | +2.64% |
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[True-True-True-True-False] |
43,192 | 44,334 | +2.64% |
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[False-True-True-False-False] |
57,453 | 58,962 | +2.63% |
benchmarks/test_rnn_reset_backends_benchmark.py::test_rnn_rollout_with_intermediate_resets[b256-t128-i32-h512-scan-False-0-lstm] |
2.0105 | 2.0630 | +2.61% |
benchmarks/test_objectives_benchmarks.py::test_values[vec_td_lambda_return_estimate-True-False] |
55.75 | 54.30 | -2.60% |
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[True-True-True-True-True] |
24,544 | 23,921 | -2.54% |
benchmarks/test_storage_write_benchmark.py::TestStorageWriteBenchmark::test_collector_lazystack_then_write[50-img_shape0-small] |
3,555 | 3,469 | -2.41% |
benchmarks/test_objectives_benchmarks.py::test_redq_deprec_speed[True-backward] |
133.86 | 137.08 | +2.40% |
benchmarks/test_storage_write_benchmark.py::TestStorageWriteBenchmark::test_collector_lazystack_then_write[100-img_shape1-atari] |
635.95 | 651.21 | +2.40% |
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[True-True-False-True-True] |
23,211 | 22,677 | -2.30% |
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[True-False-False-False-False] |
57,318 | 56,016 | -2.27% |
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[False-True-False-True-True] |
19,982 | 20,432 | +2.25% |
benchmarks/test_objectives_benchmarks.py::test_sac_speed[True-None] |
469.00 | 478.96 | +2.12% |
benchmarks/test_objectives_benchmarks.py::test_redq_deprec_speed[False-backward] |
63.44 | 62.10 | -2.12% |
benchmarks/test_storage_write_benchmark.py::TestStorageWriteBenchmark::test_collector_stack_then_write[100-img_shape2-large_img] |
170.36 | 173.95 | +2.11% |
benchmarks/test_objectives_benchmarks.py::test_cql_speed[reduce-overhead-None] |
84.41 | 82.65 | -2.08% |
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[True-True-True-False-False] |
77,909 | 79,496 | +2.04% |
benchmarks/test_storage_write_benchmark.py::TestStorageWriteBenchmark::test_storage_write_contiguous[50-img_shape0-small] |
7,362 | 7,214 | -2.02% |
benchmarks/test_non_tensor_env_benchmark.py::test_non_tensor_env_rollout_speed[1000-parallel-no-buffers-True] |
0.2154 | 0.2197 | +2.00% |
benchmarks/test_objectives_benchmarks.py::test_values[vec_generalized_advantage_estimate-True-True] |
55.37 | 54.26 | -2.00% |
benchmarks/test_objectives_benchmarks.py::test_reinforce_speed[True-backward] |
125.29 | 127.77 | +1.98% |
benchmarks/test_objectives_benchmarks.py::test_redq_deprec_speed[False-None] |
90.88 | 89.09 | -1.98% |
benchmarks/test_objectives_benchmarks.py::test_ddpg_speed[reduce-overhead-None] |
696.91 | 710.44 | +1.94% |
benchmarks/test_objectives_benchmarks.py::test_redq_deprec_speed[reduce-overhead-None] |
280.56 | 285.85 | +1.88% |
benchmarks/test_objectives_benchmarks.py::test_td3_speed[False-None] |
124.48 | 122.14 | -1.88% |
benchmarks/test_storage_write_benchmark.py::TestCollectorIntegrationBenchmark::test_collector_without_rb[100-img_shape0-atari] |
30.23 | 30.77 | +1.80% |
benchmarks/test_objectives_benchmarks.py::test_iql_speed[True-backward] |
57.96 | 59.00 | +1.79% |
benchmarks/test_replaybuffer_benchmark.py::test_rb_extend_sample[ReplayBuffer-LazyTensorStorage-RandomSampler-100000-10000-100-False] |
52.95 | 53.90 | +1.79% |
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[False-False-True-True-False] |
30,015 | 30,550 | +1.78% |
benchmarks/test_objectives_benchmarks.py::test_values[td_lambda_return_estimate-True-False] |
25.40 | 24.95 | -1.77% |
benchmarks/test_objectives_benchmarks.py::test_values[vec_td1_return_estimate-False-False] |
55.23 | 54.26 | -1.76% |
benchmarks/test_objectives_benchmarks.py::test_ddpg_speed[False-backward] |
252.03 | 247.59 | -1.76% |
benchmarks/test_objectives_benchmarks.py::test_a2c_speed[False-None] |
179.59 | 176.47 | -1.74% |
benchmarks/test_objectives_benchmarks.py::test_gae_speed[vec_generalized_advantage_estimate-True-1-512] |
657.40 | 668.65 | +1.71% |
benchmarks/test_objectives_benchmarks.py::test_dqn_speed[True-None] |
1,776 | 1,806 | +1.69% |
benchmarks/test_objectives_benchmarks.py::test_iql_speed[True-None] |
115.62 | 117.57 | +1.68% |
benchmarks/test_objectives_benchmarks.py::test_gae_speed[vec_generalized_advantage_estimate-False-1-512] |
2,214 | 2,250 | +1.62% |
benchmarks/test_collectors_benchmark.py::test_single_with_rb |
8.6701 | 8.8100 | +1.61% |
benchmarks/test_replaybuffer_benchmark.py::test_rb_sample[TensorDictPrioritizedReplayBuffer-ListStorage-None-4000] |
167.20 | 169.83 | +1.57% |
benchmarks/test_objectives_benchmarks.py::test_td3_speed[reduce-overhead-None] |
562.43 | 571.24 | +1.57% |
benchmarks/test_non_tensor_env_benchmark.py::test_non_tensor_env_rollout_speed[1000-parallel-buffers-True] |
0.5397 | 0.5314 | -1.53% |
benchmarks/test_objectives_benchmarks.py::test_iql_speed[reduce-overhead-None] |
115.98 | 117.75 | +1.53% |
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[False-True-True-True-True] |
21,794 | 21,463 | -1.52% |
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[False-False-True-False-False] |
50,496 | 51,253 | +1.50% |
benchmarks/test_storage_write_benchmark.py::TestStorageWriteBenchmark::test_storage_write_lazystack[200-img_shape3-large_batch] |
324.61 | 329.29 | +1.44% |
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[False-True-False-False-False] |
50,943 | 51,675 | +1.44% |
benchmarks/test_non_tensor_env_benchmark.py::test_non_tensor_env_rollout_speed[1000-parallel-buffers-False] |
0.6296 | 0.6206 | -1.44% |
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[False-False-False-False-False] |
45,645 | 46,260 | +1.35% |
benchmarks/test_rnn_reset_backends_benchmark.py::test_rnn_rollout_with_intermediate_resets[b256-t128-i32-h512-scan-False-0-gru] |
2.9879 | 3.0274 | +1.32% |
benchmarks/test_rnn_reset_backends_benchmark.py::test_rnn_rollout_with_intermediate_resets[b256-t128-i32-h512-scan-True-0-gru] |
4.2439 | 4.2990 | +1.30% |
benchmarks/test_replaybuffer_benchmark.py::TestPrioritizedReplayBufferBenchmark::test_sampler_sample_scale[1000000-cpu] |
96.51 | 97.75 | +1.29% |
benchmarks/test_replaybuffer_benchmark.py::test_rb_iterate[TensorDictPrioritizedReplayBuffer-ListStorage-None-4000] |
168.37 | 170.47 | +1.25% |
benchmarks/test_storage_write_benchmark.py::TestStorageWriteBenchmark::test_collector_stack_then_write[50-img_shape0-small] |
882.70 | 893.61 | +1.24% |
benchmarks/test_compressed_storage_benchmark.py::TestCompressedStorageBenchmark::test_tensor_to_bytestream_speed[torch.save] |
7,130 | 7,045 | -1.19% |
benchmarks/test_compressed_storage_benchmark.py::TestCompressedStorageBenchmark::test_tensor_to_bytestream_speed[numpy] |
383,429 | 379,173 | -1.11% |
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[True-True-True-False-True] |
43,385 | 42,910 | -1.09% |
benchmarks/test_replaybuffer_benchmark.py::test_rb_extend_sample[ReplayBuffer-LazyTensorStorage-RandomSampler-1000000-10000-100-False] |
48.55 | 49.08 | +1.08% |
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[False-True-False-False-True] |
31,804 | 31,464 | -1.07% |
benchmarks/test_objectives_benchmarks.py::test_redq_speed[False-None] |
97.01 | 96.00 | -1.05% |
benchmarks/test_rnn_reset_backends_benchmark.py::test_rnn_rollout_with_intermediate_resets[b256-t128-i32-h512-cudnn-True-0-gru] |
1.4212 | 1.4357 | +1.02% |
benchmarks/test_objectives_benchmarks.py::test_sac_speed[reduce-overhead-None] |
487.30 | 482.48 | -0.99% |
benchmarks/test_storage_write_benchmark.py::TestCollectorIntegrationBenchmark::test_collector_with_rb[200-img_shape1-large_batch] |
13.64 | 13.51 | -0.97% |
| ... | ... | ... | Showing 120 of 187 comparisons, sorted by absolute change. |
GPU
Compared 197 benchmarks. Regressions over 5%: 15. Improvements over 5%: 9.
| Benchmark | main ops | PR ops | Change |
|---|---|---|---|
benchmarks/test_replaybuffer_benchmark.py::test_rb_populate[TensorDictReplayBuffer-ListStorage-SamplerWithoutReplacement-400] |
43.00 | 191.51 | +345.41% |
benchmarks/test_replaybuffer_benchmark.py::test_rb_populate[TensorDictReplayBuffer-ListStorage-RandomSampler-400] |
193.50 | 38.11 | -80.31% |
benchmarks/test_replaybuffer_benchmark.py::test_rb_extend_sample[ReplayBuffer-LazyTensorStorage-RandomSampler-1000000-10000-100-False] |
28.09 | 47.14 | +67.83% |
benchmarks/test_objectives_benchmarks.py::test_redq_deprec_speed[reduce-overhead-None] |
82.63 | 105.49 | +27.65% |
benchmarks/test_replaybuffer_benchmark.py::test_rb_sample[TensorDictReplayBuffer-LazyMemmapStorage-SamplerWithoutReplacement-10000] |
2,671 | 3,383 | +26.64% |
benchmarks/test_replaybuffer_benchmark.py::test_rb_iterate[TensorDictReplayBuffer-LazyTensorStorage-SamplerWithoutReplacement-10000] |
3,653 | 2,718 | -25.58% |
benchmarks/test_replaybuffer_benchmark.py::test_rb_iterate[TensorDictReplayBuffer-LazyMemmapStorage-RandomSampler-10000] |
3,370 | 2,512 | -25.46% |
benchmarks/test_replaybuffer_benchmark.py::test_rb_sample[TensorDictReplayBuffer-LazyTensorStorage-RandomSampler-10000] |
2,838 | 3,426 | +20.74% |
benchmarks/test_replaybuffer_benchmark.py::test_rb_iterate[TensorDictReplayBuffer-LazyMemmapStorage-SamplerWithoutReplacement-10000] |
3,372 | 2,817 | -16.45% |
benchmarks/test_replaybuffer_benchmark.py::test_rb_iterate[TensorDictReplayBuffer-LazyTensorStorage-RandomSampler-10000] |
3,001 | 3,486 | +16.14% |
benchmarks/test_replaybuffer_benchmark.py::test_rb_sample[TensorDictReplayBuffer-LazyTensorStorage-SamplerWithoutReplacement-10000] |
2,787 | 3,180 | +14.10% |
benchmarks/test_collectors_benchmark.py::test_single_with_rb_pixels |
5.3278 | 4.6386 | -12.94% |
benchmarks/test_replaybuffer_benchmark.py::test_rb_populate[TensorDictPrioritizedReplayBuffer-LazyMemmapStorage-None-400] |
450.85 | 399.89 | -11.30% |
benchmarks/test_storage_write_benchmark.py::TestStorageWriteBenchmark::test_storage_write_contiguous[100-img_shape2-large_img] |
521.71 | 464.47 | -10.97% |
benchmarks/test_objectives_benchmarks.py::test_dqn_speed[True-backward] |
887.92 | 969.97 | +9.24% |
benchmarks/test_replaybuffer_benchmark.py::test_rb_sample[TensorDictPrioritizedReplayBuffer-LazyMemmapStorage-None-10000] |
1,994 | 1,822 | -8.62% |
benchmarks/test_replaybuffer_benchmark.py::test_rb_iterate[TensorDictPrioritizedReplayBuffer-LazyMemmapStorage-None-10000] |
1,837 | 1,984 | +7.96% |
benchmarks/test_replaybuffer_benchmark.py::test_rb_sample[TensorDictReplayBuffer-LazyTensorStorage-sampler7-10000] |
780.92 | 719.02 | -7.93% |
benchmarks/test_replaybuffer_benchmark.py::test_rb_sample[TensorDictReplayBuffer-ListStorage-RandomSampler-4000] |
160.35 | 148.92 | -7.13% |
benchmarks/test_storage_write_benchmark.py::TestStorageWriteBenchmark::test_storage_write_contiguous[200-img_shape3-large_batch] |
722.88 | 672.68 | -6.94% |
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[True-True-True-False-True] |
41,740 | 39,221 | -6.04% |
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[True-True-False-False-True] |
37,726 | 35,662 | -5.47% |
benchmarks/test_replaybuffer_benchmark.py::test_rb_populate[TensorDictReplayBuffer-LazyMemmapStorage-SamplerWithoutReplacement-400] |
497.56 | 471.91 | -5.16% |
benchmarks/test_rnn_reset_backends_benchmark.py::test_rnn_rollout_with_intermediate_resets[b256-t128-i32-h512-scan-False-0-lstm] |
21.06 | 19.99 | -5.08% |
benchmarks/test_envs_benchmark.py::test_simple |
1.2217 | 1.1613 | -4.95% |
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[True-False-False-True-True] |
19,992 | 19,006 | -4.93% |
benchmarks/test_rnn_reset_backends_benchmark.py::test_rnn_rollout_with_intermediate_resets[b256-t128-i32-h512-scan-False-0-gru] |
22.35 | 21.31 | -4.66% |
benchmarks/test_replaybuffer_benchmark.py::test_rb_sample[TensorDictPrioritizedReplayBuffer-LazyTensorStorage-None-10000] |
1,915 | 2,001 | +4.50% |
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[True-False-True-False-True] |
38,215 | 36,556 | -4.34% |
benchmarks/test_objectives_benchmarks.py::test_reinforce_speed[False-backward] |
266.18 | 276.68 | +3.94% |
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[False-False-False-False-True] |
28,703 | 27,609 | -3.81% |
benchmarks/test_replaybuffer_benchmark.py::test_rb_extend_sample[ReplayBuffer-LazyTensorStorage-RandomSampler-1000000-10000-100-True] |
21.80 | 20.98 | -3.76% |
benchmarks/test_replaybuffer_benchmark.py::test_rb_populate[TensorDictReplayBuffer-LazyMemmapStorage-RandomSampler-400] |
518.86 | 499.78 | -3.68% |
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[False-True-True-True-True] |
20,861 | 20,111 | -3.60% |
benchmarks/test_replaybuffer_benchmark.py::test_rb_extend_sample[ReplayBuffer-LazyTensorStorage-RandomSampler-100000-10000-100-True] |
22.92 | 22.14 | -3.39% |
benchmarks/test_storage_write_benchmark.py::TestCollectorIntegrationBenchmark::test_collector_without_rb_cuda[200-img_shape1-large_batch] |
8.7559 | 8.4621 | -3.36% |
benchmarks/test_compressed_storage_benchmark.py::TestCompressedStorageBenchmark::test_tensor_to_bytestream_speed[safetensors] |
23,856 | 23,066 | -3.31% |
benchmarks/test_objectives_benchmarks.py::test_a2c_speed[True-None] |
736.89 | 712.53 | -3.31% |
benchmarks/test_storage_write_benchmark.py::TestStorageWriteBenchmark::test_collector_stack_then_write[100-img_shape2-large_img] |
169.31 | 163.78 | -3.26% |
benchmarks/test_objectives_benchmarks.py::test_ppo_speed[True-None] |
668.74 | 690.19 | +3.21% |
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[False-True-False-True-True] |
19,841 | 19,207 | -3.19% |
benchmarks/test_replaybuffer_benchmark.py::test_rb_populate[TensorDictReplayBuffer-LazyTensorStorage-SamplerWithoutReplacement-400] |
981.52 | 950.33 | -3.18% |
benchmarks/test_storage_write_benchmark.py::TestCollectorIntegrationBenchmark::test_collector_without_rb[100-img_shape0-atari] |
30.32 | 29.41 | -3.02% |
benchmarks/test_storage_write_benchmark.py::TestCollectorIntegrationBenchmark::test_collector_with_rb[200-img_shape1-large_batch] |
13.52 | 13.11 | -3.02% |
benchmarks/test_storage_write_benchmark.py::TestCollectorIntegrationBenchmark::test_collector_with_rb_cuda[200-img_shape1-large_batch] |
8.4166 | 8.1645 | -2.99% |
benchmarks/test_rnn_reset_backends_benchmark.py::test_rnn_rollout_with_intermediate_resets[b256-t128-i32-h512-scan-True-0-gru] |
46.95 | 48.35 | +2.97% |
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[False-False-True-True-True] |
18,771 | 18,215 | -2.97% |
benchmarks/test_objectives_benchmarks.py::test_cql_speed[True-backward] |
213.20 | 219.44 | +2.93% |
benchmarks/test_objectives_benchmarks.py::test_values[generalized_advantage_estimate-True-True] |
47.08 | 45.71 | -2.90% |
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[False-False-False-True-True] |
18,193 | 17,673 | -2.86% |
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[False-True-True-True-False] |
34,657 | 33,690 | -2.79% |
benchmarks/test_storage_write_benchmark.py::TestStorageWriteBenchmark::test_storage_write_lazystack[100-img_shape1-atari] |
707.46 | 687.75 | -2.79% |
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[True-False-False-False-True] |
34,343 | 33,393 | -2.77% |
benchmarks/test_replaybuffer_benchmark.py::test_rb_sample[TensorDictReplayBuffer-LazyMemmapStorage-sampler6-10000] |
742.46 | 762.60 | +2.71% |
benchmarks/test_storage_write_benchmark.py::TestStorageWriteBenchmark::test_storage_write_lazystack[50-img_shape0-small] |
4,326 | 4,211 | -2.67% |
benchmarks/test_storage_write_benchmark.py::TestStorageWriteBenchmark::test_storage_write_contiguous[50-img_shape0-small] |
6,067 | 5,905 | -2.66% |
benchmarks/test_collectors_benchmark.py::test_single_pixels |
6.2465 | 6.0817 | -2.64% |
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[False-False-False-True-False] |
27,568 | 26,866 | -2.55% |
benchmarks/test_storage_write_benchmark.py::TestCollectorIntegrationBenchmark::test_collector_with_rb_cuda[100-img_shape0-atari] |
16.73 | 16.32 | -2.49% |
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[True-False-True-True-False] |
34,593 | 33,741 | -2.46% |
benchmarks/test_objectives_benchmarks.py::test_dqn_speed[True-None] |
1,906 | 1,859 | -2.44% |
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[True-True-False-False-False] |
63,951 | 62,417 | -2.40% |
benchmarks/test_objectives_benchmarks.py::test_values[vec_generalized_advantage_estimate-True-True] |
303.33 | 310.58 | +2.39% |
benchmarks/test_objectives_benchmarks.py::test_redq_deprec_speed[False-None] |
97.16 | 94.85 | -2.38% |
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[True-True-False-True-False] |
38,483 | 37,580 | -2.35% |
benchmarks/test_storage_write_benchmark.py::TestCollectorIntegrationBenchmark::test_collector_without_rb_cuda[100-img_shape0-atari] |
17.50 | 17.09 | -2.33% |
benchmarks/test_storage_write_benchmark.py::TestStorageWriteBenchmark::test_storage_write_lazystack[200-img_shape3-large_batch] |
325.52 | 318.11 | -2.28% |
benchmarks/test_storage_write_benchmark.py::TestCollectorIntegrationBenchmark::test_collector_with_rb[100-img_shape0-atari] |
26.40 | 25.80 | -2.27% |
benchmarks/test_storage_write_benchmark.py::TestCollectorIntegrationBenchmark::test_collector_without_rb[200-img_shape1-large_batch] |
15.35 | 15.00 | -2.24% |
benchmarks/test_replaybuffer_benchmark.py::test_rb_extend_sample[ReplayBuffer-LazyTensorStorage-RandomSampler-100000-10000-100-False] |
52.20 | 51.05 | -2.21% |
benchmarks/test_replaybuffer_benchmark.py::test_rb_populate[TensorDictReplayBuffer-LazyTensorStorage-RandomSampler-400] |
980.56 | 958.91 | -2.21% |
benchmarks/test_objectives_benchmarks.py::test_values[td1_return_estimate-False-False] |
19.80 | 19.37 | -2.17% |
benchmarks/test_objectives_benchmarks.py::test_ppo_speed[reduce-overhead-None] |
795.70 | 812.70 | +2.14% |
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[True-False-True-False-False] |
63,923 | 62,595 | -2.08% |
benchmarks/test_replaybuffer_benchmark.py::TestPrioritizedReplayBufferBenchmark::test_sample_mixed_devices[1000000-cuda_storage_cuda_samp... |
1,467 | 1,438 | -1.92% |
benchmarks/test_objectives_benchmarks.py::test_ppo_speed[True-backward] |
345.02 | 351.53 | +1.89% |
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[True-False-False-True-False] |
31,869 | 31,288 | -1.82% |
benchmarks/test_non_tensor_env_benchmark.py::test_non_tensor_env_rollout_speed[1000-serial-no-buffers-False] |
0.6875 | 0.6752 | -1.79% |
benchmarks/test_non_tensor_env_benchmark.py::test_non_tensor_env_rollout_speed[1000-serial-buffers-False] |
0.5972 | 0.5865 | -1.79% |
benchmarks/test_objectives_benchmarks.py::test_values[td_lambda_return_estimate-True-False] |
12.00 | 11.79 | -1.78% |
benchmarks/test_collectors_benchmark.py::test_single |
6.6925 | 6.5742 | -1.77% |
benchmarks/test_storage_write_benchmark.py::TestStorageWriteBenchmark::test_collector_lazystack_then_write[100-img_shape2-large_img] |
404.75 | 397.65 | -1.75% |
benchmarks/test_collectors_benchmark.py::test_single_with_rb |
5.9239 | 5.8226 | -1.71% |
benchmarks/test_objectives_benchmarks.py::test_values[td0_return_estimate-False-False] |
11,426 | 11,236 | -1.67% |
benchmarks/test_envs_benchmark.py::test_transformed |
0.6908 | 0.7023 | +1.66% |
benchmarks/test_objectives_benchmarks.py::test_a2c_speed[True-backward] |
357.87 | 363.36 | +1.53% |
benchmarks/test_objectives_benchmarks.py::test_ddpg_speed[True-backward] |
466.93 | 473.61 | +1.43% |
benchmarks/test_non_tensor_env_benchmark.py::test_non_tensor_env_rollout_speed[1000-parallel-buffers-False] |
0.6045 | 0.5960 | -1.41% |
benchmarks/test_replaybuffer_benchmark.py::test_rb_sample[TensorDictReplayBuffer-ListStorage-SamplerWithoutReplacement-4000] |
164.37 | 166.60 | +1.36% |
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[False-False-True-False-True] |
29,995 | 29,597 | -1.33% |
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[False-False-True-True-False] |
29,255 | 28,875 | -1.30% |
benchmarks/test_objectives_benchmarks.py::test_ddpg_speed[True-None] |
807.63 | 818.02 | +1.29% |
benchmarks/test_objectives_benchmarks.py::test_redq_deprec_speed[False-backward] |
70.70 | 69.79 | -1.28% |
benchmarks/test_objectives_benchmarks.py::test_gae_speed[vec_generalized_advantage_estimate-True-1-512] |
1,245 | 1,229 | -1.27% |
benchmarks/test_replaybuffer_benchmark.py::test_rb_iterate[TensorDictPrioritizedReplayBuffer-ListStorage-None-4000] |
160.06 | 162.09 | +1.27% |
benchmarks/test_objectives_benchmarks.py::test_sac_speed[False-backward] |
77.98 | 78.96 | +1.25% |
benchmarks/test_objectives_benchmarks.py::test_gae_speed[vec_generalized_advantage_estimate-False-32-512] |
1,292 | 1,276 | -1.25% |
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[False-True-True-False-True] |
32,136 | 31,737 | -1.24% |
benchmarks/test_objectives_benchmarks.py::test_dqn_speed[reduce-overhead-None] |
1,863 | 1,886 | +1.23% |
benchmarks/test_objectives_benchmarks.py::test_sac_speed[False-None] |
109.01 | 110.35 | +1.23% |
benchmarks/test_objectives_benchmarks.py::test_cql_speed[False-None] |
53.25 | 52.61 | -1.22% |
benchmarks/test_compressed_storage_benchmark.py::TestCompressedStorageBenchmark::test_tensor_to_bytestream_speed[untyped_storage] |
7.8932 | 7.7986 | -1.20% |
benchmarks/test_non_tensor_env_benchmark.py::test_non_tensor_env_rollout_speed[1000-serial-no-buffers-True] |
0.5956 | 0.5887 | -1.15% |
benchmarks/test_objectives_benchmarks.py::test_a2c_speed[False-None] |
271.40 | 268.33 | -1.13% |
benchmarks/test_objectives_benchmarks.py::test_ppo_speed[False-backward] |
129.71 | 131.16 | +1.11% |
benchmarks/test_objectives_benchmarks.py::test_values[vec_td1_return_estimate-False-False] |
841.90 | 832.56 | -1.11% |
benchmarks/test_storage_write_benchmark.py::TestStorageWriteBenchmark::test_collector_lazystack_then_write[100-img_shape1-atari] |
654.73 | 647.69 | -1.08% |
benchmarks/test_non_tensor_env_benchmark.py::test_non_tensor_env_rollout_speed[1000-serial-buffers-True] |
0.5151 | 0.5096 | -1.07% |
benchmarks/test_objectives_benchmarks.py::test_redq_deprec_speed[True-None] |
415.83 | 411.40 | -1.07% |
benchmarks/test_objectives_benchmarks.py::test_iql_speed[reduce-overhead-None] |
107.53 | 106.41 | -1.05% |
benchmarks/test_non_tensor_env_benchmark.py::test_non_tensor_env_rollout_speed[1000-single-False] |
1.5904 | 1.5739 | -1.04% |
benchmarks/test_replaybuffer_benchmark.py::test_rb_sample[TensorDictPrioritizedReplayBuffer-ListStorage-None-4000] |
160.06 | 161.71 | +1.03% |
benchmarks/test_objectives_benchmarks.py::test_td3_speed[True-None] |
739.45 | 732.06 | -1.00% |
benchmarks/test_non_tensor_env_benchmark.py::test_non_tensor_env_rollout_speed[1000-single-True] |
1.3081 | 1.2950 | -1.00% |
benchmarks/test_objectives_benchmarks.py::test_iql_speed[False-None] |
96.96 | 96.01 | -0.98% |
benchmarks/test_envs_benchmark.py::test_step_mdp_speed[False-True-False-False-True] |
29,829 | 30,114 | +0.96% |
benchmarks/test_objectives_benchmarks.py::test_ddpg_speed[reduce-overhead-None] |
814.52 | 822.12 | +0.93% |
benchmarks/test_replaybuffer_benchmark.py::test_rb_extend_sample[ReplayBuffer-LazyTensorStorage-RandomSampler-10000-10000-100-False] |
53.42 | 52.93 | -0.91% |
benchmarks/test_replaybuffer_benchmark.py::TestPrioritizedReplayBufferBenchmark::test_sampler_sample_scale[1000000-cuda] |
2,205 | 2,224 | +0.85% |
benchmarks/test_storage_write_benchmark.py::TestStorageWriteBenchmark::test_collector_stack_then_write[50-img_shape0-small] |
860.09 | 867.32 | +0.84% |
| ... | ... | ... | Showing 120 of 197 comparisons, sorted by absolute change. |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Stack from ghstack (oldest at bottom):