[Refactor] Faster envs (2) #1457

vmoens · 2023-08-11T09:52:58Z

A minimalist version of #1448

github-actions · 2023-08-11T10:19:39Z

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

Total Benchmarks: 89. Improved: $\large\color{#35bf28}11$. Worsened: $\large\color{#d91a1a}4$.

Expand to view detailed results

Name	Max	Mean	Ops	Ops on Repo `HEAD`	Change
test_single	0.1393s	0.1384s	7.2279 Ops/s	6.7754 Ops/s	$\textbf{\color{#35bf28}+6.68\%}$
test_sync	0.1479s	77.3141ms	12.9342 Ops/s	12.0750 Ops/s	$\textbf{\color{#35bf28}+7.12\%}$
test_async	0.1938s	73.0239ms	13.6942 Ops/s	13.0088 Ops/s	$\textbf{\color{#35bf28}+5.27\%}$
test_simple	0.6867s	0.6170s	1.6207 Ops/s	1.6093 Ops/s	$\color{#35bf28}+0.70\%$
test_transformed	1.6663s	1.6043s	0.6233 Ops/s	0.5920 Ops/s	$\textbf{\color{#35bf28}+5.29\%}$
test_serial	1.7591s	1.7011s	0.5879 Ops/s	0.5093 Ops/s	$\textbf{\color{#35bf28}+15.42\%}$
test_parallel	1.5634s	1.4942s	0.6693 Ops/s	0.6204 Ops/s	$\textbf{\color{#35bf28}+7.87\%}$
test_step_mdp_speed[True-True-True-True-True]	0.1606ms	44.6101μs	22.4164 KOps/s	22.6276 KOps/s	$\color{#d91a1a}-0.93\%$
test_step_mdp_speed[True-True-True-True-False]	93.2020μs	25.3293μs	39.4799 KOps/s	39.8819 KOps/s	$\color{#d91a1a}-1.01\%$
test_step_mdp_speed[True-True-True-False-True]	0.1092ms	31.1811μs	32.0708 KOps/s	32.1355 KOps/s	$\color{#d91a1a}-0.20\%$
test_step_mdp_speed[True-True-True-False-False]	45.2010μs	17.4488μs	57.3104 KOps/s	58.2652 KOps/s	$\color{#d91a1a}-1.64\%$
test_step_mdp_speed[True-True-False-True-True]	0.1340ms	45.9581μs	21.7590 KOps/s	21.7292 KOps/s	$\color{#35bf28}+0.14\%$
test_step_mdp_speed[True-True-False-True-False]	63.6010μs	27.0889μs	36.9154 KOps/s	37.3445 KOps/s	$\color{#d91a1a}-1.15\%$
test_step_mdp_speed[True-True-False-False-True]	0.1086ms	33.4016μs	29.9387 KOps/s	29.9290 KOps/s	$\color{#35bf28}+0.03\%$
test_step_mdp_speed[True-True-False-False-False]	44.2000μs	19.3657μs	51.6378 KOps/s	52.4679 KOps/s	$\color{#d91a1a}-1.58\%$
test_step_mdp_speed[True-False-True-True-True]	0.1252ms	48.3980μs	20.6620 KOps/s	20.9121 KOps/s	$\color{#d91a1a}-1.20\%$
test_step_mdp_speed[True-False-True-True-False]	0.1022ms	28.8974μs	34.6052 KOps/s	34.5226 KOps/s	$\color{#35bf28}+0.24\%$
test_step_mdp_speed[True-False-True-False-True]	0.1592ms	33.8083μs	29.5785 KOps/s	30.1931 KOps/s	$\color{#d91a1a}-2.04\%$
test_step_mdp_speed[True-False-True-False-False]	97.5020μs	19.2759μs	51.8783 KOps/s	52.0894 KOps/s	$\color{#d91a1a}-0.41\%$
test_step_mdp_speed[True-False-False-True-True]	97.9020μs	50.0849μs	19.9661 KOps/s	20.2904 KOps/s	$\color{#d91a1a}-1.60\%$
test_step_mdp_speed[True-False-False-True-False]	0.1024ms	30.3109μs	32.9914 KOps/s	32.9354 KOps/s	$\color{#35bf28}+0.17\%$
test_step_mdp_speed[True-False-False-False-True]	0.1616ms	35.3969μs	28.2511 KOps/s	28.9924 KOps/s	$\color{#d91a1a}-2.56\%$
test_step_mdp_speed[True-False-False-False-False]	48.3000μs	20.8032μs	48.0696 KOps/s	48.2027 KOps/s	$\color{#d91a1a}-0.28\%$
test_step_mdp_speed[False-True-True-True-True]	0.1137ms	48.1004μs	20.7898 KOps/s	20.9114 KOps/s	$\color{#d91a1a}-0.58\%$
test_step_mdp_speed[False-True-True-True-False]	50.5010μs	28.7913μs	34.7327 KOps/s	34.0017 KOps/s	$\color{#35bf28}+2.15\%$
test_step_mdp_speed[False-True-True-False-True]	0.1201ms	37.4881μs	26.6752 KOps/s	25.6100 KOps/s	$\color{#35bf28}+4.16\%$
test_step_mdp_speed[False-True-True-False-False]	3.2953ms	21.6042μs	46.2874 KOps/s	45.4794 KOps/s	$\color{#35bf28}+1.78\%$
test_step_mdp_speed[False-True-False-True-True]	0.1248ms	49.8007μs	20.0800 KOps/s	20.3685 KOps/s	$\color{#d91a1a}-1.42\%$
test_step_mdp_speed[False-True-False-True-False]	97.4020μs	30.5748μs	32.7067 KOps/s	32.7694 KOps/s	$\color{#d91a1a}-0.19\%$
test_step_mdp_speed[False-True-False-False-True]	0.1186ms	38.8381μs	25.7479 KOps/s	24.6716 KOps/s	$\color{#35bf28}+4.36\%$
test_step_mdp_speed[False-True-False-False-False]	97.2020μs	23.3945μs	42.7452 KOps/s	42.1579 KOps/s	$\color{#35bf28}+1.39\%$
test_step_mdp_speed[False-False-True-True-True]	78.2020μs	51.0446μs	19.5907 KOps/s	19.5684 KOps/s	$\color{#35bf28}+0.11\%$
test_step_mdp_speed[False-False-True-True-False]	0.1019ms	32.6807μs	30.5991 KOps/s	30.9838 KOps/s	$\color{#d91a1a}-1.24\%$
test_step_mdp_speed[False-False-True-False-True]	0.1152ms	38.7226μs	25.8247 KOps/s	24.7414 KOps/s	$\color{#35bf28}+4.38\%$
test_step_mdp_speed[False-False-True-False-False]	56.5010μs	23.3576μs	42.8126 KOps/s	42.8149 KOps/s	$-0.01\%$
test_step_mdp_speed[False-False-False-True-True]	0.1438ms	52.1779μs	19.1652 KOps/s	19.1596 KOps/s	$\color{#35bf28}+0.03\%$
test_step_mdp_speed[False-False-False-True-False]	89.2020μs	33.8109μs	29.5762 KOps/s	29.6812 KOps/s	$\color{#d91a1a}-0.35\%$
test_step_mdp_speed[False-False-False-False-True]	70.9010μs	39.8541μs	25.0915 KOps/s	23.1099 KOps/s	$\textbf{\color{#35bf28}+8.57\%}$
test_step_mdp_speed[False-False-False-False-False]	96.5010μs	24.6764μs	40.5246 KOps/s	40.2648 KOps/s	$\color{#35bf28}+0.65\%$
test_values[generalized_advantage_estimate-True-True]	15.7383ms	13.5918ms	73.5737 Ops/s	73.3193 Ops/s	$\color{#35bf28}+0.35\%$
test_values[vec_generalized_advantage_estimate-True-True]	45.9083ms	40.8992ms	24.4503 Ops/s	23.9171 Ops/s	$\color{#35bf28}+2.23\%$
test_values[td0_return_estimate-False-False]	0.3003ms	0.1860ms	5.3768 KOps/s	4.7050 KOps/s	$\textbf{\color{#35bf28}+14.28\%}$
test_values[td1_return_estimate-False-False]	13.5835ms	13.2667ms	75.3767 Ops/s	76.3167 Ops/s	$\color{#d91a1a}-1.23\%$
test_values[vec_td1_return_estimate-False-False]	46.7013ms	41.3187ms	24.2021 Ops/s	24.5061 Ops/s	$\color{#d91a1a}-1.24\%$
test_values[td_lambda_return_estimate-True-False]	32.9919ms	31.5605ms	31.6852 Ops/s	31.5362 Ops/s	$\color{#35bf28}+0.47\%$
test_values[vec_td_lambda_return_estimate-True-False]	46.5164ms	41.4118ms	24.1477 Ops/s	24.3243 Ops/s	$\color{#d91a1a}-0.73\%$
test_gae_speed[generalized_advantage_estimate-False-1-512]	13.1433ms	11.9557ms	83.6420 Ops/s	84.0546 Ops/s	$\color{#d91a1a}-0.49\%$
test_gae_speed[vec_generalized_advantage_estimate-True-1-512]	8.4312ms	3.3242ms	300.8235 Ops/s	288.3139 Ops/s	$\color{#35bf28}+4.34\%$
test_gae_speed[vec_generalized_advantage_estimate-False-1-512]	0.5901ms	0.4541ms	2.2021 KOps/s	2.1204 KOps/s	$\color{#35bf28}+3.85\%$
test_gae_speed[vec_generalized_advantage_estimate-True-32-512]	62.5153ms	56.2641ms	17.7733 Ops/s	17.8985 Ops/s	$\color{#d91a1a}-0.70\%$
test_gae_speed[vec_generalized_advantage_estimate-False-32-512]	7.3074ms	2.7554ms	362.9238 Ops/s	359.5459 Ops/s	$\color{#35bf28}+0.94\%$
test_dqn_speed	8.8302ms	1.8392ms	543.7140 Ops/s	547.8821 Ops/s	$\color{#d91a1a}-0.76\%$
test_ddpg_speed	9.6086ms	2.6534ms	376.8743 Ops/s	367.0492 Ops/s	$\color{#35bf28}+2.68\%$
test_sac_speed	17.2164ms	7.8280ms	127.7463 Ops/s	123.3549 Ops/s	$\color{#35bf28}+3.56\%$
test_redq_speed	22.7394ms	15.9568ms	62.6694 Ops/s	62.4717 Ops/s	$\color{#35bf28}+0.32\%$
test_redq_deprec_speed	16.8215ms	12.2293ms	81.7706 Ops/s	76.8589 Ops/s	$\textbf{\color{#35bf28}+6.39\%}$
test_td3_speed	10.8532ms	9.8453ms	101.5710 Ops/s	96.0183 Ops/s	$\textbf{\color{#35bf28}+5.78\%}$
test_cql_speed	42.3939ms	36.6898ms	27.2555 Ops/s	38.0057 Ops/s	$\textbf{\color{#d91a1a}-28.29\%}$
test_a2c_speed	11.3330ms	5.2170ms	191.6802 Ops/s	189.3897 Ops/s	$\color{#35bf28}+1.21\%$
test_ppo_speed	41.4040ms	5.7131ms	175.0376 Ops/s	182.8101 Ops/s	$\color{#d91a1a}-4.25\%$
test_reinforce_speed	11.0398ms	4.0813ms	245.0214 Ops/s	245.0349 Ops/s	$-0.01\%$
test_iql_speed	32.7330ms	23.3990ms	42.7369 Ops/s	46.0300 Ops/s	$\textbf{\color{#d91a1a}-7.15\%}$
test_sample_rb[TensorDictReplayBuffer-ListStorage-RandomSampler-4000]	3.3918ms	2.6365ms	379.2877 Ops/s	378.9104 Ops/s	$\color{#35bf28}+0.10\%$
test_sample_rb[TensorDictReplayBuffer-LazyMemmapStorage-RandomSampler-10000]	5.1394ms	2.8096ms	355.9191 Ops/s	360.8460 Ops/s	$\color{#d91a1a}-1.37\%$
test_sample_rb[TensorDictReplayBuffer-LazyTensorStorage-RandomSampler-10000]	4.7708ms	2.8414ms	351.9367 Ops/s	354.6221 Ops/s	$\color{#d91a1a}-0.76\%$
test_sample_rb[TensorDictReplayBuffer-ListStorage-SamplerWithoutReplacement-4000]	3.6404ms	2.6794ms	373.2245 Ops/s	378.2628 Ops/s	$\color{#d91a1a}-1.33\%$
test_sample_rb[TensorDictReplayBuffer-LazyMemmapStorage-SamplerWithoutReplacement-10000]	5.3296ms	2.8291ms	353.4745 Ops/s	355.9805 Ops/s	$\color{#d91a1a}-0.70\%$
test_sample_rb[TensorDictReplayBuffer-LazyTensorStorage-SamplerWithoutReplacement-10000]	4.3288ms	2.8099ms	355.8840 Ops/s	358.6406 Ops/s	$\color{#d91a1a}-0.77\%$
test_sample_rb[TensorDictPrioritizedReplayBuffer-ListStorage-None-4000]	3.6365ms	2.6348ms	379.5307 Ops/s	387.2584 Ops/s	$\color{#d91a1a}-2.00\%$
test_sample_rb[TensorDictPrioritizedReplayBuffer-LazyMemmapStorage-None-10000]	0.1417s	3.2116ms	311.3740 Ops/s	359.8914 Ops/s	$\textbf{\color{#d91a1a}-13.48\%}$
test_sample_rb[TensorDictPrioritizedReplayBuffer-LazyTensorStorage-None-10000]	5.2880ms	2.8446ms	351.5433 Ops/s	353.8041 Ops/s	$\color{#d91a1a}-0.64\%$
test_iterate_rb[TensorDictReplayBuffer-ListStorage-RandomSampler-4000]	3.5652ms	2.7108ms	368.8987 Ops/s	380.0087 Ops/s	$\color{#d91a1a}-2.92\%$
test_iterate_rb[TensorDictReplayBuffer-LazyMemmapStorage-RandomSampler-10000]	5.1808ms	2.9410ms	340.0224 Ops/s	358.1901 Ops/s	$\textbf{\color{#d91a1a}-5.07\%}$
test_iterate_rb[TensorDictReplayBuffer-LazyTensorStorage-RandomSampler-10000]	5.3371ms	2.9584ms	338.0169 Ops/s	353.2290 Ops/s	$\color{#d91a1a}-4.31\%$
test_iterate_rb[TensorDictReplayBuffer-ListStorage-SamplerWithoutReplacement-4000]	3.4765ms	2.6998ms	370.3962 Ops/s	378.8107 Ops/s	$\color{#d91a1a}-2.22\%$
test_iterate_rb[TensorDictReplayBuffer-LazyMemmapStorage-SamplerWithoutReplacement-10000]	4.7352ms	2.8686ms	348.5967 Ops/s	359.3995 Ops/s	$\color{#d91a1a}-3.01\%$
test_iterate_rb[TensorDictReplayBuffer-LazyTensorStorage-SamplerWithoutReplacement-10000]	4.9174ms	2.8607ms	349.5603 Ops/s	355.2383 Ops/s	$\color{#d91a1a}-1.60\%$
test_iterate_rb[TensorDictPrioritizedReplayBuffer-ListStorage-None-4000]	3.5962ms	2.6742ms	373.9450 Ops/s	377.4031 Ops/s	$\color{#d91a1a}-0.92\%$
test_iterate_rb[TensorDictPrioritizedReplayBuffer-LazyMemmapStorage-None-10000]	4.7858ms	2.8520ms	350.6280 Ops/s	359.8586 Ops/s	$\color{#d91a1a}-2.57\%$
test_iterate_rb[TensorDictPrioritizedReplayBuffer-LazyTensorStorage-None-10000]	5.9294ms	2.9027ms	344.5106 Ops/s	356.3901 Ops/s	$\color{#d91a1a}-3.33\%$
test_populate_rb[TensorDictReplayBuffer-ListStorage-RandomSampler-400]	0.2903s	30.6421ms	32.6348 Ops/s	31.9249 Ops/s	$\color{#35bf28}+2.22\%$
test_populate_rb[TensorDictReplayBuffer-LazyMemmapStorage-RandomSampler-400]	0.1471s	29.6074ms	33.7753 Ops/s	34.8381 Ops/s	$\color{#d91a1a}-3.05\%$
test_populate_rb[TensorDictReplayBuffer-LazyTensorStorage-RandomSampler-400]	0.1423s	27.0494ms	36.9695 Ops/s	37.5991 Ops/s	$\color{#d91a1a}-1.67\%$
test_populate_rb[TensorDictReplayBuffer-ListStorage-SamplerWithoutReplacement-400]	0.1435s	29.5092ms	33.8878 Ops/s	33.4265 Ops/s	$\color{#35bf28}+1.38\%$
test_populate_rb[TensorDictReplayBuffer-LazyMemmapStorage-SamplerWithoutReplacement-400]	0.1433s	27.1496ms	36.8330 Ops/s	37.9529 Ops/s	$\color{#d91a1a}-2.95\%$
test_populate_rb[TensorDictReplayBuffer-LazyTensorStorage-SamplerWithoutReplacement-400]	0.1416s	26.8441ms	37.2521 Ops/s	34.1288 Ops/s	$\textbf{\color{#35bf28}+9.15\%}$
test_populate_rb[TensorDictPrioritizedReplayBuffer-ListStorage-None-400]	0.1420s	27.1229ms	36.8692 Ops/s	37.9302 Ops/s	$\color{#d91a1a}-2.80\%$
test_populate_rb[TensorDictPrioritizedReplayBuffer-LazyMemmapStorage-None-400]	0.1462s	29.7785ms	33.5813 Ops/s	34.4476 Ops/s	$\color{#d91a1a}-2.51\%$
test_populate_rb[TensorDictPrioritizedReplayBuffer-LazyTensorStorage-None-400]	0.1445s	26.9509ms	37.1046 Ops/s	37.9909 Ops/s	$\color{#d91a1a}-2.33\%$

# Conflicts: # torchrl/collectors/collectors.py # torchrl/envs/common.py # torchrl/envs/transforms/transforms.py # torchrl/envs/utils.py # torchrl/envs/vec_env.py

matteobettini · 2023-08-31T12:50:51Z

torchrl/envs/vec_env.py

+        # output keys after reset
+        self._selected_reset_keys = {
+            _unravel_key_to_tuple(key)
+            for key in self._env_obs_keys + self.done_keys + ["_reset"]


reset keys should not include "_reset" right?
if so there will be multiple _reset keys (one per done)

We need the "_reset" keys in the reset_keys as these need to be synced across processes.
I edited the logic a bit, you can have a look

matteobettini · 2023-08-31T12:53:27Z

torchrl/envs/vec_env.py

            for key in self._selected_reset_keys:
-                _set_single_key(self.shared_tensordict_parent, out, key, clone=True)
+                if _unravel_key_to_tuple(key)[-1] != "_reset":


again, no need to check here if you just exclude the reset key from selected reset keys

but if you wanna check, there could be multiple reset keys

same response as above: we need to sync them across processes (line 627)

matteobettini · 2023-08-31T12:54:39Z

torchrl/envs/vec_env.py

            for key in self._selected_reset_keys:
-                _set_single_key(self.shared_tensordict_parent, out, key, clone=True)
+                if _unravel_key_to_tuple(key)[-1] != "_reset":


same as above

matteobettini · 2023-08-31T12:57:16Z

torchrl/collectors/collectors.py

-                self._tensordict[traj_done_or_terminated] = td_reset[
-                    traj_done_or_terminated
-                ]
+                self._tensordict = torch.where(


is this compatible with lazy stacks?

matteobettini

LGTM

Let's remember about the multiple reset keys across vec_env.

And it seems like some "next": {} might still be there in some steps of mocking envs (for example MultiKeyConutngEnv)

matteobettini · 2023-08-31T13:03:04Z

torchrl/envs/utils.py

@@ -562,6 +568,101 @@ def make_composite_from_td(data):
    return composite


+def _fuse_tensordicts(*tds, excluded, selected=None, total=None):


Is this used anywhere?

vmoens · 2023-08-31T13:07:00Z

Let's remember about the multiple reset keys across vec_env.

What do you mean?

matteobettini · 2023-08-31T13:40:26Z

Let's remember about the multiple reset keys across vec_env.

What do you mean?

i think you updated the pr, i was just saying: let’s remeber there are multiple reset keys

skandermoalla · 2023-09-01T13:09:41Z

test/test_collector.py

+# Deprecated reset_when_done
+# @pytest.mark.parametrize("num_env", [1, 2])
+# @pytest.mark.parametrize("env_name", ["vec"])
+# def test_collector_done_persist(num_env, env_name, seed=5):
+#     if num_env == 1:
+#
+#         def env_fn(seed):
+#             env = MockSerialEnv(device="cpu")
+#             env.set_seed(seed)
+#             return env
+#
+#     else:
+#
+#         def env_fn(seed):
+#             def make_env(seed):
+#                 env = MockSerialEnv(device="cpu")
+#                 env.set_seed(seed)
+#                 return env
+#
+#             env = ParallelEnv(
+#                 num_workers=num_env,
+#                 create_env_fn=make_env,
+#                 create_env_kwargs=[{"seed": i} for i in range(seed, seed + num_env)],
+#             )
+#             env.set_seed(seed)
+#             return env
+#
+#     policy = make_policy(env_name)
+#
+#     collector = SyncDataCollector(
+#         create_env_fn=env_fn,
+#         create_env_kwargs={"seed": seed},
+#         policy=policy,
+#         frames_per_batch=200 * num_env,
+#         max_frames_per_traj=2000,
+#         total_frames=20000,
+#         device="cpu",
+#         reset_when_done=False,
+#     )
+#     for _, d in enumerate(collector):  # noqa
+#         break
+#
+#     assert (d["done"].sum(-2) >= 1).all()
+#     assert torch.unique(d["collector", "traj_ids"], dim=-1).shape[-1] == 1
+#
+#     del collector


Left out comments?

We deprecated that feature but I wanted to make sure that we were not going to integrate it back so I kept the commented code.
It'll be removed from the code base before the next release.
Thanks for spotting this!

init

5fcb42c

facebook-github-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 Aug 11, 2023

vmoens added the performance Performance issue or suggestion for improvement label Aug 11, 2023

vmoens added 2 commits August 11, 2023 05:58

amend

871589e

amend

858da29

vmoens added 2 commits August 31, 2023 08:31

Merge branch 'main' into event_mp

473b200

# Conflicts: # torchrl/collectors/collectors.py # torchrl/envs/common.py # torchrl/envs/transforms/transforms.py # torchrl/envs/utils.py # torchrl/envs/vec_env.py

amend

871bdc7

vmoens changed the title ~~[PoC] Faster envs (2)~~ [Refactor] Faster envs (2) Aug 31, 2023

vmoens marked this pull request as ready for review August 31, 2023 12:40

amend

bb352e5

matteobettini reviewed Aug 31, 2023

View reviewed changes

amend

28cb428

matteobettini reviewed Aug 31, 2023

View reviewed changes

matteobettini approved these changes Aug 31, 2023

View reviewed changes

amend

26b53bc

amend

77872a6

vmoens added 9 commits August 31, 2023 11:31

amend

c7ed82e

lint

fd089d1

fixes

d6f304a

amend

df0210e

amend

78a06d5

amend

45615c3

lint

76c5805

amend

9f25da3

amend

3baccdb

amend

be3ccfb

vmoens merged commit e3b3879 into main Sep 1, 2023
49 of 55 checks passed

vmoens deleted the event_mp branch September 1, 2023 09:09

skandermoalla reviewed Sep 1, 2023

View reviewed changes

vmoens mentioned this pull request Sep 5, 2023

[PoC] Faster envs #1448

Closed

vmoens added a commit to hyerra/rl that referenced this pull request Oct 10, 2023

[Refactor] Faster envs (2) (pytorch#1457)

7254f83

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[Refactor] Faster envs (2) #1457

[Refactor] Faster envs (2) #1457

vmoens commented Aug 11, 2023 •

edited

Loading

github-actions bot commented Aug 11, 2023 •

edited

Loading

matteobettini Aug 31, 2023

vmoens Aug 31, 2023 •

edited

Loading

matteobettini Aug 31, 2023

matteobettini Aug 31, 2023

vmoens Aug 31, 2023

matteobettini Aug 31, 2023

matteobettini Aug 31, 2023

vmoens Aug 31, 2023

matteobettini left a comment

matteobettini Aug 31, 2023

vmoens commented Aug 31, 2023

matteobettini commented Aug 31, 2023

skandermoalla Sep 1, 2023

vmoens Sep 1, 2023

		@@ -562,6 +568,101 @@ def make_composite_from_td(data):
		return composite


		def _fuse_tensordicts(*tds, excluded, selected=None, total=None):

[Refactor] Faster envs (2) #1457

[Refactor] Faster envs (2) #1457

Conversation

vmoens commented Aug 11, 2023 • edited Loading

github-actions bot commented Aug 11, 2023 • edited Loading

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

Total Benchmarks: 89. Improved: $\large\color{#35bf28}11$. Worsened: $\large\color{#d91a1a}4$.

Choose a reason for hiding this comment

vmoens Aug 31, 2023 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

matteobettini left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

vmoens commented Aug 31, 2023

matteobettini commented Aug 31, 2023

Choose a reason for hiding this comment

Choose a reason for hiding this comment

vmoens commented Aug 11, 2023 •

edited

Loading

github-actions bot commented Aug 11, 2023 •

edited

Loading

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

vmoens Aug 31, 2023 •

edited

Loading