[DTensor] Add unified dtensor_send/dtensor_recv API with Strategy A by vmoens · Pull Request #1640 · pytorch/tensordict

vmoens · 2026-03-06T15:02:37Z

Stack from ghstack (oldest at bottom):

Add dtensor_send() and dtensor_recv() methods to TensorDictBase with:

strategy parameter: "materialize" (A), "redistribute" (B), "optimal" (C), "auto"
transport parameter: "torch_distributed", "ucxx", "auto"
Transport auto-detection based on dst/src type (int -> torch.distributed,
TensorDictPipe -> UCXX)

Strategy A (materialize) implementation:

Sender: materializes DTensors via full_tensor(), sends metadata + full tensors
Receiver: receives full tensors, stores them in the TensorDict

Strategies B and C are stubbed with NotImplementedError (implemented in
subsequent PRs).

Made-with: Cursor

[ghstack-poisoned]

github-actions · 2026-03-06T15:02:45Z

PR Title Label Error

Unknown or invalid prefix [DTensor].

Current title: [DTensor] Add unified dtensor_send/dtensor_recv API with Strategy A

Supported Prefixes

Your PR title must start with exactly one of these prefixes (case-insensitive):

Prefix	Label Applied	Example
`[BugFix]` or `[Fix]`	bug	`[BugFix] Fix memory leak in TensorDict`
`[Feature]`	Feature	`[Feature] Add new storage backend`
`[Doc]` or `[Docs]`	documentation	`[Doc] Update installation guide`
`[Refactor]`	Refactor	`[Refactor] Clean up module imports`
`[CI]`	CI	`[CI] Fix workflow permissions`
`[Test]` or `[Tests]`	Test	`[Test] Add unit tests for nn module`
`[Compile]`	Compile	`[Compile] Fix torch.compile issue`
`[Performance]` or `[Perf]`	Performance	`[Perf] Optimize tensor operations`
`[Deprecation]`	Deprecation	`[Deprecation] Mark old function`
`[Setup]`	setup	`[Setup] Update build configuration`
`[Distributed]` or `[Dist]`	Distributed	`[Distributed] Add scatter collective`
`[Benchmark]` or `[Bench]`	Benchmarks	`[Benchmark] Add compile benchmark`
`[Typing]` or `[Type]`	Typing	`[Typing] Add type stubs`
`[BC-breaking]` or `[BC]`	BC-breaking	`[BC-breaking] Remove deprecated API`
`[Formatting]` or `[Format]`	Formatting	`[Format] Fix code style`
`[Quality]`	Quality	`[Quality] Improve error messages`

Note: Matching is case-insensitive. Common variations (singular/plural) are supported.

github-actions · 2026-03-06T15:02:56Z

PR Title Label Error

Unknown or invalid prefix [DTensor].

Current title: [DTensor] Add unified dtensor_send/dtensor_recv API with Strategy A

Supported Prefixes

Your PR title must start with exactly one of these prefixes (case-insensitive):

Prefix	Label Applied	Example
`[BugFix]` or `[Fix]`	bug	`[BugFix] Fix memory leak in TensorDict`
`[Feature]`	Feature	`[Feature] Add new storage backend`
`[Doc]` or `[Docs]`	documentation	`[Doc] Update installation guide`
`[Refactor]`	Refactor	`[Refactor] Clean up module imports`
`[CI]`	CI	`[CI] Fix workflow permissions`
`[Test]` or `[Tests]`	Test	`[Test] Add unit tests for nn module`
`[Compile]`	Compile	`[Compile] Fix torch.compile issue`
`[Performance]` or `[Perf]`	Performance	`[Perf] Optimize tensor operations`
`[Deprecation]`	Deprecation	`[Deprecation] Mark old function`
`[Setup]`	setup	`[Setup] Update build configuration`
`[Distributed]` or `[Dist]`	Distributed	`[Distributed] Add scatter collective`
`[Benchmark]` or `[Bench]`	Benchmarks	`[Benchmark] Add compile benchmark`
`[Typing]` or `[Type]`	Typing	`[Typing] Add type stubs`
`[BC-breaking]` or `[BC]`	BC-breaking	`[BC-breaking] Remove deprecated API`
`[Formatting]` or `[Format]`	Formatting	`[Format] Fix code style`
`[Quality]`	Quality	`[Quality] Improve error messages`

Note: Matching is case-insensitive. Common variations (singular/plural) are supported.

github-actions · 2026-03-06T15:20:38Z

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

Total Benchmarks: 261. Improved: $\large\color{#35bf28}35$. Worsened: $\large\color{#d91a1a}6$.

Expand to view detailed results

Name	Max	Mean	Ops	Ops on Repo `HEAD`	Change
test_plain_set_nested	43.9500μs	14.3876μs	69.5044 KOps/s	70.4946 KOps/s	$\color{#d91a1a}-1.40\%$
test_plain_set_stack_nested	37.2910μs	14.7438μs	67.8250 KOps/s	68.7282 KOps/s	$\color{#d91a1a}-1.31\%$
test_plain_set_nested_inplace	42.4910μs	16.0103μs	62.4597 KOps/s	62.2523 KOps/s	$\color{#35bf28}+0.33\%$
test_plain_set_stack_nested_inplace	62.5210μs	15.8845μs	62.9544 KOps/s	62.4856 KOps/s	$\color{#35bf28}+0.75\%$
test_items	27.0910μs	5.4805μs	182.4636 KOps/s	182.0408 KOps/s	$\color{#35bf28}+0.23\%$
test_items_nested	0.5698ms	0.4462ms	2.2409 KOps/s	2.2634 KOps/s	$\color{#d91a1a}-0.99\%$
test_items_nested_locked	0.5638ms	0.4518ms	2.2134 KOps/s	2.2362 KOps/s	$\color{#d91a1a}-1.02\%$
test_items_nested_leaf	0.1665ms	91.3404μs	10.9481 KOps/s	10.8012 KOps/s	$\color{#35bf28}+1.36\%$
test_items_stack_nested	0.4805ms	0.4472ms	2.2359 KOps/s	2.2537 KOps/s	$\color{#d91a1a}-0.79\%$
test_items_stack_nested_leaf	0.1301ms	93.7867μs	10.6625 KOps/s	10.6662 KOps/s	$\color{#d91a1a}-0.03\%$
test_items_stack_nested_locked	0.5176ms	0.4493ms	2.2258 KOps/s	2.2304 KOps/s	$\color{#d91a1a}-0.21\%$
test_keys	33.9100μs	4.1361μs	241.7738 KOps/s	246.1273 KOps/s	$\color{#d91a1a}-1.77\%$
test_keys_nested	0.1937ms	0.1276ms	7.8350 KOps/s	7.8925 KOps/s	$\color{#d91a1a}-0.73\%$
test_keys_nested_locked	2.1438ms	0.1374ms	7.2793 KOps/s	7.4172 KOps/s	$\color{#d91a1a}-1.86\%$
test_keys_nested_leaf	0.1678ms	0.1189ms	8.4132 KOps/s	8.5649 KOps/s	$\color{#d91a1a}-1.77\%$
test_keys_stack_nested	0.2127ms	0.1292ms	7.7417 KOps/s	7.8971 KOps/s	$\color{#d91a1a}-1.97\%$
test_keys_stack_nested_leaf	0.1886ms	0.1189ms	8.4098 KOps/s	8.5845 KOps/s	$\color{#d91a1a}-2.03\%$
test_keys_stack_nested_locked	0.1630ms	0.1380ms	7.2474 KOps/s	7.4417 KOps/s	$\color{#d91a1a}-2.61\%$
test_values	5.4782μs	1.0093μs	990.8121 KOps/s	999.7890 KOps/s	$\color{#d91a1a}-0.90\%$
test_values_nested	92.6810μs	51.7589μs	19.3204 KOps/s	19.5699 KOps/s	$\color{#d91a1a}-1.28\%$
test_values_nested_locked	87.6210μs	55.2306μs	18.1059 KOps/s	18.5448 KOps/s	$\color{#d91a1a}-2.37\%$
test_values_nested_leaf	89.3110μs	59.3172μs	16.8585 KOps/s	17.2467 KOps/s	$\color{#d91a1a}-2.25\%$
test_values_stack_nested	81.6310μs	51.8681μs	19.2797 KOps/s	19.5830 KOps/s	$\color{#d91a1a}-1.55\%$
test_values_stack_nested_leaf	86.8820μs	59.1589μs	16.9036 KOps/s	17.2172 KOps/s	$\color{#d91a1a}-1.82\%$
test_values_stack_nested_locked	0.1030ms	55.7281μs	17.9443 KOps/s	18.3803 KOps/s	$\color{#d91a1a}-2.37\%$
test_membership	9.5518μs	0.7978μs	1.2534 MOps/s	1.2521 MOps/s	$\color{#35bf28}+0.11\%$
test_membership_nested	33.3110μs	2.7173μs	368.0186 KOps/s	370.0032 KOps/s	$\color{#d91a1a}-0.54\%$
test_membership_nested_leaf	15.4150μs	2.6389μs	378.9392 KOps/s	362.2450 KOps/s	$\color{#35bf28}+4.61\%$
test_membership_stacked_nested	33.6400μs	2.7621μs	362.0380 KOps/s	369.0774 KOps/s	$\color{#d91a1a}-1.91\%$
test_membership_stacked_nested_leaf	39.8610μs	2.7499μs	363.6506 KOps/s	364.1836 KOps/s	$\color{#d91a1a}-0.15\%$
test_membership_nested_last	35.8010μs	4.1316μs	242.0356 KOps/s	244.2148 KOps/s	$\color{#d91a1a}-0.89\%$
test_membership_nested_leaf_last	29.6210μs	4.1077μs	243.4443 KOps/s	244.9041 KOps/s	$\color{#d91a1a}-0.60\%$
test_membership_stacked_nested_last	73.4710μs	4.0802μs	245.0843 KOps/s	244.1228 KOps/s	$\color{#35bf28}+0.39\%$
test_membership_stacked_nested_leaf_last	36.2310μs	4.1006μs	243.8696 KOps/s	246.6123 KOps/s	$\color{#d91a1a}-1.11\%$
test_nested_getleaf	51.2110μs	20.6709μs	48.3771 KOps/s	48.9640 KOps/s	$\color{#d91a1a}-1.20\%$
test_nested_get	61.8610μs	19.2225μs	52.0224 KOps/s	51.7182 KOps/s	$\color{#35bf28}+0.59\%$
test_stacked_getleaf	48.6210μs	20.2948μs	49.2738 KOps/s	48.1768 KOps/s	$\color{#35bf28}+2.28\%$
test_stacked_get	50.1410μs	19.4363μs	51.4502 KOps/s	51.0424 KOps/s	$\color{#35bf28}+0.80\%$
test_nested_getitemleaf	85.2010μs	21.0590μs	47.4855 KOps/s	46.8230 KOps/s	$\color{#35bf28}+1.42\%$
test_nested_getitem	46.9510μs	19.7722μs	50.5760 KOps/s	49.9813 KOps/s	$\color{#35bf28}+1.19\%$
test_stacked_getitemleaf	48.7810μs	20.7738μs	48.1374 KOps/s	47.2947 KOps/s	$\color{#35bf28}+1.78\%$
test_stacked_getitem	46.9910μs	19.9313μs	50.1724 KOps/s	49.4553 KOps/s	$\color{#35bf28}+1.45\%$
test_lock_nested	0.5208ms	0.4586ms	2.1808 KOps/s	2.1894 KOps/s	$\color{#d91a1a}-0.39\%$
test_lock_stack_nested	0.5230ms	0.4622ms	2.1635 KOps/s	2.1527 KOps/s	$\color{#35bf28}+0.50\%$
test_unlock_nested	0.5241ms	0.3726ms	2.6842 KOps/s	2.6912 KOps/s	$\color{#d91a1a}-0.26\%$
test_unlock_stack_nested	0.4330ms	0.3743ms	2.6717 KOps/s	2.6395 KOps/s	$\color{#35bf28}+1.22\%$
test_flatten_speed	0.1596ms	0.1152ms	8.6834 KOps/s	8.6092 KOps/s	$\color{#35bf28}+0.86\%$
test_unflatten_speed	0.6151ms	0.5436ms	1.8396 KOps/s	1.7923 KOps/s	$\color{#35bf28}+2.64\%$
test_common_ops	0.8383ms	0.6969ms	1.4349 KOps/s	1.4781 KOps/s	$\color{#d91a1a}-2.92\%$
test_creation	69.0610μs	2.9521μs	338.7461 KOps/s	340.8953 KOps/s	$\color{#d91a1a}-0.63\%$
test_creation_empty	32.2200μs	6.5272μs	153.2040 KOps/s	151.5111 KOps/s	$\color{#35bf28}+1.12\%$
test_creation_nested_1	57.3110μs	11.0053μs	90.8650 KOps/s	91.0301 KOps/s	$\color{#d91a1a}-0.18\%$
test_creation_nested_2	38.6110μs	12.5652μs	79.5851 KOps/s	78.9655 KOps/s	$\color{#35bf28}+0.78\%$
test_creation_many_keys[10]	48.3310μs	19.5516μs	51.1466 KOps/s	50.5929 KOps/s	$\color{#35bf28}+1.09\%$
test_creation_many_keys[50]	0.1489ms	83.5534μs	11.9684 KOps/s	11.6980 KOps/s	$\color{#35bf28}+2.31\%$
test_creation_many_keys[100]	0.2267ms	0.1642ms	6.0917 KOps/s	5.9790 KOps/s	$\color{#35bf28}+1.88\%$
test_creation_nested_many_keys[10]	75.4310μs	41.7514μs	23.9513 KOps/s	23.3619 KOps/s	$\color{#35bf28}+2.52\%$
test_creation_nested_many_keys[50]	0.2343ms	0.1710ms	5.8469 KOps/s	5.6919 KOps/s	$\color{#35bf28}+2.72\%$
test_clone	42.6010μs	12.7681μs	78.3202 KOps/s	75.4074 KOps/s	$\color{#35bf28}+3.86\%$
test_getitem[int]	1.7072ms	14.5209μs	68.8664 KOps/s	61.9057 KOps/s	$\textbf{\color{#35bf28}+11.24\%}$
test_getitem[slice_int]	0.1372ms	24.5522μs	40.7296 KOps/s	43.3453 KOps/s	$\textbf{\color{#d91a1a}-6.03\%}$
test_getitem[range]	0.1762ms	65.0297μs	15.3776 KOps/s	15.2469 KOps/s	$\color{#35bf28}+0.86\%$
test_getitem[tuple]	0.1455ms	23.1887μs	43.1245 KOps/s	41.2563 KOps/s	$\color{#35bf28}+4.53\%$
test_getitem[list]	0.1775ms	59.6661μs	16.7599 KOps/s	16.4677 KOps/s	$\color{#35bf28}+1.77\%$
test_setitem_dim[int]	59.3110μs	27.4442μs	36.4376 KOps/s	37.4142 KOps/s	$\color{#d91a1a}-2.61\%$
test_setitem_dim[slice_int]	67.2610μs	43.0431μs	23.2325 KOps/s	23.0178 KOps/s	$\color{#35bf28}+0.93\%$
test_setitem_dim[range]	0.1225ms	97.6786μs	10.2377 KOps/s	10.3676 KOps/s	$\color{#d91a1a}-1.25\%$
test_setitem_dim[tuple]	70.2410μs	38.7339μs	25.8172 KOps/s	24.1116 KOps/s	$\textbf{\color{#35bf28}+7.07\%}$
test_setitem	49.9000μs	17.6363μs	56.7012 KOps/s	53.7185 KOps/s	$\textbf{\color{#35bf28}+5.55\%}$
test_set	58.3110μs	17.0492μs	58.6539 KOps/s	56.4200 KOps/s	$\color{#35bf28}+3.96\%$
test_set_shared	0.4909ms	0.2083ms	4.8016 KOps/s	4.8322 KOps/s	$\color{#d91a1a}-0.63\%$
test_update	0.3316ms	22.2901μs	44.8629 KOps/s	44.2841 KOps/s	$\color{#35bf28}+1.31\%$
test_update_nested	72.7310μs	32.0420μs	31.2091 KOps/s	29.3695 KOps/s	$\textbf{\color{#35bf28}+6.26\%}$
test_update__nested	0.4756ms	33.7405μs	29.6380 KOps/s	28.3864 KOps/s	$\color{#35bf28}+4.41\%$
test_set_nested	58.7100μs	18.8534μs	53.0408 KOps/s	49.6968 KOps/s	$\textbf{\color{#35bf28}+6.73\%}$
test_set_nested_new	72.3310μs	23.6633μs	42.2595 KOps/s	39.3543 KOps/s	$\textbf{\color{#35bf28}+7.38\%}$
test_select	80.8010μs	39.8606μs	25.0874 KOps/s	23.7536 KOps/s	$\textbf{\color{#35bf28}+5.62\%}$
test_select_nested	0.1068ms	70.3085μs	14.2230 KOps/s	14.1926 KOps/s	$\color{#35bf28}+0.21\%$
test_exclude_nested	0.1354ms	87.1797μs	11.4706 KOps/s	11.7028 KOps/s	$\color{#d91a1a}-1.98\%$
test_empty[True]	0.4469ms	0.3877ms	2.5790 KOps/s	2.6127 KOps/s	$\color{#d91a1a}-1.29\%$
test_empty[False]	8.2325μs	1.2523μs	798.5551 KOps/s	808.4200 KOps/s	$\color{#d91a1a}-1.22\%$
test_to	0.1023ms	71.0523μs	14.0741 KOps/s	14.0506 KOps/s	$\color{#35bf28}+0.17\%$
test_to_nonblocking	0.1141ms	65.4895μs	15.2696 KOps/s	16.1060 KOps/s	$\textbf{\color{#d91a1a}-5.19\%}$
test_unbind_speed	0.3637ms	0.3222ms	3.1039 KOps/s	3.1749 KOps/s	$\color{#d91a1a}-2.24\%$
test_unbind_speed_stack0	0.3957ms	0.3172ms	3.1522 KOps/s	3.1636 KOps/s	$\color{#d91a1a}-0.36\%$
test_unbind_speed_stack1	0.1049s	0.8889ms	1.1250 KOps/s	1.2313 KOps/s	$\textbf{\color{#d91a1a}-8.64\%}$
test_split	1.1472ms	1.0897ms	917.6620 Ops/s	810.0656 Ops/s	$\textbf{\color{#35bf28}+13.28\%}$
test_chunk	0.1055s	1.1590ms	862.7786 Ops/s	957.4254 Ops/s	$\textbf{\color{#d91a1a}-9.89\%}$
test_to_cpu_blocking	28.5776ms	28.1296ms	35.5497 Ops/s	53.3079 Ops/s	$\textbf{\color{#d91a1a}-33.31\%}$
test_to_cpu_global_sync	11.2611ms	11.1424ms	89.7471 Ops/s	81.4204 Ops/s	$\textbf{\color{#35bf28}+10.23\%}$
test_to_cpu_event_sync	12.3055ms	12.0195ms	83.1985 Ops/s	83.9742 Ops/s	$\color{#d91a1a}-0.92\%$
test_to_cpu_default	12.3259ms	12.0569ms	82.9402 Ops/s	83.9019 Ops/s	$\color{#d91a1a}-1.15\%$
test_consolidate[False-None]	4.0514ms	3.9607ms	252.4792 Ops/s	224.2915 Ops/s	$\textbf{\color{#35bf28}+12.57\%}$
test_consolidate[default-None]	2.2210ms	1.9313ms	517.7986 Ops/s	494.3408 Ops/s	$\color{#35bf28}+4.75\%$
test_consolidate[reduce-overhead-None]	1.9435ms	1.8660ms	535.8932 Ops/s	515.3975 Ops/s	$\color{#35bf28}+3.98\%$
test_consolidate_njt[False-None]	8.4550ms	8.2501ms	121.2106 Ops/s	120.0504 Ops/s	$\color{#35bf28}+0.97\%$
test_to[False-False-None]	2.1710ms	2.0248ms	493.8807 Ops/s	487.9463 Ops/s	$\color{#35bf28}+1.22\%$
test_to[True-False-None]	2.0172ms	1.8714ms	534.3733 Ops/s	530.1603 Ops/s	$\color{#35bf28}+0.79\%$
test_to[within-False-None]	6.2363ms	5.9472ms	168.1465 Ops/s	165.9848 Ops/s	$\color{#35bf28}+1.30\%$
test_to[True-default-None]	9.0054ms	8.7144ms	114.7521 Ops/s	112.5332 Ops/s	$\color{#35bf28}+1.97\%$
test_to_njt[False-False-None]	8.5388ms	8.2590ms	121.0807 Ops/s	120.3476 Ops/s	$\color{#35bf28}+0.61\%$
test_to_njt[True-False-None]	6.8938ms	6.7835ms	147.4159 Ops/s	146.6121 Ops/s	$\color{#35bf28}+0.55\%$
test_to_njt[within-False-None]	15.7899ms	15.2026ms	65.7783 Ops/s	65.4721 Ops/s	$\color{#35bf28}+0.47\%$
test_creation[device0]	0.3528ms	0.1101ms	9.0791 KOps/s	8.8788 KOps/s	$\color{#35bf28}+2.26\%$
test_creation_from_tensor	0.3579ms	0.1097ms	9.1169 KOps/s	8.9933 KOps/s	$\color{#35bf28}+1.37\%$
test_add_one[memmap_tensor0]	0.2790ms	6.3189μs	158.2557 KOps/s	156.5768 KOps/s	$\color{#35bf28}+1.07\%$
test_contiguous[memmap_tensor0]	15.2700μs	0.6077μs	1.6456 MOps/s	2.2697 MOps/s	$\textbf{\color{#d91a1a}-27.50\%}$
test_stack[memmap_tensor0]	32.8500μs	4.4642μs	224.0060 KOps/s	219.1616 KOps/s	$\color{#35bf28}+2.21\%$
test_memmaptd_index	1.0126ms	0.2644ms	3.7818 KOps/s	3.8193 KOps/s	$\color{#d91a1a}-0.98\%$
test_memmaptd_index_astensor	0.5162ms	0.3625ms	2.7590 KOps/s	2.7748 KOps/s	$\color{#d91a1a}-0.57\%$
test_memmaptd_index_op	0.8939ms	0.5988ms	1.6701 KOps/s	1.6556 KOps/s	$\color{#35bf28}+0.88\%$
test_serialize_model	0.1392s	0.1368s	7.3108 Ops/s	7.3416 Ops/s	$\color{#d91a1a}-0.42\%$
test_serialize_model_pickle	1.3619s	1.2125s	0.8248 Ops/s	0.8262 Ops/s	$\color{#d91a1a}-0.17\%$
test_serialize_weights	0.1387s	0.1367s	7.3155 Ops/s	7.3918 Ops/s	$\color{#d91a1a}-1.03\%$
test_serialize_weights_returnearly	0.3978s	85.8888ms	11.6430 Ops/s	6.0512 Ops/s	$\textbf{\color{#35bf28}+92.41\%}$
test_serialize_weights_pickle	1.3698s	1.2141s	0.8236 Ops/s	0.8224 Ops/s	$\color{#35bf28}+0.15\%$
test_reshape_pytree	0.2161ms	30.9432μs	32.3173 KOps/s	30.2491 KOps/s	$\textbf{\color{#35bf28}+6.84\%}$
test_reshape_td	90.2720μs	43.0876μs	23.2086 KOps/s	22.6174 KOps/s	$\color{#35bf28}+2.61\%$
test_view_pytree	0.2125ms	30.7738μs	32.4951 KOps/s	30.7558 KOps/s	$\textbf{\color{#35bf28}+5.66\%}$
test_view_td	91.7320μs	51.9014μs	19.2673 KOps/s	19.0076 KOps/s	$\color{#35bf28}+1.37\%$
test_unbind_pytree	0.2328ms	34.6836μs	28.8321 KOps/s	27.2902 KOps/s	$\textbf{\color{#35bf28}+5.65\%}$
test_unbind_td	0.1891ms	47.7887μs	20.9255 KOps/s	20.2093 KOps/s	$\color{#35bf28}+3.54\%$
test_split_pytree	0.2397ms	40.2995μs	24.8142 KOps/s	23.6289 KOps/s	$\textbf{\color{#35bf28}+5.02\%}$
test_split_td	0.1229ms	61.8757μs	16.1614 KOps/s	15.3843 KOps/s	$\textbf{\color{#35bf28}+5.05\%}$
test_add_pytree	0.2274ms	41.9811μs	23.8203 KOps/s	23.6734 KOps/s	$\color{#35bf28}+0.62\%$
test_add_td	99.3020μs	57.7142μs	17.3267 KOps/s	17.4002 KOps/s	$\color{#d91a1a}-0.42\%$
test_compile_add_one_nested[tensordict-compile]	0.1898ms	0.1360ms	7.3548 KOps/s	6.7497 KOps/s	$\textbf{\color{#35bf28}+8.96\%}$
test_compile_add_one_nested[tensordict-eager]	0.4364ms	0.1936ms	5.1656 KOps/s	5.2034 KOps/s	$\color{#d91a1a}-0.73\%$
test_compile_add_one_nested[pytree-compile]	0.1500ms	0.1050ms	9.5233 KOps/s	9.3811 KOps/s	$\color{#35bf28}+1.52\%$
test_compile_add_one_nested[pytree-eager]	0.4301ms	0.1709ms	5.8525 KOps/s	5.8162 KOps/s	$\color{#35bf28}+0.62\%$
test_compile_copy_nested[tensordict-compile]	0.2520ms	9.6702μs	103.4107 KOps/s	97.9012 KOps/s	$\textbf{\color{#35bf28}+5.63\%}$
test_compile_copy_nested[tensordict-eager]	0.1107ms	50.5786μs	19.7712 KOps/s	19.5222 KOps/s	$\color{#35bf28}+1.28\%$
test_compile_copy_nested[pytree-compile]	0.1472ms	9.5602μs	104.6007 KOps/s	105.1866 KOps/s	$\color{#d91a1a}-0.56\%$
test_compile_copy_nested[pytree-eager]	0.4607ms	64.5649μs	15.4883 KOps/s	15.5073 KOps/s	$\color{#d91a1a}-0.12\%$
test_compile_add_one_flat[tensordict-compile]	0.2273ms	0.1735ms	5.7640 KOps/s	5.4699 KOps/s	$\textbf{\color{#35bf28}+5.38\%}$
test_compile_add_one_flat[tensordict-eager]	0.3116ms	0.2753ms	3.6321 KOps/s	3.6114 KOps/s	$\color{#35bf28}+0.57\%$
test_compile_add_one_flat[tensorclass-compile]	0.3064ms	0.1150ms	8.6985 KOps/s	8.5564 KOps/s	$\color{#35bf28}+1.66\%$
test_compile_add_one_flat[tensorclass-eager]	0.1159ms	75.3891μs	13.2645 KOps/s	13.7948 KOps/s	$\color{#d91a1a}-3.84\%$
test_compile_add_one_flat[pytree-compile]	0.2042ms	0.1551ms	6.4481 KOps/s	6.3510 KOps/s	$\color{#35bf28}+1.53\%$
test_compile_add_one_flat[pytree-eager]	0.7669ms	0.5039ms	1.9845 KOps/s	1.9475 KOps/s	$\color{#35bf28}+1.90\%$
test_compile_add_self_flat[tensordict-eager]	0.4848ms	0.3275ms	3.0535 KOps/s	3.0247 KOps/s	$\color{#35bf28}+0.95\%$
test_compile_add_self_flat[tensordict-compile]	0.3082ms	0.1755ms	5.6973 KOps/s	5.4049 KOps/s	$\textbf{\color{#35bf28}+5.41\%}$
test_compile_add_self_flat[tensorclass-eager]	0.1292ms	87.4687μs	11.4327 KOps/s	11.3693 KOps/s	$\color{#35bf28}+0.56\%$
test_compile_add_self_flat[tensorclass-compile]	0.3294ms	0.1169ms	8.5551 KOps/s	7.9955 KOps/s	$\textbf{\color{#35bf28}+7.00\%}$
test_compile_add_self_flat[pytree-eager]	0.6337ms	0.4186ms	2.3889 KOps/s	2.3068 KOps/s	$\color{#35bf28}+3.56\%$
test_compile_add_self_flat[pytree-compile]	0.1887ms	0.1557ms	6.4209 KOps/s	6.2437 KOps/s	$\color{#35bf28}+2.84\%$
test_compile_copy_flat[tensordict-compile]	55.3710μs	13.1493μs	76.0499 KOps/s	73.4540 KOps/s	$\color{#35bf28}+3.53\%$
test_compile_copy_flat[tensordict-eager]	65.5310μs	40.0824μs	24.9486 KOps/s	24.9993 KOps/s	$\color{#d91a1a}-0.20\%$
test_compile_copy_flat[pytree-compile]	0.1190ms	10.5573μs	94.7208 KOps/s	96.3560 KOps/s	$\color{#d91a1a}-1.70\%$
test_compile_copy_flat[pytree-eager]	0.4036ms	51.5072μs	19.4148 KOps/s	19.5658 KOps/s	$\color{#d91a1a}-0.77\%$
test_compile_assign_and_add[tensordict-compile]	1.9706ms	0.1714ms	5.8327 KOps/s	5.5493 KOps/s	$\textbf{\color{#35bf28}+5.11\%}$
test_compile_assign_and_add[tensordict-eager]	3.5348ms	3.3142ms	301.7274 Ops/s	307.5907 Ops/s	$\color{#d91a1a}-1.91\%$
test_compile_assign_and_add[pytree-compile]	1.9179ms	0.1576ms	6.3438 KOps/s	6.0741 KOps/s	$\color{#35bf28}+4.44\%$
test_compile_assign_and_add[pytree-eager]	2.9230ms	2.7507ms	363.5463 Ops/s	368.5598 Ops/s	$\color{#d91a1a}-1.36\%$
test_compile_indexing[tensor-tensordict-compile]	0.2190ms	0.1118ms	8.9473 KOps/s	8.9589 KOps/s	$\color{#d91a1a}-0.13\%$
test_compile_indexing[tensor-tensordict-eager]	0.3156ms	75.6438μs	13.2198 KOps/s	13.6862 KOps/s	$\color{#d91a1a}-3.41\%$
test_compile_indexing[tensor-tensorclass-compile]	0.2139ms	97.7599μs	10.2291 KOps/s	10.3475 KOps/s	$\color{#d91a1a}-1.14\%$
test_compile_indexing[tensor-tensorclass-eager]	0.2501ms	44.7380μs	22.3524 KOps/s	23.3598 KOps/s	$\color{#d91a1a}-4.31\%$
test_compile_indexing[tensor-pytree-compile]	0.1737ms	98.6168μs	10.1403 KOps/s	10.5005 KOps/s	$\color{#d91a1a}-3.43\%$
test_compile_indexing[tensor-pytree-eager]	0.2592ms	44.8896μs	22.2769 KOps/s	23.3753 KOps/s	$\color{#d91a1a}-4.70\%$
test_compile_indexing[slice-tensordict-compile]	0.1975ms	56.6533μs	17.6512 KOps/s	16.4805 KOps/s	$\textbf{\color{#35bf28}+7.10\%}$
test_compile_indexing[slice-tensordict-eager]	0.2221ms	26.9639μs	37.0866 KOps/s	37.9288 KOps/s	$\color{#d91a1a}-2.22\%$
test_compile_indexing[slice-tensorclass-compile]	0.1537ms	44.5785μs	22.4324 KOps/s	21.9284 KOps/s	$\color{#35bf28}+2.30\%$
test_compile_indexing[slice-tensorclass-eager]	0.2568ms	21.2579μs	47.0412 KOps/s	46.3510 KOps/s	$\color{#35bf28}+1.49\%$
test_compile_indexing[slice-pytree-compile]	85.7710μs	45.7084μs	21.8778 KOps/s	22.5172 KOps/s	$\color{#d91a1a}-2.84\%$
test_compile_indexing[slice-pytree-eager]	0.2744ms	21.5654μs	46.3705 KOps/s	46.4178 KOps/s	$\color{#d91a1a}-0.10\%$
test_compile_indexing[int-tensordict-compile]	0.1362ms	57.2840μs	17.4569 KOps/s	17.3533 KOps/s	$\color{#35bf28}+0.60\%$
test_compile_indexing[int-tensordict-eager]	0.2389ms	27.5982μs	36.2342 KOps/s	37.5533 KOps/s	$\color{#d91a1a}-3.51\%$
test_compile_indexing[int-tensorclass-compile]	0.1022ms	44.3722μs	22.5366 KOps/s	21.6460 KOps/s	$\color{#35bf28}+4.11\%$
test_compile_indexing[int-tensorclass-eager]	0.2604ms	21.2694μs	47.0158 KOps/s	46.6142 KOps/s	$\color{#35bf28}+0.86\%$
test_compile_indexing[int-pytree-compile]	94.1510μs	44.7105μs	22.3661 KOps/s	21.9285 KOps/s	$\color{#35bf28}+2.00\%$
test_compile_indexing[int-pytree-eager]	0.2608ms	21.3546μs	46.8282 KOps/s	46.4967 KOps/s	$\color{#35bf28}+0.71\%$
test_compile_replace[single-eager]	0.1095ms	47.1690μs	21.2004 KOps/s	22.1953 KOps/s	$\color{#d91a1a}-4.48\%$
test_compile_replace[single-compile]	0.1714ms	0.1053ms	9.4939 KOps/s	9.4245 KOps/s	$\color{#35bf28}+0.74\%$
test_compile_replace[multi-eager]	0.6229ms	0.5627ms	1.7773 KOps/s	1.8377 KOps/s	$\color{#d91a1a}-3.29\%$
test_compile_replace[multi-compile]	0.2628ms	0.1102ms	9.0756 KOps/s	8.6270 KOps/s	$\textbf{\color{#35bf28}+5.20\%}$
test_compile_tc_getattr_20[eager]	0.2151ms	0.1628ms	6.1411 KOps/s	6.1038 KOps/s	$\color{#35bf28}+0.61\%$
test_compile_tc_getattr_20[compile]	0.3152ms	0.1189ms	8.4077 KOps/s	8.4732 KOps/s	$\color{#d91a1a}-0.77\%$
test_compile_clone_shallow[20-eager]	54.3110μs	18.6009μs	53.7609 KOps/s	53.8095 KOps/s	$\color{#d91a1a}-0.09\%$
test_compile_clone_shallow[20-compile]	67.3010μs	11.0020μs	90.8922 KOps/s	90.5705 KOps/s	$\color{#35bf28}+0.36\%$
test_compile_clone_shallow[40-eager]	90.6810μs	32.8757μs	30.4176 KOps/s	30.7664 KOps/s	$\color{#d91a1a}-1.13\%$
test_compile_clone_shallow[40-compile]	0.1806ms	12.1626μs	82.2191 KOps/s	80.4860 KOps/s	$\color{#35bf28}+2.15\%$
test_compile_clone_shallow[80-eager]	0.1199ms	61.3453μs	16.3012 KOps/s	16.5544 KOps/s	$\color{#d91a1a}-1.53\%$
test_compile_clone_shallow[80-compile]	48.3610μs	15.0108μs	66.6188 KOps/s	67.1884 KOps/s	$\color{#d91a1a}-0.85\%$
test_compile_update_inplace[eager]	0.1069ms	57.7756μs	17.3083 KOps/s	17.3916 KOps/s	$\color{#d91a1a}-0.48\%$
test_compile_update_inplace[compile]	0.2039ms	0.1331ms	7.5106 KOps/s	7.0600 KOps/s	$\textbf{\color{#35bf28}+6.38\%}$
test_mod_add[eager]	88.9220μs	47.2403μs	21.1684 KOps/s	21.3474 KOps/s	$\color{#d91a1a}-0.84\%$
test_mod_add[compile]	0.5193ms	0.1003ms	9.9738 KOps/s	9.2071 KOps/s	$\textbf{\color{#35bf28}+8.33\%}$
test_mod_add[compile-overhead]	0.3762ms	0.1477ms	6.7689 KOps/s	6.7233 KOps/s	$\color{#35bf28}+0.68\%$
test_mod_wrap[eager]	0.3520ms	0.2809ms	3.5601 KOps/s	3.3860 KOps/s	$\textbf{\color{#35bf28}+5.14\%}$
test_mod_wrap[compile]	0.5081ms	0.3405ms	2.9366 KOps/s	2.9176 KOps/s	$\color{#35bf28}+0.65\%$
test_mod_wrap[compile-overhead]	7.3268ms	4.0242ms	248.4955 Ops/s	252.3634 Ops/s	$\color{#d91a1a}-1.53\%$
test_mod_wrap_and_backward[eager]	1.6764ms	1.4742ms	678.3110 Ops/s	668.9375 Ops/s	$\color{#35bf28}+1.40\%$
test_mod_wrap_and_backward[compile]	1.5428ms	1.4097ms	709.3700 Ops/s	705.8707 Ops/s	$\color{#35bf28}+0.50\%$
test_mod_wrap_and_backward[compile-overhead]	1.2367ms	0.8709ms	1.1483 KOps/s	1.1411 KOps/s	$\color{#35bf28}+0.63\%$
test_seq_add[eager]	0.2088ms	0.1473ms	6.7879 KOps/s	6.6645 KOps/s	$\color{#35bf28}+1.85\%$
test_seq_add[compile]	0.2914ms	0.1099ms	9.1029 KOps/s	8.7859 KOps/s	$\color{#35bf28}+3.61\%$
test_seq_add[compile-overhead]	0.2192ms	0.1490ms	6.7126 KOps/s	6.4874 KOps/s	$\color{#35bf28}+3.47\%$
test_seq_wrap[eager]	0.5736ms	0.5052ms	1.9793 KOps/s	1.9879 KOps/s	$\color{#d91a1a}-0.44\%$
test_seq_wrap[compile]	0.4343ms	0.3550ms	2.8173 KOps/s	2.7879 KOps/s	$\color{#35bf28}+1.05\%$
test_seq_wrap[compile-overhead]	0.3265ms	0.2585ms	3.8687 KOps/s	3.8490 KOps/s	$\color{#35bf28}+0.51\%$
test_func_call_runtime[False-eager]	0.8814ms	0.8063ms	1.2402 KOps/s	1.2462 KOps/s	$\color{#d91a1a}-0.48\%$
test_func_call_runtime[False-compile]	0.9837ms	0.8785ms	1.1383 KOps/s	1.1298 KOps/s	$\color{#35bf28}+0.75\%$
test_func_call_runtime[False-compile-overhead]	0.5202ms	0.4437ms	2.2536 KOps/s	2.2269 KOps/s	$\color{#35bf28}+1.20\%$
test_func_call_runtime[True-eager]	1.1815ms	1.0375ms	963.8985 Ops/s	958.1684 Ops/s	$\color{#35bf28}+0.60\%$
test_func_call_runtime[True-compile]	0.9594ms	0.8865ms	1.1281 KOps/s	1.0836 KOps/s	$\color{#35bf28}+4.11\%$
test_func_call_runtime[True-compile-overhead]	0.5144ms	0.4550ms	2.1977 KOps/s	2.1543 KOps/s	$\color{#35bf28}+2.02\%$
test_func_call_cm_runtime[False-eager]	1.1505ms	0.8059ms	1.2409 KOps/s	1.2331 KOps/s	$\color{#35bf28}+0.63\%$
test_func_call_cm_runtime[False-compile]	0.9734ms	0.8788ms	1.1380 KOps/s	1.1258 KOps/s	$\color{#35bf28}+1.08\%$
test_func_call_cm_runtime[False-compile-overhead]	0.5000ms	0.4470ms	2.2373 KOps/s	2.2117 KOps/s	$\color{#35bf28}+1.16\%$
test_func_call_cm_runtime[True-eager]	1.2881ms	1.1767ms	849.8311 Ops/s	831.0437 Ops/s	$\color{#35bf28}+2.26\%$
test_func_call_cm_runtime[True-compile]	1.0643ms	0.9338ms	1.0709 KOps/s	1.0631 KOps/s	$\color{#35bf28}+0.74\%$
test_func_call_cm_runtime[True-compile-overhead]	0.5738ms	0.4873ms	2.0522 KOps/s	2.0066 KOps/s	$\color{#35bf28}+2.27\%$
test_vmap_func_call_cm_runtime[eager]	2.7736ms	2.3063ms	433.6018 Ops/s	429.9375 Ops/s	$\color{#35bf28}+0.85\%$
test_vmap_func_call_cm_runtime[compile]	1.0087ms	0.9400ms	1.0638 KOps/s	1.0470 KOps/s	$\color{#35bf28}+1.61\%$
test_vmap_func_call_cm_runtime[compile-overhead]	0.5340ms	0.4924ms	2.0307 KOps/s	1.9913 KOps/s	$\color{#35bf28}+1.98\%$
test_distributed	2.9237ms	0.1564ms	6.3951 KOps/s	6.5930 KOps/s	$\color{#d91a1a}-3.00\%$
test_tdmodule	0.7404ms	27.4274μs	36.4599 KOps/s	37.1106 KOps/s	$\color{#d91a1a}-1.75\%$
test_tdmodule_dispatch	73.1310μs	43.8796μs	22.7896 KOps/s	22.5539 KOps/s	$\color{#35bf28}+1.05\%$
test_tdseq	45.5310μs	25.9340μs	38.5594 KOps/s	38.1209 KOps/s	$\color{#35bf28}+1.15\%$
test_tdseq_dispatch	79.3520μs	45.7623μs	21.8521 KOps/s	21.7433 KOps/s	$\color{#35bf28}+0.50\%$
test_instantiation_functorch	2.0467ms	1.9741ms	506.5563 Ops/s	502.2256 Ops/s	$\color{#35bf28}+0.86\%$
test_exec_functorch	0.2497ms	0.1781ms	5.6145 KOps/s	5.7718 KOps/s	$\color{#d91a1a}-2.72\%$
test_exec_functional_call	0.1894ms	0.1558ms	6.4171 KOps/s	6.4885 KOps/s	$\color{#d91a1a}-1.10\%$
test_exec_td_decorator	0.4306ms	0.2272ms	4.4011 KOps/s	4.3895 KOps/s	$\color{#35bf28}+0.26\%$
test_vmap_mlp_speed_decorator[True-True]	1.0045ms	0.7997ms	1.2504 KOps/s	1.2338 KOps/s	$\color{#35bf28}+1.35\%$
test_vmap_mlp_speed_decorator[True-False]	0.9742ms	0.7981ms	1.2530 KOps/s	1.2371 KOps/s	$\color{#35bf28}+1.28\%$
test_vmap_mlp_speed_decorator[False-True]	0.8865ms	0.6912ms	1.4468 KOps/s	1.4186 KOps/s	$\color{#35bf28}+1.99\%$
test_vmap_mlp_speed_decorator[False-False]	0.8765ms	0.6930ms	1.4431 KOps/s	1.3940 KOps/s	$\color{#35bf28}+3.52\%$
test_vmap_transformer_speed_decorator[True-True]	20.1729ms	20.0469ms	49.8831 Ops/s	49.4307 Ops/s	$\color{#35bf28}+0.92\%$
test_vmap_transformer_speed_decorator[True-False]	20.2045ms	20.0664ms	49.8345 Ops/s	49.3949 Ops/s	$\color{#35bf28}+0.89\%$
test_vmap_transformer_speed_decorator[False-True]	20.5757ms	19.8974ms	50.2577 Ops/s	49.8765 Ops/s	$\color{#35bf28}+0.76\%$
test_vmap_transformer_speed_decorator[False-False]	20.3919ms	19.8943ms	50.2657 Ops/s	49.8832 Ops/s	$\color{#35bf28}+0.77\%$
test_to_module_speed[True]	1.9485ms	1.4091ms	709.6781 Ops/s	700.2111 Ops/s	$\color{#35bf28}+1.35\%$
test_to_module_speed[False]	1.8542ms	1.3738ms	727.9298 Ops/s	714.0486 Ops/s	$\color{#35bf28}+1.94\%$
test_tc_init	70.8710μs	42.7568μs	23.3881 KOps/s	23.0945 KOps/s	$\color{#35bf28}+1.27\%$
test_tc_init_tensor_only	33.9910μs	9.2502μs	108.1057 KOps/s	108.8304 KOps/s	$\color{#d91a1a}-0.67\%$
test_tc_init_nested	0.3643ms	85.3820μs	11.7121 KOps/s	11.5583 KOps/s	$\color{#35bf28}+1.33\%$
test_tc_init_many_fields	78.5810μs	15.7123μs	63.6446 KOps/s	64.2410 KOps/s	$\color{#d91a1a}-0.93\%$
test_tc_first_layer_tensor	27.4810μs	1.7198μs	581.4536 KOps/s	584.9094 KOps/s	$\color{#d91a1a}-0.59\%$
test_tc_first_layer_tensor_only	2.2700μs	0.3818μs	2.6192 MOps/s	2.5742 MOps/s	$\color{#35bf28}+1.75\%$
test_tc_first_layer_tensor_set	39.8210μs	3.6915μs	270.8908 KOps/s	268.6168 KOps/s	$\color{#35bf28}+0.85\%$
test_tc_first_layer_tensor_only_set	24.6710μs	3.0889μs	323.7362 KOps/s	319.5773 KOps/s	$\color{#35bf28}+1.30\%$
test_tc_first_layer_nontensor	30.4800μs	5.8585μs	170.6914 KOps/s	172.3147 KOps/s	$\color{#d91a1a}-0.94\%$
test_tc_second_layer_tensor	45.7010μs	4.1549μs	240.6797 KOps/s	235.7931 KOps/s	$\color{#35bf28}+2.07\%$
test_tc_second_layer_nontensor	31.2710μs	8.3184μs	120.2154 KOps/s	120.7440 KOps/s	$\color{#d91a1a}-0.44\%$
test_unbind	0.2639s	17.0437ms	58.6728 Ops/s	55.1392 Ops/s	$\textbf{\color{#35bf28}+6.41\%}$
test_full_like	5.0348ms	4.4048ms	227.0243 Ops/s	59.3602 Ops/s	$\textbf{\color{#35bf28}+282.45\%}$
test_zeros_like	4.9890ms	4.3868ms	227.9555 Ops/s	59.4565 Ops/s	$\textbf{\color{#35bf28}+283.40\%}$
test_ones_like	4.8867ms	4.3952ms	227.5231 Ops/s	59.3844 Ops/s	$\textbf{\color{#35bf28}+283.14\%}$
test_clone	6.8351ms	6.5965ms	151.5948 Ops/s	55.4774 Ops/s	$\textbf{\color{#35bf28}+173.26\%}$
test_squeeze	0.1767ms	13.5540μs	73.7791 KOps/s	72.1011 KOps/s	$\color{#35bf28}+2.33\%$
test_unsqueeze	0.1803ms	0.1093ms	9.1486 KOps/s	8.8205 KOps/s	$\color{#35bf28}+3.72\%$
test_split	0.3444ms	0.1775ms	5.6348 KOps/s	5.4029 KOps/s	$\color{#35bf28}+4.29\%$
test_permute	0.2627ms	0.2003ms	4.9934 KOps/s	4.8037 KOps/s	$\color{#35bf28}+3.95\%$
test_stack	35.5903ms	35.2742ms	28.3493 Ops/s	19.1189 Ops/s	$\textbf{\color{#35bf28}+48.28\%}$
test_cat	35.5180ms	35.1998ms	28.4093 Ops/s	19.1711 Ops/s	$\textbf{\color{#35bf28}+48.19\%}$
test_sequential_tensordict	0.2649ms	0.2079ms	4.8103 KOps/s	4.6210 KOps/s	$\color{#35bf28}+4.10\%$
test_sequential_graph_module	0.1610ms	0.1143ms	8.7508 KOps/s	8.2879 KOps/s	$\textbf{\color{#35bf28}+5.59\%}$
test_nested_tensordict	0.3727ms	0.2780ms	3.5972 KOps/s	3.4827 KOps/s	$\color{#35bf28}+3.29\%$
test_nested_graph_module	0.2242ms	0.1318ms	7.5857 KOps/s	7.6489 KOps/s	$\color{#d91a1a}-0.83\%$

github-actions · 2026-03-06T15:20:49Z

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

Total Benchmarks: 261. Improved: $\large\color{#35bf28}24$. Worsened: $\large\color{#d91a1a}7$.

Expand to view detailed results

Name	Max	Mean	Ops	Ops on Repo `HEAD`	Change
test_plain_set_nested	29.1210μs	14.9734μs	66.7849 KOps/s	66.7369 KOps/s	$\color{#35bf28}+0.07\%$
test_plain_set_stack_nested	31.9500μs	15.2942μs	65.3845 KOps/s	66.2761 KOps/s	$\color{#d91a1a}-1.35\%$
test_plain_set_nested_inplace	41.0610μs	16.7826μs	59.5855 KOps/s	59.1133 KOps/s	$\color{#35bf28}+0.80\%$
test_plain_set_stack_nested_inplace	45.2710μs	16.8056μs	59.5041 KOps/s	59.7575 KOps/s	$\color{#d91a1a}-0.42\%$
test_items	81.1620μs	5.9896μs	166.9553 KOps/s	166.2827 KOps/s	$\color{#35bf28}+0.40\%$
test_items_nested	0.5164ms	0.4633ms	2.1585 KOps/s	2.1332 KOps/s	$\color{#35bf28}+1.18\%$
test_items_nested_locked	0.5268ms	0.4702ms	2.1268 KOps/s	2.1316 KOps/s	$\color{#d91a1a}-0.23\%$
test_items_nested_leaf	0.1298ms	99.6165μs	10.0385 KOps/s	10.0889 KOps/s	$\color{#d91a1a}-0.50\%$
test_items_stack_nested	0.4983ms	0.4659ms	2.1466 KOps/s	2.1326 KOps/s	$\color{#35bf28}+0.66\%$
test_items_stack_nested_leaf	0.1860ms	98.6724μs	10.1345 KOps/s	10.0941 KOps/s	$\color{#35bf28}+0.40\%$
test_items_stack_nested_locked	0.5017ms	0.4738ms	2.1108 KOps/s	2.1326 KOps/s	$\color{#d91a1a}-1.02\%$
test_keys	46.5000μs	4.2335μs	236.2131 KOps/s	234.4990 KOps/s	$\color{#35bf28}+0.73\%$
test_keys_nested	0.1745ms	0.1292ms	7.7429 KOps/s	7.7069 KOps/s	$\color{#35bf28}+0.47\%$
test_keys_nested_locked	2.1171ms	0.1389ms	7.1976 KOps/s	7.1915 KOps/s	$\color{#35bf28}+0.09\%$
test_keys_nested_leaf	0.1552ms	0.1207ms	8.2817 KOps/s	8.2495 KOps/s	$\color{#35bf28}+0.39\%$
test_keys_stack_nested	0.1647ms	0.1303ms	7.6745 KOps/s	7.7220 KOps/s	$\color{#d91a1a}-0.61\%$
test_keys_stack_nested_leaf	0.1521ms	0.1208ms	8.2759 KOps/s	8.3017 KOps/s	$\color{#d91a1a}-0.31\%$
test_keys_stack_nested_locked	0.1878ms	0.1380ms	7.2475 KOps/s	7.2418 KOps/s	$\color{#35bf28}+0.08\%$
test_values	6.8760μs	1.0325μs	968.5682 KOps/s	976.4249 KOps/s	$\color{#d91a1a}-0.80\%$
test_values_nested	78.7710μs	52.5351μs	19.0349 KOps/s	19.1824 KOps/s	$\color{#d91a1a}-0.77\%$
test_values_nested_locked	81.5220μs	56.3740μs	17.7387 KOps/s	17.8245 KOps/s	$\color{#d91a1a}-0.48\%$
test_values_nested_leaf	98.6220μs	59.9604μs	16.6777 KOps/s	16.5542 KOps/s	$\color{#35bf28}+0.75\%$
test_values_stack_nested	91.7020μs	52.2525μs	19.1379 KOps/s	18.9917 KOps/s	$\color{#35bf28}+0.77\%$
test_values_stack_nested_leaf	0.1115ms	59.6800μs	16.7560 KOps/s	16.5378 KOps/s	$\color{#35bf28}+1.32\%$
test_values_stack_nested_locked	97.6920μs	55.8526μs	17.9043 KOps/s	18.1336 KOps/s	$\color{#d91a1a}-1.26\%$
test_membership	6.2583μs	0.8464μs	1.1814 MOps/s	1.1733 MOps/s	$\color{#35bf28}+0.69\%$
test_membership_nested	29.9200μs	2.8913μs	345.8608 KOps/s	347.6444 KOps/s	$\color{#d91a1a}-0.51\%$
test_membership_nested_leaf	67.0910μs	2.9148μs	343.0713 KOps/s	346.2628 KOps/s	$\color{#d91a1a}-0.92\%$
test_membership_stacked_nested	23.6400μs	2.9125μs	343.3479 KOps/s	344.1282 KOps/s	$\color{#d91a1a}-0.23\%$
test_membership_stacked_nested_leaf	33.8510μs	2.9153μs	343.0157 KOps/s	347.8675 KOps/s	$\color{#d91a1a}-1.39\%$
test_membership_nested_last	27.5000μs	4.2837μs	233.4443 KOps/s	231.1597 KOps/s	$\color{#35bf28}+0.99\%$
test_membership_nested_leaf_last	24.4800μs	4.3562μs	229.5593 KOps/s	231.7868 KOps/s	$\color{#d91a1a}-0.96\%$
test_membership_stacked_nested_last	28.9400μs	4.3651μs	229.0915 KOps/s	232.2604 KOps/s	$\color{#d91a1a}-1.36\%$
test_membership_stacked_nested_leaf_last	46.9510μs	4.3266μs	231.1276 KOps/s	231.2833 KOps/s	$\color{#d91a1a}-0.07\%$
test_nested_getleaf	90.0410μs	21.8823μs	45.6990 KOps/s	46.3581 KOps/s	$\color{#d91a1a}-1.42\%$
test_nested_get	55.4810μs	20.3703μs	49.0910 KOps/s	49.2792 KOps/s	$\color{#d91a1a}-0.38\%$
test_stacked_getleaf	53.5410μs	21.4981μs	46.5157 KOps/s	46.5158 KOps/s	$-0.00\%$
test_stacked_get	51.9710μs	20.7002μs	48.3088 KOps/s	49.4772 KOps/s	$\color{#d91a1a}-2.36\%$
test_nested_getitemleaf	0.1298ms	21.3850μs	46.7617 KOps/s	45.7059 KOps/s	$\color{#35bf28}+2.31\%$
test_nested_getitem	53.4110μs	20.8258μs	48.0173 KOps/s	47.8463 KOps/s	$\color{#35bf28}+0.36\%$
test_stacked_getitemleaf	46.3010μs	21.8770μs	45.7101 KOps/s	45.6299 KOps/s	$\color{#35bf28}+0.18\%$
test_stacked_getitem	47.0410μs	20.7768μs	48.1306 KOps/s	47.2081 KOps/s	$\color{#35bf28}+1.95\%$
test_lock_nested	0.5883ms	0.4777ms	2.0935 KOps/s	2.0905 KOps/s	$\color{#35bf28}+0.14\%$
test_lock_stack_nested	0.5299ms	0.4797ms	2.0848 KOps/s	2.0637 KOps/s	$\color{#35bf28}+1.02\%$
test_unlock_nested	0.4722ms	0.3896ms	2.5667 KOps/s	2.5609 KOps/s	$\color{#35bf28}+0.23\%$
test_unlock_stack_nested	0.4531ms	0.3902ms	2.5627 KOps/s	2.5267 KOps/s	$\color{#35bf28}+1.43\%$
test_flatten_speed	0.1770ms	0.1237ms	8.0814 KOps/s	8.1201 KOps/s	$\color{#d91a1a}-0.48\%$
test_unflatten_speed	0.6208ms	0.5688ms	1.7580 KOps/s	1.7634 KOps/s	$\color{#d91a1a}-0.30\%$
test_common_ops	0.9383ms	0.6938ms	1.4413 KOps/s	1.4177 KOps/s	$\color{#35bf28}+1.67\%$
test_creation	0.1145ms	3.1748μs	314.9785 KOps/s	318.0065 KOps/s	$\color{#d91a1a}-0.95\%$
test_creation_empty	33.1510μs	7.0011μs	142.8355 KOps/s	143.3830 KOps/s	$\color{#d91a1a}-0.38\%$
test_creation_nested_1	41.4510μs	11.5752μs	86.3919 KOps/s	86.7919 KOps/s	$\color{#d91a1a}-0.46\%$
test_creation_nested_2	40.6810μs	13.2396μs	75.5309 KOps/s	74.9807 KOps/s	$\color{#35bf28}+0.73\%$
test_creation_many_keys[10]	64.2420μs	21.2317μs	47.0993 KOps/s	47.4688 KOps/s	$\color{#d91a1a}-0.78\%$
test_creation_many_keys[50]	0.1366ms	92.2125μs	10.8445 KOps/s	10.9727 KOps/s	$\color{#d91a1a}-1.17\%$
test_creation_many_keys[100]	0.2243ms	0.1809ms	5.5286 KOps/s	5.6273 KOps/s	$\color{#d91a1a}-1.75\%$
test_creation_nested_many_keys[10]	72.9910μs	45.4590μs	21.9979 KOps/s	22.1264 KOps/s	$\color{#d91a1a}-0.58\%$
test_creation_nested_many_keys[50]	0.2183ms	0.1852ms	5.3987 KOps/s	5.4211 KOps/s	$\color{#d91a1a}-0.41\%$
test_clone	42.1500μs	13.1777μs	75.8859 KOps/s	74.1090 KOps/s	$\color{#35bf28}+2.40\%$
test_getitem[int]	1.5547ms	15.0112μs	66.6170 KOps/s	59.8943 KOps/s	$\textbf{\color{#35bf28}+11.22\%}$
test_getitem[slice_int]	0.1390ms	24.1160μs	41.4662 KOps/s	41.1270 KOps/s	$\color{#35bf28}+0.82\%$
test_getitem[range]	0.1697ms	61.8070μs	16.1794 KOps/s	15.0874 KOps/s	$\textbf{\color{#35bf28}+7.24\%}$
test_getitem[tuple]	0.1521ms	23.9193μs	41.8072 KOps/s	41.9820 KOps/s	$\color{#d91a1a}-0.42\%$
test_getitem[list]	0.1790ms	56.6272μs	17.6594 KOps/s	17.1190 KOps/s	$\color{#35bf28}+3.16\%$
test_setitem_dim[int]	48.8310μs	25.1914μs	39.6961 KOps/s	37.8830 KOps/s	$\color{#35bf28}+4.79\%$
test_setitem_dim[slice_int]	62.4910μs	41.7970μs	23.9252 KOps/s	23.2862 KOps/s	$\color{#35bf28}+2.74\%$
test_setitem_dim[range]	0.1225ms	93.5775μs	10.6863 KOps/s	10.6868 KOps/s	$-0.00\%$
test_setitem_dim[tuple]	66.3710μs	38.7307μs	25.8193 KOps/s	25.3527 KOps/s	$\color{#35bf28}+1.84\%$
test_setitem	47.9010μs	17.3828μs	57.5280 KOps/s	55.8101 KOps/s	$\color{#35bf28}+3.08\%$
test_set	40.2510μs	16.7394μs	59.7393 KOps/s	58.5777 KOps/s	$\color{#35bf28}+1.98\%$
test_set_shared	0.6334ms	0.2067ms	4.8389 KOps/s	4.7085 KOps/s	$\color{#35bf28}+2.77\%$
test_update	0.4515ms	21.3619μs	46.8124 KOps/s	45.4338 KOps/s	$\color{#35bf28}+3.03\%$
test_update_nested	69.1210μs	32.7373μs	30.5462 KOps/s	29.9491 KOps/s	$\color{#35bf28}+1.99\%$
test_update__nested	0.4889ms	33.3281μs	30.0047 KOps/s	28.6655 KOps/s	$\color{#35bf28}+4.67\%$
test_set_nested	54.1710μs	18.7751μs	53.2622 KOps/s	52.4980 KOps/s	$\color{#35bf28}+1.46\%$
test_set_nested_new	60.9510μs	23.7880μs	42.0381 KOps/s	41.8450 KOps/s	$\color{#35bf28}+0.46\%$
test_select	73.2810μs	40.0874μs	24.9455 KOps/s	24.7061 KOps/s	$\color{#35bf28}+0.97\%$
test_select_nested	0.1023ms	74.3953μs	13.4417 KOps/s	13.6216 KOps/s	$\color{#d91a1a}-1.32\%$
test_exclude_nested	0.1377ms	91.1171μs	10.9749 KOps/s	11.0470 KOps/s	$\color{#d91a1a}-0.65\%$
test_empty[True]	0.4259ms	0.3977ms	2.5142 KOps/s	2.5177 KOps/s	$\color{#d91a1a}-0.14\%$
test_empty[False]	7.8102μs	1.3197μs	757.7758 KOps/s	773.5999 KOps/s	$\color{#d91a1a}-2.05\%$
test_to	0.1056ms	72.7765μs	13.7407 KOps/s	13.7530 KOps/s	$\color{#d91a1a}-0.09\%$
test_to_nonblocking	0.1131ms	65.3397μs	15.3046 KOps/s	15.1682 KOps/s	$\color{#35bf28}+0.90\%$
test_unbind_speed	0.3717ms	0.3317ms	3.0149 KOps/s	2.9919 KOps/s	$\color{#35bf28}+0.77\%$
test_unbind_speed_stack0	0.4030ms	0.3296ms	3.0337 KOps/s	3.0163 KOps/s	$\color{#35bf28}+0.58\%$
test_unbind_speed_stack1	0.1068s	0.9201ms	1.0869 KOps/s	1.1714 KOps/s	$\textbf{\color{#d91a1a}-7.22\%}$
test_split	1.1984ms	1.1397ms	877.3913 Ops/s	785.5224 Ops/s	$\textbf{\color{#35bf28}+11.70\%}$
test_chunk	0.1072s	1.2137ms	823.9068 Ops/s	925.7417 Ops/s	$\textbf{\color{#d91a1a}-11.00\%}$
test_to_cpu_blocking	18.9162ms	18.6117ms	53.7297 Ops/s	52.6000 Ops/s	$\color{#35bf28}+2.15\%$
test_to_cpu_global_sync	11.3384ms	11.2231ms	89.1017 Ops/s	77.8927 Ops/s	$\textbf{\color{#35bf28}+14.39\%}$
test_to_cpu_event_sync	12.4673ms	12.0596ms	82.9213 Ops/s	80.7280 Ops/s	$\color{#35bf28}+2.72\%$
test_to_cpu_default	12.3148ms	12.0894ms	82.7173 Ops/s	80.6695 Ops/s	$\color{#35bf28}+2.54\%$
test_consolidate[False-None]	4.3301ms	4.1199ms	242.7232 Ops/s	215.0100 Ops/s	$\textbf{\color{#35bf28}+12.89\%}$
test_consolidate[default-None]	2.1291ms	2.0440ms	489.2291 Ops/s	479.3427 Ops/s	$\color{#35bf28}+2.06\%$
test_consolidate[reduce-overhead-None]	2.0426ms	1.9467ms	513.6862 Ops/s	499.8178 Ops/s	$\color{#35bf28}+2.77\%$
test_consolidate_njt[False-None]	8.8481ms	8.4786ms	117.9444 Ops/s	117.4216 Ops/s	$\color{#35bf28}+0.45\%$
test_to[False-False-None]	2.2331ms	2.0537ms	486.9179 Ops/s	473.3487 Ops/s	$\color{#35bf28}+2.87\%$
test_to[True-False-None]	2.2645ms	1.8980ms	526.8760 Ops/s	527.0959 Ops/s	$\color{#d91a1a}-0.04\%$
test_to[within-False-None]	6.3199ms	6.1320ms	163.0782 Ops/s	162.9656 Ops/s	$\color{#35bf28}+0.07\%$
test_to[True-default-None]	8.8644ms	8.6824ms	115.1754 Ops/s	112.8102 Ops/s	$\color{#35bf28}+2.10\%$
test_to_njt[False-False-None]	8.5953ms	8.4503ms	118.3392 Ops/s	116.7291 Ops/s	$\color{#35bf28}+1.38\%$
test_to_njt[True-False-None]	7.0457ms	6.9107ms	144.7037 Ops/s	141.4012 Ops/s	$\color{#35bf28}+2.34\%$
test_to_njt[within-False-None]	15.8740ms	15.6648ms	63.8374 Ops/s	63.1720 Ops/s	$\color{#35bf28}+1.05\%$
test_creation[device0]	0.3914ms	0.1156ms	8.6517 KOps/s	8.3176 KOps/s	$\color{#35bf28}+4.02\%$
test_creation_from_tensor	0.4030ms	0.1125ms	8.8912 KOps/s	8.7836 KOps/s	$\color{#35bf28}+1.23\%$
test_add_one[memmap_tensor0]	0.1383ms	6.4558μs	154.9006 KOps/s	149.6523 KOps/s	$\color{#35bf28}+3.51\%$
test_contiguous[memmap_tensor0]	19.5710μs	0.6782μs	1.4745 MOps/s	2.1320 MOps/s	$\textbf{\color{#d91a1a}-30.84\%}$
test_stack[memmap_tensor0]	31.3800μs	4.6482μs	215.1354 KOps/s	217.6022 KOps/s	$\color{#d91a1a}-1.13\%$
test_memmaptd_index	1.0481ms	0.2712ms	3.6868 KOps/s	3.6748 KOps/s	$\color{#35bf28}+0.33\%$
test_memmaptd_index_astensor	0.5374ms	0.3739ms	2.6742 KOps/s	2.6754 KOps/s	$\color{#d91a1a}-0.05\%$
test_memmaptd_index_op	0.7635ms	0.6200ms	1.6128 KOps/s	1.5992 KOps/s	$\color{#35bf28}+0.85\%$
test_serialize_model	0.1401s	0.1373s	7.2846 Ops/s	7.2912 Ops/s	$\color{#d91a1a}-0.09\%$
test_serialize_model_pickle	1.3493s	1.2107s	0.8260 Ops/s	0.8261 Ops/s	$\color{#d91a1a}-0.02\%$
test_serialize_weights	0.1376s	0.1359s	7.3583 Ops/s	7.3249 Ops/s	$\color{#35bf28}+0.46\%$
test_serialize_weights_returnearly	0.4476s	94.9856ms	10.5279 Ops/s	14.8097 Ops/s	$\textbf{\color{#d91a1a}-28.91\%}$
test_serialize_weights_pickle	1.3746s	1.1911s	0.8396 Ops/s	0.8181 Ops/s	$\color{#35bf28}+2.62\%$
test_reshape_pytree	0.2067ms	32.8905μs	30.4039 KOps/s	30.4817 KOps/s	$\color{#d91a1a}-0.26\%$
test_reshape_td	87.3120μs	46.1950μs	21.6474 KOps/s	21.9999 KOps/s	$\color{#d91a1a}-1.60\%$
test_view_pytree	0.2258ms	32.8348μs	30.4555 KOps/s	31.0965 KOps/s	$\color{#d91a1a}-2.06\%$
test_view_td	89.9420μs	54.5786μs	18.3222 KOps/s	19.0684 KOps/s	$\color{#d91a1a}-3.91\%$
test_unbind_pytree	0.2371ms	36.1129μs	27.6910 KOps/s	27.3272 KOps/s	$\color{#35bf28}+1.33\%$
test_unbind_td	0.1982ms	49.4892μs	20.2064 KOps/s	19.7120 KOps/s	$\color{#35bf28}+2.51\%$
test_split_pytree	0.2550ms	42.2740μs	23.6552 KOps/s	23.8254 KOps/s	$\color{#d91a1a}-0.71\%$
test_split_td	0.1689ms	63.4914μs	15.7502 KOps/s	15.5655 KOps/s	$\color{#35bf28}+1.19\%$
test_add_pytree	0.2330ms	42.2940μs	23.6440 KOps/s	23.5172 KOps/s	$\color{#35bf28}+0.54\%$
test_add_td	0.1110ms	54.9641μs	18.1937 KOps/s	17.9952 KOps/s	$\color{#35bf28}+1.10\%$
test_compile_add_one_nested[tensordict-compile]	0.2022ms	0.1401ms	7.1358 KOps/s	6.8324 KOps/s	$\color{#35bf28}+4.44\%$
test_compile_add_one_nested[tensordict-eager]	0.3122ms	0.2027ms	4.9334 KOps/s	4.9464 KOps/s	$\color{#d91a1a}-0.26\%$
test_compile_add_one_nested[pytree-compile]	0.1603ms	0.1111ms	9.0020 KOps/s	8.9872 KOps/s	$\color{#35bf28}+0.16\%$
test_compile_add_one_nested[pytree-eager]	0.4442ms	0.1782ms	5.6130 KOps/s	5.5894 KOps/s	$\color{#35bf28}+0.42\%$
test_compile_copy_nested[tensordict-compile]	0.2559ms	11.1881μs	89.3810 KOps/s	98.7399 KOps/s	$\textbf{\color{#d91a1a}-9.48\%}$
test_compile_copy_nested[tensordict-eager]	86.5120μs	53.6030μs	18.6557 KOps/s	18.4986 KOps/s	$\color{#35bf28}+0.85\%$
test_compile_copy_nested[pytree-compile]	0.1642ms	9.7085μs	103.0030 KOps/s	103.1693 KOps/s	$\color{#d91a1a}-0.16\%$
test_compile_copy_nested[pytree-eager]	0.4712ms	68.1504μs	14.6734 KOps/s	14.6958 KOps/s	$\color{#d91a1a}-0.15\%$
test_compile_add_one_flat[tensordict-compile]	0.3204ms	0.1766ms	5.6609 KOps/s	5.1586 KOps/s	$\textbf{\color{#35bf28}+9.74\%}$
test_compile_add_one_flat[tensordict-eager]	0.3644ms	0.2809ms	3.5604 KOps/s	3.4914 KOps/s	$\color{#35bf28}+1.98\%$
test_compile_add_one_flat[tensorclass-compile]	0.1973ms	0.1203ms	8.3134 KOps/s	7.8609 KOps/s	$\textbf{\color{#35bf28}+5.76\%}$
test_compile_add_one_flat[tensorclass-eager]	0.1233ms	74.6816μs	13.3902 KOps/s	13.2373 KOps/s	$\color{#35bf28}+1.16\%$
test_compile_add_one_flat[pytree-compile]	0.3857ms	0.1594ms	6.2724 KOps/s	5.9968 KOps/s	$\color{#35bf28}+4.60\%$
test_compile_add_one_flat[pytree-eager]	0.8122ms	0.5194ms	1.9252 KOps/s	1.8973 KOps/s	$\color{#35bf28}+1.47\%$
test_compile_add_self_flat[tensordict-eager]	0.3909ms	0.3328ms	3.0044 KOps/s	2.9569 KOps/s	$\color{#35bf28}+1.61\%$
test_compile_add_self_flat[tensordict-compile]	0.2382ms	0.1798ms	5.5631 KOps/s	5.0841 KOps/s	$\textbf{\color{#35bf28}+9.42\%}$
test_compile_add_self_flat[tensorclass-eager]	0.1489ms	90.7747μs	11.0163 KOps/s	11.1786 KOps/s	$\color{#d91a1a}-1.45\%$
test_compile_add_self_flat[tensorclass-compile]	0.2354ms	0.1223ms	8.1770 KOps/s	7.7289 KOps/s	$\textbf{\color{#35bf28}+5.80\%}$
test_compile_add_self_flat[pytree-eager]	0.6689ms	0.4348ms	2.3000 KOps/s	2.2967 KOps/s	$\color{#35bf28}+0.14\%$
test_compile_add_self_flat[pytree-compile]	0.2164ms	0.1598ms	6.2586 KOps/s	6.0495 KOps/s	$\color{#35bf28}+3.46\%$
test_compile_copy_flat[tensordict-compile]	0.1065ms	13.2543μs	75.4474 KOps/s	75.5765 KOps/s	$\color{#d91a1a}-0.17\%$
test_compile_copy_flat[tensordict-eager]	85.7020μs	41.2387μs	24.2491 KOps/s	24.0866 KOps/s	$\color{#35bf28}+0.67\%$
test_compile_copy_flat[pytree-compile]	0.1135ms	10.7158μs	93.3199 KOps/s	93.4844 KOps/s	$\color{#d91a1a}-0.18\%$
test_compile_copy_flat[pytree-eager]	0.4152ms	52.7723μs	18.9493 KOps/s	19.0813 KOps/s	$\color{#d91a1a}-0.69\%$
test_compile_assign_and_add[tensordict-compile]	2.0142ms	0.1739ms	5.7501 KOps/s	5.4517 KOps/s	$\textbf{\color{#35bf28}+5.47\%}$
test_compile_assign_and_add[tensordict-eager]	3.4723ms	3.2949ms	303.5016 Ops/s	302.5634 Ops/s	$\color{#35bf28}+0.31\%$
test_compile_assign_and_add[pytree-compile]	2.0119ms	0.1608ms	6.2186 KOps/s	5.9648 KOps/s	$\color{#35bf28}+4.26\%$
test_compile_assign_and_add[pytree-eager]	2.8630ms	2.7515ms	363.4394 Ops/s	357.3338 Ops/s	$\color{#35bf28}+1.71\%$
test_compile_indexing[tensor-tensordict-compile]	0.2253ms	0.1090ms	9.1724 KOps/s	8.3599 KOps/s	$\textbf{\color{#35bf28}+9.72\%}$
test_compile_indexing[tensor-tensordict-eager]	0.3164ms	73.0917μs	13.6814 KOps/s	13.4581 KOps/s	$\color{#35bf28}+1.66\%$
test_compile_indexing[tensor-tensorclass-compile]	0.1992ms	98.0979μs	10.1939 KOps/s	10.1075 KOps/s	$\color{#35bf28}+0.85\%$
test_compile_indexing[tensor-tensorclass-eager]	0.2600ms	45.4529μs	22.0008 KOps/s	22.8179 KOps/s	$\color{#d91a1a}-3.58\%$
test_compile_indexing[tensor-pytree-compile]	0.1483ms	97.5423μs	10.2520 KOps/s	9.5309 KOps/s	$\textbf{\color{#35bf28}+7.57\%}$
test_compile_indexing[tensor-pytree-eager]	0.2564ms	44.2970μs	22.5749 KOps/s	22.0274 KOps/s	$\color{#35bf28}+2.49\%$
test_compile_indexing[slice-tensordict-compile]	0.1103ms	54.9331μs	18.2040 KOps/s	17.2833 KOps/s	$\textbf{\color{#35bf28}+5.33\%}$
test_compile_indexing[slice-tensordict-eager]	0.2204ms	27.2677μs	36.6735 KOps/s	36.9183 KOps/s	$\color{#d91a1a}-0.66\%$
test_compile_indexing[slice-tensorclass-compile]	0.1463ms	43.2706μs	23.1104 KOps/s	21.7881 KOps/s	$\textbf{\color{#35bf28}+6.07\%}$
test_compile_indexing[slice-tensorclass-eager]	0.2572ms	22.1995μs	45.0460 KOps/s	45.0873 KOps/s	$\color{#d91a1a}-0.09\%$
test_compile_indexing[slice-pytree-compile]	96.6320μs	45.3527μs	22.0494 KOps/s	21.1751 KOps/s	$\color{#35bf28}+4.13\%$
test_compile_indexing[slice-pytree-eager]	0.2684ms	21.9986μs	45.4574 KOps/s	45.1415 KOps/s	$\color{#35bf28}+0.70\%$
test_compile_indexing[int-tensordict-compile]	0.1061ms	55.0894μs	18.1523 KOps/s	16.9716 KOps/s	$\textbf{\color{#35bf28}+6.96\%}$
test_compile_indexing[int-tensordict-eager]	0.2075ms	26.8142μs	37.2937 KOps/s	36.7996 KOps/s	$\color{#35bf28}+1.34\%$
test_compile_indexing[int-tensorclass-compile]	92.1920μs	43.7046μs	22.8809 KOps/s	21.4838 KOps/s	$\textbf{\color{#35bf28}+6.50\%}$
test_compile_indexing[int-tensorclass-eager]	0.2634ms	22.1738μs	45.0982 KOps/s	45.2912 KOps/s	$\color{#d91a1a}-0.43\%$
test_compile_indexing[int-pytree-compile]	0.1013ms	44.1522μs	22.6489 KOps/s	21.7056 KOps/s	$\color{#35bf28}+4.35\%$
test_compile_indexing[int-pytree-eager]	0.2659ms	21.9895μs	45.4763 KOps/s	45.2488 KOps/s	$\color{#35bf28}+0.50\%$
test_compile_replace[single-eager]	99.3420μs	46.6874μs	21.4191 KOps/s	21.5229 KOps/s	$\color{#d91a1a}-0.48\%$
test_compile_replace[single-compile]	0.1842ms	0.1052ms	9.5024 KOps/s	9.2608 KOps/s	$\color{#35bf28}+2.61\%$
test_compile_replace[multi-eager]	0.7133ms	0.5794ms	1.7258 KOps/s	1.7912 KOps/s	$\color{#d91a1a}-3.65\%$
test_compile_replace[multi-compile]	0.2588ms	0.1117ms	8.9500 KOps/s	8.7469 KOps/s	$\color{#35bf28}+2.32\%$
test_compile_tc_getattr_20[eager]	0.2239ms	0.1659ms	6.0263 KOps/s	6.0302 KOps/s	$\color{#d91a1a}-0.06\%$
test_compile_tc_getattr_20[compile]	0.3271ms	0.1199ms	8.3412 KOps/s	8.2301 KOps/s	$\color{#35bf28}+1.35\%$
test_compile_clone_shallow[20-eager]	47.8000μs	19.2590μs	51.9237 KOps/s	53.1324 KOps/s	$\color{#d91a1a}-2.27\%$
test_compile_clone_shallow[20-compile]	62.0820μs	11.2650μs	88.7708 KOps/s	87.2189 KOps/s	$\color{#35bf28}+1.78\%$
test_compile_clone_shallow[40-eager]	77.7310μs	33.4404μs	29.9040 KOps/s	29.6623 KOps/s	$\color{#35bf28}+0.81\%$
test_compile_clone_shallow[40-compile]	63.9010μs	12.4582μs	80.2682 KOps/s	76.1360 KOps/s	$\textbf{\color{#35bf28}+5.43\%}$
test_compile_clone_shallow[80-eager]	99.2820μs	62.3292μs	16.0438 KOps/s	15.7490 KOps/s	$\color{#35bf28}+1.87\%$
test_compile_clone_shallow[80-compile]	57.6610μs	14.7163μs	67.9521 KOps/s	69.5440 KOps/s	$\color{#d91a1a}-2.29\%$
test_compile_update_inplace[eager]	0.1028ms	58.7403μs	17.0241 KOps/s	16.7911 KOps/s	$\color{#35bf28}+1.39\%$
test_compile_update_inplace[compile]	0.2582ms	0.1389ms	7.1972 KOps/s	6.7085 KOps/s	$\textbf{\color{#35bf28}+7.28\%}$
test_mod_add[eager]	0.1220ms	49.4539μs	20.2208 KOps/s	20.4172 KOps/s	$\color{#d91a1a}-0.96\%$
test_mod_add[compile]	0.1721ms	0.1041ms	9.6031 KOps/s	9.3163 KOps/s	$\color{#35bf28}+3.08\%$
test_mod_add[compile-overhead]	0.3135ms	0.1490ms	6.7125 KOps/s	6.4987 KOps/s	$\color{#35bf28}+3.29\%$
test_mod_wrap[eager]	0.3646ms	0.2914ms	3.4318 KOps/s	3.4484 KOps/s	$\color{#d91a1a}-0.48\%$
test_mod_wrap[compile]	0.6309ms	0.3583ms	2.7906 KOps/s	2.8147 KOps/s	$\color{#d91a1a}-0.85\%$
test_mod_wrap[compile-overhead]	7.4590ms	4.1025ms	243.7534 Ops/s	248.2331 Ops/s	$\color{#d91a1a}-1.80\%$
test_mod_wrap_and_backward[eager]	1.6271ms	1.5043ms	664.7544 Ops/s	671.4262 Ops/s	$\color{#d91a1a}-0.99\%$
test_mod_wrap_and_backward[compile]	1.6054ms	1.4397ms	694.6018 Ops/s	689.4579 Ops/s	$\color{#35bf28}+0.75\%$
test_mod_wrap_and_backward[compile-overhead]	1.2448ms	0.8885ms	1.1255 KOps/s	1.0995 KOps/s	$\color{#35bf28}+2.37\%$
test_seq_add[eager]	0.2303ms	0.1569ms	6.3735 KOps/s	6.5007 KOps/s	$\color{#d91a1a}-1.96\%$
test_seq_add[compile]	0.5575ms	0.1144ms	8.7397 KOps/s	8.1318 KOps/s	$\textbf{\color{#35bf28}+7.48\%}$
test_seq_add[compile-overhead]	0.2111ms	0.1576ms	6.3463 KOps/s	5.9324 KOps/s	$\textbf{\color{#35bf28}+6.98\%}$
test_seq_wrap[eager]	0.7037ms	0.5350ms	1.8691 KOps/s	1.8848 KOps/s	$\color{#d91a1a}-0.84\%$
test_seq_wrap[compile]	0.4748ms	0.3743ms	2.6716 KOps/s	2.6332 KOps/s	$\color{#35bf28}+1.46\%$
test_seq_wrap[compile-overhead]	0.3377ms	0.2708ms	3.6926 KOps/s	3.6915 KOps/s	$\color{#35bf28}+0.03\%$
test_func_call_runtime[False-eager]	0.9710ms	0.8942ms	1.1184 KOps/s	1.1898 KOps/s	$\textbf{\color{#d91a1a}-6.00\%}$
test_func_call_runtime[False-compile]	1.1608ms	0.9267ms	1.0792 KOps/s	1.0881 KOps/s	$\color{#d91a1a}-0.82\%$
test_func_call_runtime[False-compile-overhead]	0.5167ms	0.4631ms	2.1596 KOps/s	2.1232 KOps/s	$\color{#35bf28}+1.72\%$
test_func_call_runtime[True-eager]	1.1885ms	1.0776ms	928.0098 Ops/s	925.7568 Ops/s	$\color{#35bf28}+0.24\%$
test_func_call_runtime[True-compile]	1.0103ms	0.9203ms	1.0866 KOps/s	1.0645 KOps/s	$\color{#35bf28}+2.07\%$
test_func_call_runtime[True-compile-overhead]	0.5385ms	0.4762ms	2.1001 KOps/s	2.0600 KOps/s	$\color{#35bf28}+1.95\%$
test_func_call_cm_runtime[False-eager]	1.2957ms	0.8865ms	1.1280 KOps/s	1.2007 KOps/s	$\textbf{\color{#d91a1a}-6.05\%}$
test_func_call_cm_runtime[False-compile]	0.9970ms	0.9086ms	1.1006 KOps/s	1.0892 KOps/s	$\color{#35bf28}+1.05\%$
test_func_call_cm_runtime[False-compile-overhead]	0.5416ms	0.4675ms	2.1390 KOps/s	2.1286 KOps/s	$\color{#35bf28}+0.49\%$
test_func_call_cm_runtime[True-eager]	1.3434ms	1.2310ms	812.3533 Ops/s	822.9822 Ops/s	$\color{#d91a1a}-1.29\%$
test_func_call_cm_runtime[True-compile]	0.9965ms	0.9473ms	1.0557 KOps/s	1.0296 KOps/s	$\color{#35bf28}+2.53\%$
test_func_call_cm_runtime[True-compile-overhead]	0.5554ms	0.5079ms	1.9690 KOps/s	1.9106 KOps/s	$\color{#35bf28}+3.06\%$
test_vmap_func_call_cm_runtime[eager]	2.8804ms	2.3742ms	421.1882 Ops/s	419.0351 Ops/s	$\color{#35bf28}+0.51\%$
test_vmap_func_call_cm_runtime[compile]	1.0390ms	0.9704ms	1.0305 KOps/s	1.0093 KOps/s	$\color{#35bf28}+2.11\%$
test_vmap_func_call_cm_runtime[compile-overhead]	0.5661ms	0.5179ms	1.9309 KOps/s	1.8976 KOps/s	$\color{#35bf28}+1.76\%$
test_distributed	0.8121ms	0.1532ms	6.5273 KOps/s	6.5408 KOps/s	$\color{#d91a1a}-0.21\%$
test_tdmodule	0.3072ms	27.6923μs	36.1111 KOps/s	35.7193 KOps/s	$\color{#35bf28}+1.10\%$
test_tdmodule_dispatch	72.6810μs	44.4640μs	22.4901 KOps/s	21.9698 KOps/s	$\color{#35bf28}+2.37\%$
test_tdseq	48.6710μs	26.9789μs	37.0660 KOps/s	37.0605 KOps/s	$\color{#35bf28}+0.01\%$
test_tdseq_dispatch	69.0010μs	47.6851μs	20.9709 KOps/s	21.1648 KOps/s	$\color{#d91a1a}-0.92\%$
test_instantiation_functorch	2.1768ms	2.0700ms	483.0873 Ops/s	479.8831 Ops/s	$\color{#35bf28}+0.67\%$
test_exec_functorch	0.2189ms	0.1781ms	5.6161 KOps/s	5.4980 KOps/s	$\color{#35bf28}+2.15\%$
test_exec_functional_call	0.2071ms	0.1590ms	6.2910 KOps/s	6.2286 KOps/s	$\color{#35bf28}+1.00\%$
test_exec_td_decorator	0.4426ms	0.2341ms	4.2715 KOps/s	4.2058 KOps/s	$\color{#35bf28}+1.56\%$
test_vmap_mlp_speed_decorator[True-True]	1.0070ms	0.8260ms	1.2107 KOps/s	1.2090 KOps/s	$\color{#35bf28}+0.14\%$
test_vmap_mlp_speed_decorator[True-False]	1.0332ms	0.8274ms	1.2086 KOps/s	1.2165 KOps/s	$\color{#d91a1a}-0.65\%$
test_vmap_mlp_speed_decorator[False-True]	0.9048ms	0.7131ms	1.4023 KOps/s	1.4119 KOps/s	$\color{#d91a1a}-0.68\%$
test_vmap_mlp_speed_decorator[False-False]	0.8990ms	0.7135ms	1.4014 KOps/s	1.4083 KOps/s	$\color{#d91a1a}-0.49\%$
test_vmap_transformer_speed_decorator[True-True]	21.1480ms	20.5223ms	48.7275 Ops/s	48.5774 Ops/s	$\color{#35bf28}+0.31\%$
test_vmap_transformer_speed_decorator[True-False]	21.3982ms	20.5646ms	48.6272 Ops/s	48.6261 Ops/s	$+0.00\%$
test_vmap_transformer_speed_decorator[False-True]	20.9987ms	20.3596ms	49.1168 Ops/s	49.1641 Ops/s	$\color{#d91a1a}-0.10\%$
test_vmap_transformer_speed_decorator[False-False]	21.1997ms	20.3957ms	49.0299 Ops/s	49.0477 Ops/s	$\color{#d91a1a}-0.04\%$
test_to_module_speed[True]	1.5579ms	1.4819ms	674.8291 Ops/s	674.1735 Ops/s	$\color{#35bf28}+0.10\%$
test_to_module_speed[False]	1.5642ms	1.4589ms	685.4508 Ops/s	689.5384 Ops/s	$\color{#d91a1a}-0.59\%$
test_tc_init	68.3110μs	44.9895μs	22.2274 KOps/s	22.5047 KOps/s	$\color{#d91a1a}-1.23\%$
test_tc_init_tensor_only	40.3110μs	9.7048μs	103.0422 KOps/s	101.3112 KOps/s	$\color{#35bf28}+1.71\%$
test_tc_init_nested	0.1408ms	88.7637μs	11.2659 KOps/s	11.4285 KOps/s	$\color{#d91a1a}-1.42\%$
test_tc_init_many_fields	41.9810μs	16.3301μs	61.2365 KOps/s	59.8829 KOps/s	$\color{#35bf28}+2.26\%$
test_tc_first_layer_tensor	27.0500μs	1.8241μs	548.2130 KOps/s	540.5713 KOps/s	$\color{#35bf28}+1.41\%$
test_tc_first_layer_tensor_only	2.7034μs	0.4055μs	2.4661 MOps/s	2.5418 MOps/s	$\color{#d91a1a}-2.98\%$
test_tc_first_layer_tensor_set	42.1710μs	3.9508μs	253.1162 KOps/s	253.3258 KOps/s	$\color{#d91a1a}-0.08\%$
test_tc_first_layer_tensor_only_set	30.5710μs	3.2812μs	304.7711 KOps/s	304.4631 KOps/s	$\color{#35bf28}+0.10\%$
test_tc_first_layer_nontensor	34.0510μs	6.1591μs	162.3609 KOps/s	161.7195 KOps/s	$\color{#35bf28}+0.40\%$
test_tc_second_layer_tensor	27.7510μs	4.4256μs	225.9563 KOps/s	224.8043 KOps/s	$\color{#35bf28}+0.51\%$
test_tc_second_layer_nontensor	37.8410μs	8.6909μs	115.0624 KOps/s	114.1211 KOps/s	$\color{#35bf28}+0.82\%$
test_unbind	0.2666s	18.3167ms	54.5949 Ops/s	55.2336 Ops/s	$\color{#d91a1a}-1.16\%$
test_full_like	7.5427ms	4.4471ms	224.8650 Ops/s	225.1095 Ops/s	$\color{#d91a1a}-0.11\%$
test_zeros_like	5.1286ms	4.4142ms	226.5406 Ops/s	226.1677 Ops/s	$\color{#35bf28}+0.16\%$
test_ones_like	4.6109ms	4.4113ms	226.6917 Ops/s	225.5629 Ops/s	$\color{#35bf28}+0.50\%$
test_clone	7.3167ms	6.7254ms	148.6903 Ops/s	148.4699 Ops/s	$\color{#35bf28}+0.15\%$
test_squeeze	0.2300ms	14.2391μs	70.2291 KOps/s	69.2488 KOps/s	$\color{#35bf28}+1.42\%$
test_unsqueeze	0.1645ms	0.1123ms	8.9061 KOps/s	8.9112 KOps/s	$\color{#d91a1a}-0.06\%$
test_split	0.2351ms	0.1821ms	5.4909 KOps/s	5.4149 KOps/s	$\color{#35bf28}+1.40\%$
test_permute	0.2703ms	0.2028ms	4.9302 KOps/s	4.5704 KOps/s	$\textbf{\color{#35bf28}+7.87\%}$
test_stack	36.7227ms	35.7382ms	27.9813 Ops/s	18.5745 Ops/s	$\textbf{\color{#35bf28}+50.64\%}$
test_cat	36.2006ms	35.6175ms	28.0761 Ops/s	19.2982 Ops/s	$\textbf{\color{#35bf28}+45.49\%}$
test_sequential_tensordict	0.5246ms	0.2140ms	4.6738 KOps/s	4.4058 KOps/s	$\textbf{\color{#35bf28}+6.08\%}$
test_sequential_graph_module	0.2299ms	0.1217ms	8.2203 KOps/s	8.4604 KOps/s	$\color{#d91a1a}-2.84\%$
test_nested_tensordict	0.6846ms	0.2811ms	3.5568 KOps/s	3.5104 KOps/s	$\color{#35bf28}+1.32\%$
test_nested_graph_module	0.1849ms	0.1308ms	7.6426 KOps/s	7.6602 KOps/s	$\color{#d91a1a}-0.23\%$

Update

ccdad8b

[ghstack-poisoned]

vmoens mentioned this pull request Mar 6, 2026

[DTensor] Add transfer plan computation for cross-mesh DTensor redistribution #1639

Open

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 6, 2026

This was referenced Mar 6, 2026

[DTensor] Add Strategy B (local-shard transfer + redistribute on receiver) #1641

Open

[DTensor] Add Strategy C (optimal P2P transfer using transfer plan) #1642

Open

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[DTensor] Add unified dtensor_send/dtensor_recv API with Strategy A#1640

[DTensor] Add unified dtensor_send/dtensor_recv API with Strategy A#1640
vmoens wants to merge 1 commit intogh/vmoens/81/basefrom
gh/vmoens/81/head

vmoens commented Mar 6, 2026 •

edited

Loading

Uh oh!

github-actions bot commented Mar 6, 2026

Uh oh!

github-actions bot commented Mar 6, 2026

Uh oh!

github-actions bot commented Mar 6, 2026

Uh oh!

github-actions bot commented Mar 6, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

Conversation

vmoens commented Mar 6, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

github-actions bot commented Mar 6, 2026

PR Title Label Error

Supported Prefixes

Uh oh!

github-actions bot commented Mar 6, 2026

PR Title Label Error

Supported Prefixes

Uh oh!

github-actions bot commented Mar 6, 2026

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

Total Benchmarks: 261. Improved: $\large\color{#35bf28}35$. Worsened: $\large\color{#d91a1a}6$.

Uh oh!

github-actions bot commented Mar 6, 2026

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

Total Benchmarks: 261. Improved: $\large\color{#35bf28}24$. Worsened: $\large\color{#d91a1a}7$.

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

vmoens commented Mar 6, 2026 •

edited

Loading

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

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