[Feature] broadcast pointwise ops for tensor/tensordict mixed inputs

[vmoens]

Stack from ghstack (oldest at bottom):

• -> [Feature] broadcast pointwise ops for tensor/tensordict mixed inputs #1166

[github-actions]

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

Total Benchmarks: 217. Improved: $\large\color{#35bf28}18$. Worsened: $\large\color{#d91a1a}17$.

Expand to view detailed results

Name	Max	Mean	Ops	Ops on Repo HEAD	Change
test_plain_set_nested	47.2280μs	21.0463μs	47.5143 KOps/s	49.9637 KOps/s	$\color{#d91a1a}-4.90\%$
test_plain_set_stack_nested	67.2760μs	21.1947μs	47.1817 KOps/s	48.9615 KOps/s	$\color{#d91a1a}-3.64\%$
test_plain_set_nested_inplace	0.1317ms	22.8563μs	43.7516 KOps/s	45.4554 KOps/s	$\color{#d91a1a}-3.75\%$
test_plain_set_stack_nested_inplace	76.1920μs	22.7722μs	43.9132 KOps/s	45.3432 KOps/s	$\color{#d91a1a}-3.15\%$
test_items	43.4620μs	4.1095μs	243.3412 KOps/s	238.9702 KOps/s	$\color{#35bf28}+1.83\%$
test_items_nested	0.6945ms	0.4038ms	2.4763 KOps/s	2.4352 KOps/s	$\color{#35bf28}+1.69\%$
test_items_nested_locked	0.6023ms	0.4040ms	2.4751 KOps/s	2.4335 KOps/s	$\color{#35bf28}+1.71\%$
test_items_nested_leaf	0.1478ms	77.0581μs	12.9772 KOps/s	12.9764 KOps/s	$+0.01\%$
test_items_stack_nested	0.8413ms	0.4067ms	2.4586 KOps/s	2.4203 KOps/s	$\color{#35bf28}+1.58\%$
test_items_stack_nested_leaf	0.1510ms	79.9144μs	12.5134 KOps/s	12.2530 KOps/s	$\color{#35bf28}+2.13\%$
test_items_stack_nested_locked	0.5537ms	0.4081ms	2.4503 KOps/s	2.4299 KOps/s	$\color{#35bf28}+0.84\%$
test_keys	21.1300μs	3.4856μs	286.8914 KOps/s	283.0684 KOps/s	$\color{#35bf28}+1.35\%$
test_keys_nested	0.2721ms	0.1654ms	6.0466 KOps/s	5.8665 KOps/s	$\color{#35bf28}+3.07\%$
test_keys_nested_locked	1.8037ms	0.1721ms	5.8108 KOps/s	5.7410 KOps/s	$\color{#35bf28}+1.22\%$
test_keys_nested_leaf	0.2345ms	0.1445ms	6.9208 KOps/s	6.8734 KOps/s	$\color{#35bf28}+0.69\%$
test_keys_stack_nested	0.3263ms	0.1654ms	6.0447 KOps/s	5.9918 KOps/s	$\color{#35bf28}+0.88\%$
test_keys_stack_nested_leaf	0.2327ms	0.1439ms	6.9469 KOps/s	6.8814 KOps/s	$\color{#35bf28}+0.95\%$
test_keys_stack_nested_locked	0.2678ms	0.1710ms	5.8492 KOps/s	5.8043 KOps/s	$\color{#35bf28}+0.77\%$
test_values	5.9212μs	1.0505μs	951.9314 KOps/s	951.2131 KOps/s	$\color{#35bf28}+0.08\%$
test_values_nested	0.1330ms	63.5118μs	15.7451 KOps/s	15.5950 KOps/s	$\color{#35bf28}+0.96\%$
test_values_nested_locked	0.1243ms	63.9641μs	15.6338 KOps/s	14.8120 KOps/s	$\textbf{\color{#35bf28}+5.55\%}$
test_values_nested_leaf	0.1608ms	73.0617μs	13.6871 KOps/s	13.8100 KOps/s	$\color{#d91a1a}-0.89\%$
test_values_stack_nested	0.1236ms	65.5509μs	15.2553 KOps/s	15.7673 KOps/s	$\color{#d91a1a}-3.25\%$
test_values_stack_nested_leaf	0.1281ms	74.7077μs	13.3855 KOps/s	13.6825 KOps/s	$\color{#d91a1a}-2.17\%$
test_values_stack_nested_locked	0.1162ms	64.6818μs	15.4603 KOps/s	15.6584 KOps/s	$\color{#d91a1a}-1.26\%$
test_membership	19.6470μs	0.8840μs	1.1312 MOps/s	1.1059 MOps/s	$\color{#35bf28}+2.29\%$
test_membership_nested	29.4260μs	2.9489μs	339.1105 KOps/s	344.5831 KOps/s	$\color{#d91a1a}-1.59\%$
test_membership_nested_leaf	30.4270μs	2.9948μs	333.9176 KOps/s	334.1849 KOps/s	$\color{#d91a1a}-0.08\%$
test_membership_stacked_nested	25.0670μs	2.9422μs	339.8770 KOps/s	333.4512 KOps/s	$\color{#35bf28}+1.93\%$
test_membership_stacked_nested_leaf	33.9730μs	2.9996μs	333.3747 KOps/s	346.7772 KOps/s	$\color{#d91a1a}-3.86\%$
test_membership_nested_last	28.4540μs	4.4998μs	222.2345 KOps/s	227.7478 KOps/s	$\color{#d91a1a}-2.42\%$
test_membership_nested_leaf_last	29.2540μs	4.5492μs	219.8168 KOps/s	224.3809 KOps/s	$\color{#d91a1a}-2.03\%$
test_membership_stacked_nested_last	42.3800μs	4.4453μs	224.9572 KOps/s	227.8845 KOps/s	$\color{#d91a1a}-1.28\%$
test_membership_stacked_nested_leaf_last	31.8200μs	4.4572μs	224.3566 KOps/s	225.0590 KOps/s	$\color{#d91a1a}-0.31\%$
test_nested_getleaf	49.9030μs	11.1796μs	89.4486 KOps/s	92.0904 KOps/s	$\color{#d91a1a}-2.87\%$
test_nested_get	50.8450μs	10.5874μs	94.4518 KOps/s	95.5312 KOps/s	$\color{#d91a1a}-1.13\%$
test_stacked_getleaf	50.8550μs	11.1554μs	89.6430 KOps/s	91.4396 KOps/s	$\color{#d91a1a}-1.96\%$
test_stacked_get	51.4070μs	10.5717μs	94.5923 KOps/s	96.6658 KOps/s	$\color{#d91a1a}-2.15\%$
test_nested_getitemleaf	52.7690μs	11.5951μs	86.2434 KOps/s	90.2691 KOps/s	$\color{#d91a1a}-4.46\%$
test_nested_getitem	51.9880μs	10.7994μs	92.5975 KOps/s	94.2462 KOps/s	$\color{#d91a1a}-1.75\%$
test_stacked_getitemleaf	49.9940μs	11.6280μs	85.9997 KOps/s	90.1395 KOps/s	$\color{#d91a1a}-4.59\%$
test_stacked_getitem	50.0430μs	11.5727μs	86.4106 KOps/s	94.0139 KOps/s	$\textbf{\color{#d91a1a}-8.09\%}$
test_lock_nested	2.2180ms	0.4625ms	2.1623 KOps/s	2.1261 KOps/s	$\color{#35bf28}+1.70\%$
test_lock_stack_nested	0.5810ms	0.4359ms	2.2943 KOps/s	2.2592 KOps/s	$\color{#35bf28}+1.55\%$
test_unlock_nested	1.3990ms	0.3817ms	2.6198 KOps/s	2.5807 KOps/s	$\color{#35bf28}+1.52\%$
test_unlock_stack_nested	0.6761ms	0.3529ms	2.8336 KOps/s	2.8326 KOps/s	$\color{#35bf28}+0.03\%$
test_flatten_speed	0.2020ms	0.1015ms	9.8502 KOps/s	9.7812 KOps/s	$\color{#35bf28}+0.70\%$
test_unflatten_speed	0.9163ms	0.5416ms	1.8465 KOps/s	1.8475 KOps/s	$\color{#d91a1a}-0.05\%$
test_common_ops	1.6291ms	0.8038ms	1.2440 KOps/s	1.3051 KOps/s	$\color{#d91a1a}-4.67\%$
test_creation	0.1226ms	2.7541μs	363.0956 KOps/s	395.8291 KOps/s	$\textbf{\color{#d91a1a}-8.27\%}$
test_creation_empty	43.8820μs	12.3112μs	81.2272 KOps/s	100.1901 KOps/s	$\textbf{\color{#d91a1a}-18.93\%}$
test_creation_nested_1	55.6140μs	15.0348μs	66.5122 KOps/s	77.3335 KOps/s	$\textbf{\color{#d91a1a}-13.99\%}$
test_creation_nested_2	64.6210μs	19.9042μs	50.2407 KOps/s	56.8570 KOps/s	$\textbf{\color{#d91a1a}-11.64\%}$
test_clone	51.7370μs	13.7297μs	72.8350 KOps/s	71.9938 KOps/s	$\color{#35bf28}+1.17\%$
test_getitem[int]	1.1249ms	12.8592μs	77.7654 KOps/s	76.5643 KOps/s	$\color{#35bf28}+1.57\%$
test_getitem[slice_int]	0.1549ms	25.5141μs	39.1940 KOps/s	40.0437 KOps/s	$\color{#d91a1a}-2.12\%$
test_getitem[range]	0.1874ms	48.8755μs	20.4602 KOps/s	19.8601 KOps/s	$\color{#35bf28}+3.02\%$
test_getitem[tuple]	0.1753ms	20.4314μs	48.9442 KOps/s	48.0661 KOps/s	$\color{#35bf28}+1.83\%$
test_getitem[list]	0.1991ms	44.7157μs	22.3635 KOps/s	21.9816 KOps/s	$\color{#35bf28}+1.74\%$
test_setitem_dim[int]	48.3110μs	24.6790μs	40.5202 KOps/s	38.1125 KOps/s	$\textbf{\color{#35bf28}+6.32\%}$
test_setitem_dim[slice_int]	97.0020μs	50.7209μs	19.7158 KOps/s	19.3656 KOps/s	$\color{#35bf28}+1.81\%$
test_setitem_dim[range]	0.1213ms	73.2890μs	13.6446 KOps/s	13.3924 KOps/s	$\color{#35bf28}+1.88\%$
test_setitem_dim[tuple]	81.8430μs	39.7639μs	25.1484 KOps/s	24.3531 KOps/s	$\color{#35bf28}+3.27\%$
test_setitem	65.9630μs	20.4498μs	48.9001 KOps/s	50.0865 KOps/s	$\color{#d91a1a}-2.37\%$
test_set	0.3961ms	20.2179μs	49.4612 KOps/s	52.1361 KOps/s	$\textbf{\color{#d91a1a}-5.13\%}$
test_set_shared	2.1581ms	0.1712ms	5.8402 KOps/s	5.5960 KOps/s	$\color{#35bf28}+4.36\%$
test_update	0.3711ms	23.3425μs	42.8403 KOps/s	46.7775 KOps/s	$\textbf{\color{#d91a1a}-8.42\%}$
test_update_nested	0.3657ms	33.7708μs	29.6114 KOps/s	31.3871 KOps/s	$\textbf{\color{#d91a1a}-5.66\%}$
test_update__nested	0.7612ms	33.5288μs	29.8251 KOps/s	29.1374 KOps/s	$\color{#35bf28}+2.36\%$
test_set_nested	0.3813ms	22.5018μs	44.4410 KOps/s	46.0089 KOps/s	$\color{#d91a1a}-3.41\%$
test_set_nested_new	0.3791ms	27.2467μs	36.7017 KOps/s	37.6014 KOps/s	$\color{#d91a1a}-2.39\%$
test_select	0.3982ms	44.4787μs	22.4827 KOps/s	23.5550 KOps/s	$\color{#d91a1a}-4.55\%$
test_select_nested	0.1216ms	64.1126μs	15.5976 KOps/s	15.4363 KOps/s	$\color{#35bf28}+1.04\%$
test_exclude_nested	0.1629ms	83.4754μs	11.9796 KOps/s	12.1454 KOps/s	$\color{#d91a1a}-1.37\%$
test_empty[True]	0.5229ms	0.4136ms	2.4177 KOps/s	2.3526 KOps/s	$\color{#35bf28}+2.77\%$
test_empty[False]	13.2147μs	1.3903μs	719.2908 KOps/s	699.2205 KOps/s	$\color{#35bf28}+2.87\%$
test_unbind_speed	0.4012ms	0.2730ms	3.6632 KOps/s	3.5981 KOps/s	$\color{#35bf28}+1.81\%$
test_unbind_speed_stack0	0.3904ms	0.2732ms	3.6606 KOps/s	3.6622 KOps/s	$\color{#d91a1a}-0.05\%$
test_unbind_speed_stack1	0.1185s	0.8520ms	1.1737 KOps/s	1.3027 KOps/s	$\textbf{\color{#d91a1a}-9.90\%}$
test_split	1.7352ms	1.6074ms	622.1405 Ops/s	550.4681 Ops/s	$\textbf{\color{#35bf28}+13.02\%}$
test_chunk	0.1213s	2.0246ms	493.9230 Ops/s	554.8626 Ops/s	$\textbf{\color{#d91a1a}-10.98\%}$
test_consolidate_njt[False-None]	8.9948ms	8.2553ms	121.1341 Ops/s	116.7531 Ops/s	$\color{#35bf28}+3.75\%$
test_creation[device0]	4.5713ms	94.5646μs	10.5748 KOps/s	10.7168 KOps/s	$\color{#d91a1a}-1.33\%$
test_creation_from_tensor	0.4410ms	98.7902μs	10.1225 KOps/s	10.5411 KOps/s	$\color{#d91a1a}-3.97\%$
test_add_one[memmap_tensor0]	0.1199ms	4.9682μs	201.2802 KOps/s	199.7852 KOps/s	$\color{#35bf28}+0.75\%$
test_contiguous[memmap_tensor0]	8.1250μs	0.5214μs	1.9181 MOps/s	1.9466 MOps/s	$\color{#d91a1a}-1.47\%$
test_stack[memmap_tensor0]	0.1307ms	3.5664μs	280.3985 KOps/s	292.7340 KOps/s	$\color{#d91a1a}-4.21\%$
test_memmaptd_index	1.1395ms	0.2352ms	4.2508 KOps/s	4.1335 KOps/s	$\color{#35bf28}+2.84\%$
test_memmaptd_index_astensor	0.8136ms	0.3250ms	3.0772 KOps/s	3.0032 KOps/s	$\color{#35bf28}+2.46\%$
test_memmaptd_index_op	1.0806ms	0.6049ms	1.6531 KOps/s	1.7263 KOps/s	$\color{#d91a1a}-4.24\%$
test_serialize_model	0.1258s	0.1202s	8.3182 Ops/s	8.2356 Ops/s	$\color{#35bf28}+1.00\%$
test_serialize_model_pickle	0.4339s	0.3988s	2.5075 Ops/s	2.4975 Ops/s	$\color{#35bf28}+0.40\%$
test_serialize_weights	0.1286s	0.1215s	8.2328 Ops/s	7.1621 Ops/s	$\textbf{\color{#35bf28}+14.95\%}$
test_serialize_weights_returnearly	0.2823s	0.1832s	5.4588 Ops/s	6.4306 Ops/s	$\textbf{\color{#d91a1a}-15.11\%}$
test_serialize_weights_pickle	0.4832s	0.4069s	2.4574 Ops/s	2.3460 Ops/s	$\color{#35bf28}+4.75\%$
test_serialize_weights_filesystem	0.1595s	0.1466s	6.8209 Ops/s	6.8354 Ops/s	$\color{#d91a1a}-0.21\%$
test_serialize_model_filesystem	0.1637s	0.1541s	6.4903 Ops/s	6.5987 Ops/s	$\color{#d91a1a}-1.64\%$
test_reshape_pytree	85.7610μs	26.5151μs	37.7144 KOps/s	37.4850 KOps/s	$\color{#35bf28}+0.61\%$
test_reshape_td	78.3570μs	33.0780μs	30.2316 KOps/s	29.7203 KOps/s	$\color{#35bf28}+1.72\%$
test_view_pytree	0.1054ms	26.9401μs	37.1194 KOps/s	36.9710 KOps/s	$\color{#35bf28}+0.40\%$
test_view_td	85.4600μs	37.8907μs	26.3917 KOps/s	25.1669 KOps/s	$\color{#35bf28}+4.87\%$
test_unbind_pytree	65.0620μs	29.7801μs	33.5795 KOps/s	32.8652 KOps/s	$\color{#35bf28}+2.17\%$
test_unbind_td	0.3509ms	40.6281μs	24.6135 KOps/s	24.6339 KOps/s	$\color{#d91a1a}-0.08\%$
test_split_pytree	72.5460μs	29.6169μs	33.7645 KOps/s	33.5474 KOps/s	$\color{#35bf28}+0.65\%$
test_split_td	0.6560ms	45.7118μs	21.8762 KOps/s	22.1434 KOps/s	$\color{#d91a1a}-1.21\%$
test_add_pytree	83.4560μs	35.3287μs	28.3056 KOps/s	27.9565 KOps/s	$\color{#35bf28}+1.25\%$
test_add_td	0.1369ms	61.0805μs	16.3718 KOps/s	18.4130 KOps/s	$\textbf{\color{#d91a1a}-11.09\%}$
test_compile_add_one_nested[tensordict-compile]	0.2119ms	64.1445μs	15.5898 KOps/s	15.8017 KOps/s	$\color{#d91a1a}-1.34\%$
test_compile_add_one_nested[tensordict-eager]	0.5456ms	0.1786ms	5.6006 KOps/s	5.7980 KOps/s	$\color{#d91a1a}-3.40\%$
test_compile_add_one_nested[pytree-compile]	0.1066ms	45.4638μs	21.9955 KOps/s	21.2423 KOps/s	$\color{#35bf28}+3.55\%$
test_compile_add_one_nested[pytree-eager]	0.2624ms	0.1184ms	8.4488 KOps/s	8.3823 KOps/s	$\color{#35bf28}+0.79\%$
test_compile_copy_nested[tensordict-compile]	73.3470μs	25.9896μs	38.4770 KOps/s	36.8485 KOps/s	$\color{#35bf28}+4.42\%$
test_compile_copy_nested[tensordict-eager]	0.1205ms	59.2236μs	16.8852 KOps/s	16.6983 KOps/s	$\color{#35bf28}+1.12\%$
test_compile_copy_nested[pytree-compile]	0.3556ms	78.3428μs	12.7644 KOps/s	12.5487 KOps/s	$\color{#35bf28}+1.72\%$
test_compile_copy_nested[pytree-eager]	0.1415ms	68.0511μs	14.6948 KOps/s	14.4212 KOps/s	$\color{#35bf28}+1.90\%$
test_compile_add_one_flat[tensordict-compile]	0.1923ms	0.1054ms	9.4845 KOps/s	9.4167 KOps/s	$\color{#35bf28}+0.72\%$
test_compile_add_one_flat[tensordict-eager]	0.4486ms	0.2215ms	4.5137 KOps/s	4.5650 KOps/s	$\color{#d91a1a}-1.12\%$
test_compile_add_one_flat[tensorclass-compile]	0.1344ms	44.6744μs	22.3842 KOps/s	21.6216 KOps/s	$\color{#35bf28}+3.53\%$
test_compile_add_one_flat[tensorclass-eager]	0.4815ms	67.4723μs	14.8209 KOps/s	15.3332 KOps/s	$\color{#d91a1a}-3.34\%$
test_compile_add_one_flat[pytree-compile]	0.1842ms	0.1023ms	9.7799 KOps/s	9.7106 KOps/s	$\color{#35bf28}+0.71\%$
test_compile_add_one_flat[pytree-eager]	0.4375ms	0.2011ms	4.9725 KOps/s	4.9811 KOps/s	$\color{#d91a1a}-0.17\%$
test_compile_add_self_flat[tensordict-eager]	0.4689ms	0.2406ms	4.1555 KOps/s	4.2379 KOps/s	$\color{#d91a1a}-1.94\%$
test_compile_add_self_flat[tensordict-compile]	0.2002ms	0.1042ms	9.5924 KOps/s	9.3541 KOps/s	$\color{#35bf28}+2.55\%$
test_compile_add_self_flat[tensorclass-eager]	0.1526ms	68.5874μs	14.5799 KOps/s	16.8855 KOps/s	$\textbf{\color{#d91a1a}-13.65\%}$
test_compile_add_self_flat[tensorclass-compile]	0.1168ms	47.1648μs	21.2022 KOps/s	21.8986 KOps/s	$\color{#d91a1a}-3.18\%$
test_compile_add_self_flat[pytree-eager]	0.2527ms	0.1584ms	6.3114 KOps/s	6.3034 KOps/s	$\color{#35bf28}+0.13\%$
test_compile_add_self_flat[pytree-compile]	0.1899ms	0.1062ms	9.4156 KOps/s	9.6242 KOps/s	$\color{#d91a1a}-2.17\%$
test_compile_copy_flat[tensordict-compile]	79.0000μs	21.0505μs	47.5047 KOps/s	46.0732 KOps/s	$\color{#35bf28}+3.11\%$
test_compile_copy_flat[tensordict-eager]	0.1337ms	66.2283μs	15.0993 KOps/s	15.0454 KOps/s	$\color{#35bf28}+0.36\%$
test_compile_copy_flat[pytree-compile]	0.1506ms	83.3610μs	11.9960 KOps/s	12.0967 KOps/s	$\color{#d91a1a}-0.83\%$
test_compile_copy_flat[pytree-eager]	0.1305ms	69.0873μs	14.4744 KOps/s	14.3643 KOps/s	$\color{#35bf28}+0.77\%$
test_compile_assign_and_add[tensordict-compile]	0.3097ms	0.2054ms	4.8684 KOps/s	4.8120 KOps/s	$\color{#35bf28}+1.17\%$
test_compile_assign_and_add[tensordict-eager]	1.6271ms	1.3423ms	744.9959 Ops/s	756.9032 Ops/s	$\color{#d91a1a}-1.57\%$
test_compile_assign_and_add[pytree-compile]	0.3349ms	0.2021ms	4.9489 KOps/s	4.8556 KOps/s	$\color{#35bf28}+1.92\%$
test_compile_assign_and_add[pytree-eager]	0.9792ms	0.7759ms	1.2888 KOps/s	1.2958 KOps/s	$\color{#d91a1a}-0.54\%$
test_compile_assign_and_add_stack[compile]	0.5410ms	0.4499ms	2.2227 KOps/s	2.2030 KOps/s	$\color{#35bf28}+0.89\%$
test_compile_assign_and_add_stack[eager]	3.5396ms	2.7415ms	364.7630 Ops/s	378.2797 Ops/s	$\color{#d91a1a}-3.57\%$
test_compile_indexing[tensor-tensordict-compile]	90.1490μs	34.5361μs	28.9552 KOps/s	26.6725 KOps/s	$\textbf{\color{#35bf28}+8.56\%}$
test_compile_indexing[tensor-tensordict-eager]	0.6850ms	32.9796μs	30.3218 KOps/s	28.9533 KOps/s	$\color{#35bf28}+4.73\%$
test_compile_indexing[tensor-tensorclass-compile]	0.1088ms	28.6509μs	34.9030 KOps/s	32.6979 KOps/s	$\textbf{\color{#35bf28}+6.74\%}$
test_compile_indexing[tensor-tensorclass-eager]	0.1045ms	23.4919μs	42.5679 KOps/s	40.3790 KOps/s	$\textbf{\color{#35bf28}+5.42\%}$
test_compile_indexing[tensor-pytree-compile]	78.9380μs	29.3240μs	34.1017 KOps/s	31.9058 KOps/s	$\textbf{\color{#35bf28}+6.88\%}$
test_compile_indexing[tensor-pytree-eager]	0.1036ms	23.4003μs	42.7345 KOps/s	42.1286 KOps/s	$\color{#35bf28}+1.44\%$
test_compile_indexing[slice-tensordict-compile]	0.1324ms	50.2069μs	19.9176 KOps/s	18.7380 KOps/s	$\textbf{\color{#35bf28}+6.30\%}$
test_compile_indexing[slice-tensordict-eager]	0.5812ms	20.7127μs	48.2796 KOps/s	49.3451 KOps/s	$\color{#d91a1a}-2.16\%$
test_compile_indexing[slice-tensorclass-compile]	0.1006ms	42.4556μs	23.5540 KOps/s	22.0801 KOps/s	$\textbf{\color{#35bf28}+6.68\%}$
test_compile_indexing[slice-tensorclass-eager]	99.1060μs	18.6670μs	53.5705 KOps/s	51.9637 KOps/s	$\color{#35bf28}+3.09\%$
test_compile_indexing[slice-pytree-compile]	0.1219ms	43.4751μs	23.0017 KOps/s	21.7617 KOps/s	$\textbf{\color{#35bf28}+5.70\%}$
test_compile_indexing[slice-pytree-eager]	98.8370μs	18.4206μs	54.2872 KOps/s	52.3899 KOps/s	$\color{#35bf28}+3.62\%$
test_compile_indexing[int-tensordict-compile]	0.1587ms	51.3842μs	19.4612 KOps/s	18.2631 KOps/s	$\textbf{\color{#35bf28}+6.56\%}$
test_compile_indexing[int-tensordict-eager]	1.0636ms	20.0797μs	49.8016 KOps/s	50.3002 KOps/s	$\color{#d91a1a}-0.99\%$
test_compile_indexing[int-tensorclass-compile]	98.0640μs	43.9402μs	22.7582 KOps/s	21.4946 KOps/s	$\textbf{\color{#35bf28}+5.88\%}$
test_compile_indexing[int-tensorclass-eager]	57.1570μs	18.5616μs	53.8747 KOps/s	52.8079 KOps/s	$\color{#35bf28}+2.02\%$
test_compile_indexing[int-pytree-compile]	0.1225ms	44.1620μs	22.6439 KOps/s	21.5927 KOps/s	$\color{#35bf28}+4.87\%$
test_compile_indexing[int-pytree-eager]	57.2570μs	18.5050μs	54.0395 KOps/s	52.9247 KOps/s	$\color{#35bf28}+2.11\%$
test_mod_add[eager]	0.1268ms	33.7419μs	29.6367 KOps/s	29.4539 KOps/s	$\color{#35bf28}+0.62\%$
test_mod_add[compile]	0.1420ms	47.1806μs	21.1951 KOps/s	19.9226 KOps/s	$\textbf{\color{#35bf28}+6.39\%}$
test_mod_add[compile-overhead]	0.1028ms	46.4634μs	21.5223 KOps/s	19.7515 KOps/s	$\textbf{\color{#35bf28}+8.97\%}$
test_mod_wrap[eager]	0.3644ms	0.2279ms	4.3882 KOps/s	4.3363 KOps/s	$\color{#35bf28}+1.20\%$
test_mod_wrap[compile]	0.3012ms	0.2090ms	4.7858 KOps/s	4.8101 KOps/s	$\color{#d91a1a}-0.50\%$
test_mod_wrap[compile-overhead]	0.3941ms	0.2088ms	4.7893 KOps/s	4.8451 KOps/s	$\color{#d91a1a}-1.15\%$
test_mod_wrap_and_backward[eager]	18.1627ms	12.3833ms	80.7538 Ops/s	82.3454 Ops/s	$\color{#d91a1a}-1.93\%$
test_mod_wrap_and_backward[compile]	14.0254ms	12.8747ms	77.6716 Ops/s	85.5255 Ops/s	$\textbf{\color{#d91a1a}-9.18\%}$
test_mod_wrap_and_backward[compile-overhead]	13.5814ms	12.0057ms	83.2935 Ops/s	85.5481 Ops/s	$\color{#d91a1a}-2.64\%$
test_seq_add[eager]	0.2254ms	0.1141ms	8.7621 KOps/s	8.4732 KOps/s	$\color{#35bf28}+3.41\%$
test_seq_add[compile]	0.1710ms	62.0005μs	16.1289 KOps/s	15.5482 KOps/s	$\color{#35bf28}+3.73\%$
test_seq_add[compile-overhead]	0.1315ms	60.6624μs	16.4847 KOps/s	16.0373 KOps/s	$\color{#35bf28}+2.79\%$
test_seq_wrap[eager]	0.7984ms	0.4491ms	2.2269 KOps/s	2.2234 KOps/s	$\color{#35bf28}+0.16\%$
test_seq_wrap[compile]	0.4060ms	0.2350ms	4.2554 KOps/s	4.2670 KOps/s	$\color{#d91a1a}-0.27\%$
test_seq_wrap[compile-overhead]	0.4449ms	0.2304ms	4.3399 KOps/s	4.3555 KOps/s	$\color{#d91a1a}-0.36\%$
test_func_call_runtime[False-eager]	0.9938ms	0.5540ms	1.8050 KOps/s	1.8035 KOps/s	$\color{#35bf28}+0.08\%$
test_func_call_runtime[False-compile]	0.5507ms	0.4370ms	2.2882 KOps/s	2.3221 KOps/s	$\color{#d91a1a}-1.46\%$
test_func_call_runtime[False-compile-overhead]	0.8455ms	0.4435ms	2.2549 KOps/s	2.3176 KOps/s	$\color{#d91a1a}-2.71\%$
test_func_call_runtime[True-eager]	1.3102ms	0.7728ms	1.2940 KOps/s	1.2896 KOps/s	$\color{#35bf28}+0.34\%$
test_func_call_runtime[True-compile]	0.6567ms	0.4822ms	2.0740 KOps/s	2.1007 KOps/s	$\color{#d91a1a}-1.27\%$
test_func_call_runtime[True-compile-overhead]	0.6614ms	0.4808ms	2.0800 KOps/s	2.1202 KOps/s	$\color{#d91a1a}-1.90\%$
test_func_call_cm_runtime[False-eager]	1.0068ms	0.5528ms	1.8088 KOps/s	1.8305 KOps/s	$\color{#d91a1a}-1.18\%$
test_func_call_cm_runtime[False-compile]	0.6197ms	0.4411ms	2.2672 KOps/s	2.3194 KOps/s	$\color{#d91a1a}-2.25\%$
test_func_call_cm_runtime[False-compile-overhead]	0.6284ms	0.4389ms	2.2782 KOps/s	2.3288 KOps/s	$\color{#d91a1a}-2.17\%$
test_func_call_cm_runtime[True-eager]	1.4757ms	0.9231ms	1.0833 KOps/s	1.0941 KOps/s	$\color{#d91a1a}-0.98\%$
test_func_call_cm_runtime[True-compile]	0.6844ms	0.5041ms	1.9838 KOps/s	2.0042 KOps/s	$\color{#d91a1a}-1.02\%$
test_func_call_cm_runtime[True-compile-overhead]	0.6691ms	0.5084ms	1.9671 KOps/s	1.9935 KOps/s	$\color{#d91a1a}-1.32\%$
test_vmap_func_call_cm_runtime[eager]	2.8381ms	1.9645ms	509.0265 Ops/s	513.1605 Ops/s	$\color{#d91a1a}-0.81\%$
test_vmap_func_call_cm_runtime[compile]	0.7546ms	0.5250ms	1.9049 KOps/s	1.8831 KOps/s	$\color{#35bf28}+1.16\%$
test_vmap_func_call_cm_runtime[compile-overhead]	1.0866ms	0.5432ms	1.8408 KOps/s	1.8869 KOps/s	$\color{#d91a1a}-2.44\%$
test_distributed	0.4017ms	0.1259ms	7.9428 KOps/s	7.6339 KOps/s	$\color{#35bf28}+4.05\%$
test_tdmodule	44.3030μs	26.4544μs	37.8009 KOps/s	39.4531 KOps/s	$\color{#d91a1a}-4.19\%$
test_tdmodule_dispatch	78.1160μs	48.6631μs	20.5495 KOps/s	21.7760 KOps/s	$\textbf{\color{#d91a1a}-5.63\%}$
test_tdseq	60.1530μs	29.5336μs	33.8597 KOps/s	34.6601 KOps/s	$\color{#d91a1a}-2.31\%$
test_tdseq_dispatch	88.7960μs	54.6123μs	18.3109 KOps/s	19.0228 KOps/s	$\color{#d91a1a}-3.74\%$
test_instantiation_functorch	2.3641ms	1.5359ms	651.0947 Ops/s	634.2536 Ops/s	$\color{#35bf28}+2.66\%$
test_exec_functorch	0.3368ms	0.1789ms	5.5887 KOps/s	5.3855 KOps/s	$\color{#35bf28}+3.77\%$
test_exec_functional_call	0.3021ms	0.1756ms	5.6933 KOps/s	5.8182 KOps/s	$\color{#d91a1a}-2.15\%$
test_exec_td_decorator	0.5451ms	0.2380ms	4.2017 KOps/s	4.3249 KOps/s	$\color{#d91a1a}-2.85\%$
test_vmap_mlp_speed_decorator[True-True]	0.8908ms	0.6732ms	1.4853 KOps/s	1.5062 KOps/s	$\color{#d91a1a}-1.39\%$
test_vmap_mlp_speed_decorator[True-False]	1.1819ms	0.6745ms	1.4826 KOps/s	1.5067 KOps/s	$\color{#d91a1a}-1.60\%$
test_vmap_mlp_speed_decorator[False-True]	0.8562ms	0.5403ms	1.8507 KOps/s	1.8772 KOps/s	$\color{#d91a1a}-1.41\%$
test_vmap_mlp_speed_decorator[False-False]	0.8561ms	0.5416ms	1.8463 KOps/s	1.8719 KOps/s	$\color{#d91a1a}-1.36\%$
test_to_module_speed[True]	2.6468ms	1.3638ms	733.2409 Ops/s	738.8994 Ops/s	$\color{#d91a1a}-0.77\%$
test_to_module_speed[False]	2.0798ms	1.3304ms	751.6407 Ops/s	725.9635 Ops/s	$\color{#35bf28}+3.54\%$
test_tc_init	99.9270μs	45.6227μs	21.9189 KOps/s	22.8320 KOps/s	$\color{#d91a1a}-4.00\%$
test_tc_init_nested	0.1944ms	94.0133μs	10.6368 KOps/s	11.3242 KOps/s	$\textbf{\color{#d91a1a}-6.07\%}$
test_tc_first_layer_tensor	27.1710μs	1.5548μs	643.1902 KOps/s	652.9202 KOps/s	$\color{#d91a1a}-1.49\%$
test_tc_first_layer_nontensor	27.6210μs	4.7010μs	212.7194 KOps/s	212.2733 KOps/s	$\color{#35bf28}+0.21\%$
test_tc_second_layer_tensor	24.9870μs	2.8723μs	348.1473 KOps/s	350.8432 KOps/s	$\color{#d91a1a}-0.77\%$
test_tc_second_layer_nontensor	52.4080μs	6.0868μs	164.2907 KOps/s	165.7808 KOps/s	$\color{#d91a1a}-0.90\%$
test_unbind	0.2542s	14.2526ms	70.1629 Ops/s	62.7140 Ops/s	$\textbf{\color{#35bf28}+11.88\%}$
test_full_like	12.9306ms	9.7799ms	102.2505 Ops/s	110.0829 Ops/s	$\textbf{\color{#d91a1a}-7.12\%}$
test_zeros_like	4.2041ms	3.4926ms	286.3164 Ops/s	290.7328 Ops/s	$\color{#d91a1a}-1.52\%$
test_ones_like	5.1076ms	4.2276ms	236.5388 Ops/s	149.7090 Ops/s	$\textbf{\color{#35bf28}+58.00\%}$
test_clone	7.2973ms	6.4086ms	156.0411 Ops/s	111.2004 Ops/s	$\textbf{\color{#35bf28}+40.32\%}$
test_squeeze	0.1000ms	12.1392μs	82.3780 KOps/s	81.8939 KOps/s	$\color{#35bf28}+0.59\%$
test_unsqueeze	0.1729ms	92.7443μs	10.7823 KOps/s	10.8544 KOps/s	$\color{#d91a1a}-0.66\%$
test_split	0.3985ms	0.1929ms	5.1845 KOps/s	5.1101 KOps/s	$\color{#35bf28}+1.46\%$
test_permute	0.4027ms	0.2094ms	4.7760 KOps/s	4.7332 KOps/s	$\color{#35bf28}+0.90\%$
test_stack	34.9362ms	28.6046ms	34.9594 Ops/s	34.8049 Ops/s	$\color{#35bf28}+0.44\%$
test_cat	32.4270ms	28.2399ms	35.4109 Ops/s	35.0316 Ops/s	$\color{#35bf28}+1.08\%$

[github-actions]

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

Total Benchmarks: 229. Improved: $\large\color{#35bf28}30$. Worsened: $\large\color{#d91a1a}15$.

Expand to view detailed results

Name	Max	Mean	Ops	Ops on Repo HEAD	Change
test_plain_set_nested	28.4610μs	11.3055μs	88.4529 KOps/s	78.5192 KOps/s	$\textbf{\color{#35bf28}+12.65\%}$
test_plain_set_stack_nested	36.6100μs	11.4933μs	87.0075 KOps/s	77.1846 KOps/s	$\textbf{\color{#35bf28}+12.73\%}$
test_plain_set_nested_inplace	44.7900μs	12.3651μs	80.8725 KOps/s	72.2716 KOps/s	$\textbf{\color{#35bf28}+11.90\%}$
test_plain_set_stack_nested_inplace	38.4110μs	12.3154μs	81.1992 KOps/s	72.0625 KOps/s	$\textbf{\color{#35bf28}+12.68\%}$
test_items	26.0600μs	2.8635μs	349.2191 KOps/s	343.0273 KOps/s	$\color{#35bf28}+1.81\%$
test_items_nested	0.4022ms	0.3577ms	2.7955 KOps/s	2.7840 KOps/s	$\color{#35bf28}+0.41\%$
test_items_nested_locked	0.4225ms	0.3540ms	2.8250 KOps/s	2.7651 KOps/s	$\color{#35bf28}+2.17\%$
test_items_nested_leaf	79.5010μs	57.9045μs	17.2698 KOps/s	17.3530 KOps/s	$\color{#d91a1a}-0.48\%$
test_items_stack_nested	0.3974ms	0.3613ms	2.7679 KOps/s	2.7580 KOps/s	$\color{#35bf28}+0.36\%$
test_items_stack_nested_leaf	93.1020μs	59.9991μs	16.6669 KOps/s	16.7316 KOps/s	$\color{#d91a1a}-0.39\%$
test_items_stack_nested_locked	0.4031ms	0.3590ms	2.7859 KOps/s	2.7419 KOps/s	$\color{#35bf28}+1.60\%$
test_keys	24.6300μs	3.4399μs	290.7019 KOps/s	291.5825 KOps/s	$\color{#d91a1a}-0.30\%$
test_keys_nested	0.1119ms	80.7549μs	12.3831 KOps/s	12.0827 KOps/s	$\color{#35bf28}+2.49\%$
test_keys_nested_locked	0.8195ms	86.1642μs	11.6057 KOps/s	11.3455 KOps/s	$\color{#35bf28}+2.29\%$
test_keys_nested_leaf	2.7085ms	72.1271μs	13.8644 KOps/s	13.7198 KOps/s	$\color{#35bf28}+1.05\%$
test_keys_stack_nested	0.1159ms	82.2833μs	12.1531 KOps/s	12.0061 KOps/s	$\color{#35bf28}+1.22\%$
test_keys_stack_nested_leaf	0.1183ms	72.8543μs	13.7260 KOps/s	13.4575 KOps/s	$\color{#35bf28}+2.00\%$
test_keys_stack_nested_locked	0.1168ms	88.1604μs	11.3430 KOps/s	11.2839 KOps/s	$\color{#35bf28}+0.52\%$
test_values	3.7490μs	0.8430μs	1.1863 MOps/s	1.1781 MOps/s	$\color{#35bf28}+0.69\%$
test_values_nested	60.2110μs	34.5108μs	28.9764 KOps/s	29.0648 KOps/s	$\color{#d91a1a}-0.30\%$
test_values_nested_locked	61.9110μs	35.9280μs	27.8334 KOps/s	27.6388 KOps/s	$\color{#35bf28}+0.70\%$
test_values_nested_leaf	74.1310μs	39.1078μs	25.5704 KOps/s	25.3626 KOps/s	$\color{#35bf28}+0.82\%$
test_values_stack_nested	61.7410μs	34.9548μs	28.6084 KOps/s	28.5087 KOps/s	$\color{#35bf28}+0.35\%$
test_values_stack_nested_leaf	73.3610μs	39.6028μs	25.2507 KOps/s	25.5833 KOps/s	$\color{#d91a1a}-1.30\%$
test_values_stack_nested_locked	67.7810μs	36.5744μs	27.3416 KOps/s	27.5046 KOps/s	$\color{#d91a1a}-0.59\%$
test_membership	1.7230μs	0.5019μs	1.9922 MOps/s	1.9639 MOps/s	$\color{#35bf28}+1.44\%$
test_membership_nested	31.4600μs	2.0136μs	496.6285 KOps/s	511.0686 KOps/s	$\color{#d91a1a}-2.83\%$
test_membership_nested_leaf	15.4355μs	1.9560μs	511.2483 KOps/s	513.5656 KOps/s	$\color{#d91a1a}-0.45\%$
test_membership_stacked_nested	27.0810μs	2.0290μs	492.8630 KOps/s	478.5833 KOps/s	$\color{#35bf28}+2.98\%$
test_membership_stacked_nested_leaf	23.7510μs	2.0432μs	489.4320 KOps/s	485.8433 KOps/s	$\color{#35bf28}+0.74\%$
test_membership_nested_last	27.2800μs	3.0604μs	326.7584 KOps/s	329.4252 KOps/s	$\color{#d91a1a}-0.81\%$
test_membership_nested_leaf_last	27.6500μs	3.0155μs	331.6180 KOps/s	324.5486 KOps/s	$\color{#35bf28}+2.18\%$
test_membership_stacked_nested_last	33.1410μs	3.0805μs	324.6221 KOps/s	147.4585 KOps/s	$\textbf{\color{#35bf28}+120.14\%}$
test_membership_stacked_nested_leaf_last	36.8010μs	3.0594μs	326.8654 KOps/s	146.3000 KOps/s	$\textbf{\color{#35bf28}+123.42\%}$
test_nested_getleaf	37.2310μs	6.0682μs	164.7926 KOps/s	165.6221 KOps/s	$\color{#d91a1a}-0.50\%$
test_nested_get	29.9000μs	5.8083μs	172.1679 KOps/s	174.0449 KOps/s	$\color{#d91a1a}-1.08\%$
test_stacked_getleaf	36.9410μs	6.1066μs	163.7585 KOps/s	165.2708 KOps/s	$\color{#d91a1a}-0.92\%$
test_stacked_get	35.8410μs	5.8069μs	172.2083 KOps/s	174.0533 KOps/s	$\color{#d91a1a}-1.06\%$
test_nested_getitemleaf	37.2410μs	6.1717μs	162.0306 KOps/s	159.5021 KOps/s	$\color{#35bf28}+1.59\%$
test_nested_getitem	38.3910μs	5.9658μs	167.6211 KOps/s	170.6297 KOps/s	$\color{#d91a1a}-1.76\%$
test_stacked_getitemleaf	36.0810μs	6.2178μs	160.8296 KOps/s	161.5666 KOps/s	$\color{#d91a1a}-0.46\%$
test_stacked_getitem	35.8510μs	5.9115μs	169.1625 KOps/s	169.4064 KOps/s	$\color{#d91a1a}-0.14\%$
test_lock_nested	2.4897ms	0.3669ms	2.7254 KOps/s	2.7169 KOps/s	$\color{#35bf28}+0.31\%$
test_lock_stack_nested	0.3852ms	0.3396ms	2.9444 KOps/s	2.9781 KOps/s	$\color{#d91a1a}-1.13\%$
test_unlock_nested	0.6449ms	0.3070ms	3.2574 KOps/s	3.2733 KOps/s	$\color{#d91a1a}-0.49\%$
test_unlock_stack_nested	0.3193ms	0.2789ms	3.5849 KOps/s	3.6409 KOps/s	$\color{#d91a1a}-1.54\%$
test_flatten_speed	0.1185ms	74.7354μs	13.3805 KOps/s	13.3075 KOps/s	$\color{#35bf28}+0.55\%$
test_unflatten_speed	0.3592ms	0.3139ms	3.1861 KOps/s	3.1495 KOps/s	$\color{#35bf28}+1.16\%$
test_common_ops	1.5438ms	0.5581ms	1.7917 KOps/s	1.6011 KOps/s	$\textbf{\color{#35bf28}+11.90\%}$
test_creation	0.1049ms	1.7155μs	582.9296 KOps/s	585.1059 KOps/s	$\color{#d91a1a}-0.37\%$
test_creation_empty	35.8410μs	6.3273μs	158.0456 KOps/s	105.7940 KOps/s	$\textbf{\color{#35bf28}+49.39\%}$
test_creation_nested_1	31.7310μs	8.0371μs	124.4231 KOps/s	89.0497 KOps/s	$\textbf{\color{#35bf28}+39.72\%}$
test_creation_nested_2	45.4110μs	10.6900μs	93.5454 KOps/s	72.0865 KOps/s	$\textbf{\color{#35bf28}+29.77\%}$
test_clone	0.1339ms	10.3465μs	96.6508 KOps/s	96.3950 KOps/s	$\color{#35bf28}+0.27\%$
test_getitem[int]	1.8925ms	10.4462μs	95.7285 KOps/s	94.5667 KOps/s	$\color{#35bf28}+1.23\%$
test_getitem[slice_int]	0.1132ms	20.1614μs	49.5998 KOps/s	48.8534 KOps/s	$\color{#35bf28}+1.53\%$
test_getitem[range]	0.1273ms	35.7358μs	27.9832 KOps/s	28.5043 KOps/s	$\color{#d91a1a}-1.83\%$
test_getitem[tuple]	0.1095ms	17.8346μs	56.0708 KOps/s	56.8849 KOps/s	$\color{#d91a1a}-1.43\%$
test_getitem[list]	0.1258ms	31.0987μs	32.1557 KOps/s	31.9050 KOps/s	$\color{#35bf28}+0.79\%$
test_setitem_dim[int]	39.5410μs	19.4747μs	51.3488 KOps/s	58.3895 KOps/s	$\textbf{\color{#d91a1a}-12.06\%}$
test_setitem_dim[slice_int]	60.7110μs	39.4593μs	25.3425 KOps/s	27.7685 KOps/s	$\textbf{\color{#d91a1a}-8.74\%}$
test_setitem_dim[range]	83.3420μs	53.1100μs	18.8288 KOps/s	19.9883 KOps/s	$\textbf{\color{#d91a1a}-5.80\%}$
test_setitem_dim[tuple]	53.6610μs	32.3456μs	30.9161 KOps/s	33.4253 KOps/s	$\textbf{\color{#d91a1a}-7.51\%}$
test_setitem	0.1292ms	14.5082μs	68.9267 KOps/s	67.6088 KOps/s	$\color{#35bf28}+1.95\%$
test_set	0.1264ms	13.1353μs	76.1305 KOps/s	67.5619 KOps/s	$\textbf{\color{#35bf28}+12.68\%}$
test_set_shared	1.5529ms	0.1502ms	6.6581 KOps/s	6.6609 KOps/s	$\color{#d91a1a}-0.04\%$
test_update	0.5479ms	14.9460μs	66.9075 KOps/s	54.1541 KOps/s	$\textbf{\color{#35bf28}+23.55\%}$
test_update_nested	0.1214ms	20.1452μs	49.6396 KOps/s	42.6881 KOps/s	$\textbf{\color{#35bf28}+16.28\%}$
test_update__nested	0.5557ms	24.2966μs	41.1581 KOps/s	41.5786 KOps/s	$\color{#d91a1a}-1.01\%$
test_set_nested	0.1217ms	14.2145μs	70.3508 KOps/s	62.4676 KOps/s	$\textbf{\color{#35bf28}+12.62\%}$
test_set_nested_new	0.1252ms	16.4815μs	60.6739 KOps/s	54.4067 KOps/s	$\textbf{\color{#35bf28}+11.52\%}$
test_select	0.1626ms	27.7532μs	36.0319 KOps/s	33.1189 KOps/s	$\textbf{\color{#35bf28}+8.80\%}$
test_select_nested	67.9810μs	43.1336μs	23.1838 KOps/s	22.8462 KOps/s	$\color{#35bf28}+1.48\%$
test_exclude_nested	92.5610μs	60.9812μs	16.3985 KOps/s	16.2323 KOps/s	$\color{#35bf28}+1.02\%$
test_empty[True]	0.3180ms	0.2846ms	3.5137 KOps/s	3.4351 KOps/s	$\color{#35bf28}+2.29\%$
test_empty[False]	3.9140μs	0.8302μs	1.2045 MOps/s	1.2157 MOps/s	$\color{#d91a1a}-0.92\%$
test_to	84.2220μs	55.1751μs	18.1241 KOps/s	17.9590 KOps/s	$\color{#35bf28}+0.92\%$
test_to_nonblocking	94.8210μs	49.2319μs	20.3121 KOps/s	21.3437 KOps/s	$\color{#d91a1a}-4.83\%$
test_unbind_speed	1.6725ms	0.2327ms	4.2968 KOps/s	4.3701 KOps/s	$\color{#d91a1a}-1.68\%$
test_unbind_speed_stack0	0.2897ms	0.2350ms	4.2548 KOps/s	4.3059 KOps/s	$\color{#d91a1a}-1.19\%$
test_unbind_speed_stack1	93.4147ms	0.6653ms	1.5031 KOps/s	1.5238 KOps/s	$\color{#d91a1a}-1.36\%$
test_split	94.8030ms	1.5818ms	632.1982 Ops/s	588.0183 Ops/s	$\textbf{\color{#35bf28}+7.51\%}$
test_chunk	94.9267ms	1.5839ms	631.3413 Ops/s	701.3201 Ops/s	$\textbf{\color{#d91a1a}-9.98\%}$
test_consolidate[False-None]	97.4774ms	2.8930ms	345.6674 Ops/s	341.1321 Ops/s	$\color{#35bf28}+1.33\%$
test_consolidate[default-None]	1.7385ms	1.6287ms	613.9828 Ops/s	612.8844 Ops/s	$\color{#35bf28}+0.18\%$
test_consolidate[reduce-overhead-None]	2.0543ms	1.6567ms	603.5997 Ops/s	596.4995 Ops/s	$\color{#35bf28}+1.19\%$
test_consolidate_njt[False-None]	6.6218ms	6.1975ms	161.3543 Ops/s	161.9332 Ops/s	$\color{#d91a1a}-0.36\%$
test_to[False-False-None]	2.0846ms	1.6707ms	598.5464 Ops/s	584.6755 Ops/s	$\color{#35bf28}+2.37\%$
test_to[True-False-None]	1.6664ms	1.2325ms	811.3674 Ops/s	811.7336 Ops/s	$\color{#d91a1a}-0.05\%$
test_to[within-False-None]	4.3362ms	3.9379ms	253.9437 Ops/s	250.9985 Ops/s	$\color{#35bf28}+1.17\%$
test_to[True-default-None]	5.4126ms	5.0060ms	199.7614 Ops/s	201.4812 Ops/s	$\color{#d91a1a}-0.85\%$
test_to_njt[False-False-None]	7.1342ms	6.7495ms	148.1597 Ops/s	149.1523 Ops/s	$\color{#d91a1a}-0.67\%$
test_to_njt[True-False-None]	5.6006ms	5.1813ms	193.0029 Ops/s	193.7095 Ops/s	$\color{#d91a1a}-0.36\%$
test_to_njt[within-False-None]	11.8562ms	11.4682ms	87.1979 Ops/s	87.9750 Ops/s	$\color{#d91a1a}-0.88\%$
test_creation[device0]	0.5437ms	78.6331μs	12.7173 KOps/s	12.8023 KOps/s	$\color{#d91a1a}-0.66\%$
test_creation_from_tensor	0.6700ms	81.7413μs	12.2337 KOps/s	12.2106 KOps/s	$\color{#35bf28}+0.19\%$
test_add_one[memmap_tensor0]	0.3183ms	6.4353μs	155.3920 KOps/s	156.6005 KOps/s	$\color{#d91a1a}-0.77\%$
test_contiguous[memmap_tensor0]	20.2948μs	0.3952μs	2.5306 MOps/s	2.4889 MOps/s	$\color{#35bf28}+1.67\%$
test_stack[memmap_tensor0]	21.9110μs	4.1377μs	241.6785 KOps/s	241.0583 KOps/s	$\color{#35bf28}+0.26\%$
test_memmaptd_index	1.6827ms	0.2371ms	4.2171 KOps/s	4.1992 KOps/s	$\color{#35bf28}+0.43\%$
test_memmaptd_index_astensor	0.5696ms	0.2959ms	3.3793 KOps/s	3.3601 KOps/s	$\color{#35bf28}+0.57\%$
test_memmaptd_index_op	0.9905ms	0.5235ms	1.9104 KOps/s	1.7387 KOps/s	$\textbf{\color{#35bf28}+9.88\%}$
test_serialize_model	0.1320s	0.1310s	7.6319 Ops/s	7.6805 Ops/s	$\color{#d91a1a}-0.63\%$
test_serialize_model_pickle	1.3661s	1.2149s	0.8231 Ops/s	0.8240 Ops/s	$\color{#d91a1a}-0.11\%$
test_serialize_weights	0.4169s	0.1713s	5.8365 Ops/s	7.7004 Ops/s	$\textbf{\color{#d91a1a}-24.21\%}$
test_serialize_weights_returnearly	0.3353s	53.5928ms	18.6592 Ops/s	11.2767 Ops/s	$\textbf{\color{#35bf28}+65.47\%}$
test_serialize_weights_pickle	1.3779s	1.2236s	0.8172 Ops/s	0.8376 Ops/s	$\color{#d91a1a}-2.43\%$
test_reshape_pytree	65.5710μs	21.7255μs	46.0288 KOps/s	46.3895 KOps/s	$\color{#d91a1a}-0.78\%$
test_reshape_td	47.4510μs	25.9695μs	38.5067 KOps/s	38.4272 KOps/s	$\color{#35bf28}+0.21\%$
test_view_pytree	48.7810μs	21.7494μs	45.9784 KOps/s	46.8612 KOps/s	$\color{#d91a1a}-1.88\%$
test_view_td	56.2010μs	28.8519μs	34.6598 KOps/s	32.2523 KOps/s	$\textbf{\color{#35bf28}+7.46\%}$
test_unbind_pytree	55.1910μs	27.5688μs	36.2729 KOps/s	36.6336 KOps/s	$\color{#d91a1a}-0.98\%$
test_unbind_td	0.7599ms	35.6447μs	28.0547 KOps/s	28.4816 KOps/s	$\color{#d91a1a}-1.50\%$
test_split_pytree	58.2510μs	29.4295μs	33.9795 KOps/s	34.2073 KOps/s	$\color{#d91a1a}-0.67\%$
test_split_td	0.9588ms	37.3500μs	26.7737 KOps/s	26.4918 KOps/s	$\color{#35bf28}+1.06\%$
test_add_pytree	64.2210μs	33.3374μs	29.9963 KOps/s	30.5967 KOps/s	$\color{#d91a1a}-1.96\%$
test_add_td	85.1010μs	44.5908μs	22.4261 KOps/s	21.1410 KOps/s	$\textbf{\color{#35bf28}+6.08\%}$
test_compile_add_one_nested[tensordict-compile]	0.1689ms	0.1216ms	8.2228 KOps/s	8.2794 KOps/s	$\color{#d91a1a}-0.68\%$
test_compile_add_one_nested[tensordict-eager]	0.2184ms	0.1287ms	7.7723 KOps/s	7.8883 KOps/s	$\color{#d91a1a}-1.47\%$
test_compile_add_one_nested[pytree-compile]	0.1390ms	96.6842μs	10.3430 KOps/s	10.4600 KOps/s	$\color{#d91a1a}-1.12\%$
test_compile_add_one_nested[pytree-eager]	2.2948ms	0.1451ms	6.8908 KOps/s	6.9369 KOps/s	$\color{#d91a1a}-0.66\%$
test_compile_copy_nested[tensordict-compile]	62.8410μs	23.8847μs	41.8678 KOps/s	47.2592 KOps/s	$\textbf{\color{#d91a1a}-11.41\%}$
test_compile_copy_nested[tensordict-eager]	52.5210μs	28.7803μs	34.7460 KOps/s	34.5092 KOps/s	$\color{#35bf28}+0.69\%$
test_compile_copy_nested[pytree-compile]	0.3892ms	63.3215μs	15.7924 KOps/s	15.6874 KOps/s	$\color{#35bf28}+0.67\%$
test_compile_copy_nested[pytree-eager]	83.4810μs	48.8205μs	20.4832 KOps/s	20.2041 KOps/s	$\color{#35bf28}+1.38\%$
test_compile_add_one_flat[tensordict-compile]	0.1829ms	0.1414ms	7.0742 KOps/s	7.1401 KOps/s	$\color{#d91a1a}-0.92\%$
test_compile_add_one_flat[tensordict-eager]	0.3118ms	0.2142ms	4.6692 KOps/s	4.7505 KOps/s	$\color{#d91a1a}-1.71\%$
test_compile_add_one_flat[tensorclass-compile]	0.1493ms	0.1001ms	9.9919 KOps/s	10.4791 KOps/s	$\color{#d91a1a}-4.65\%$
test_compile_add_one_flat[tensorclass-eager]	0.1151ms	55.2482μs	18.1001 KOps/s	19.3240 KOps/s	$\textbf{\color{#d91a1a}-6.33\%}$
test_compile_add_one_flat[pytree-compile]	0.2735ms	0.1342ms	7.4500 KOps/s	7.4865 KOps/s	$\color{#d91a1a}-0.49\%$
test_compile_add_one_flat[pytree-eager]	0.4979ms	0.4605ms	2.1716 KOps/s	2.1576 KOps/s	$\color{#35bf28}+0.65\%$
test_compile_add_self_flat[tensordict-eager]	0.3710ms	0.2578ms	3.8784 KOps/s	3.9053 KOps/s	$\color{#d91a1a}-0.69\%$
test_compile_add_self_flat[tensordict-compile]	0.1893ms	0.1422ms	7.0301 KOps/s	7.1303 KOps/s	$\color{#d91a1a}-1.40\%$
test_compile_add_self_flat[tensorclass-eager]	0.1485ms	68.3549μs	14.6295 KOps/s	15.9619 KOps/s	$\textbf{\color{#d91a1a}-8.35\%}$
test_compile_add_self_flat[tensorclass-compile]	0.1376ms	98.9682μs	10.1043 KOps/s	10.3925 KOps/s	$\color{#d91a1a}-2.77\%$
test_compile_add_self_flat[pytree-eager]	0.4575ms	0.3936ms	2.5404 KOps/s	2.5162 KOps/s	$\color{#35bf28}+0.96\%$
test_compile_add_self_flat[pytree-compile]	0.1712ms	0.1342ms	7.4508 KOps/s	7.5820 KOps/s	$\color{#d91a1a}-1.73\%$
test_compile_copy_flat[tensordict-compile]	46.6710μs	19.4782μs	51.3394 KOps/s	58.8694 KOps/s	$\textbf{\color{#d91a1a}-12.79\%}$
test_compile_copy_flat[tensordict-eager]	58.8810μs	31.0452μs	32.2111 KOps/s	32.3612 KOps/s	$\color{#d91a1a}-0.46\%$
test_compile_copy_flat[pytree-compile]	0.2198ms	70.6344μs	14.1574 KOps/s	14.4122 KOps/s	$\color{#d91a1a}-1.77\%$
test_compile_copy_flat[pytree-eager]	81.3110μs	51.9152μs	19.2622 KOps/s	19.5329 KOps/s	$\color{#d91a1a}-1.39\%$
test_compile_assign_and_add[tensordict-compile]	1.5901ms	0.3833ms	2.6093 KOps/s	2.2918 KOps/s	$\textbf{\color{#35bf28}+13.85\%}$
test_compile_assign_and_add[tensordict-eager]	2.6451ms	2.5521ms	391.8343 Ops/s	386.9012 Ops/s	$\color{#35bf28}+1.28\%$
test_compile_assign_and_add[pytree-compile]	1.5714ms	0.4311ms	2.3197 KOps/s	2.3203 KOps/s	$\color{#d91a1a}-0.03\%$
test_compile_assign_and_add[pytree-eager]	2.7349ms	2.5585ms	390.8599 Ops/s	384.8160 Ops/s	$\color{#35bf28}+1.57\%$
test_compile_indexing[tensor-tensordict-compile]	0.1687ms	0.1142ms	8.7552 KOps/s	9.0499 KOps/s	$\color{#d91a1a}-3.26\%$
test_compile_indexing[tensor-tensordict-eager]	0.5791ms	78.3966μs	12.7557 KOps/s	12.8341 KOps/s	$\color{#d91a1a}-0.61\%$
test_compile_indexing[tensor-tensorclass-compile]	0.2136ms	0.1071ms	9.3342 KOps/s	9.7249 KOps/s	$\color{#d91a1a}-4.02\%$
test_compile_indexing[tensor-tensorclass-eager]	0.5082ms	68.7782μs	14.5395 KOps/s	14.8178 KOps/s	$\color{#d91a1a}-1.88\%$
test_compile_indexing[tensor-pytree-compile]	0.5607ms	0.1101ms	9.0855 KOps/s	9.6647 KOps/s	$\textbf{\color{#d91a1a}-5.99\%}$
test_compile_indexing[tensor-pytree-eager]	0.4706ms	68.3683μs	14.6267 KOps/s	14.1395 KOps/s	$\color{#35bf28}+3.45\%$
test_compile_indexing[slice-tensordict-compile]	0.5180ms	0.1030ms	9.7113 KOps/s	10.2092 KOps/s	$\color{#d91a1a}-4.88\%$
test_compile_indexing[slice-tensordict-eager]	0.1428ms	15.8363μs	63.1460 KOps/s	60.7160 KOps/s	$\color{#35bf28}+4.00\%$
test_compile_indexing[slice-tensorclass-compile]	0.3531ms	95.0011μs	10.5262 KOps/s	10.7237 KOps/s	$\color{#d91a1a}-1.84\%$
test_compile_indexing[slice-tensorclass-eager]	41.3410μs	15.4608μs	64.6797 KOps/s	63.2509 KOps/s	$\color{#35bf28}+2.26\%$
test_compile_indexing[slice-pytree-compile]	0.1396ms	98.6098μs	10.1410 KOps/s	10.5923 KOps/s	$\color{#d91a1a}-4.26\%$
test_compile_indexing[slice-pytree-eager]	50.0710μs	15.4365μs	64.7816 KOps/s	64.2667 KOps/s	$\color{#35bf28}+0.80\%$
test_compile_indexing[int-tensordict-compile]	0.1401ms	98.6363μs	10.1383 KOps/s	10.0931 KOps/s	$\color{#35bf28}+0.45\%$
test_compile_indexing[int-tensordict-eager]	0.5705ms	16.3988μs	60.9800 KOps/s	58.5174 KOps/s	$\color{#35bf28}+4.21\%$
test_compile_indexing[int-tensorclass-compile]	0.1389ms	95.3209μs	10.4909 KOps/s	10.5853 KOps/s	$\color{#d91a1a}-0.89\%$
test_compile_indexing[int-tensorclass-eager]	51.4400μs	15.2819μs	65.4367 KOps/s	64.5751 KOps/s	$\color{#35bf28}+1.33\%$
test_compile_indexing[int-pytree-compile]	0.1479ms	95.0216μs	10.5239 KOps/s	10.6002 KOps/s	$\color{#d91a1a}-0.72\%$
test_compile_indexing[int-pytree-eager]	44.8610μs	15.3757μs	65.0378 KOps/s	64.4444 KOps/s	$\color{#35bf28}+0.92\%$
test_mod_add[eager]	0.1684ms	35.1677μs	28.4352 KOps/s	26.3182 KOps/s	$\textbf{\color{#35bf28}+8.04\%}$
test_mod_add[compile]	0.2132ms	81.6481μs	12.2477 KOps/s	12.8050 KOps/s	$\color{#d91a1a}-4.35\%$
test_mod_add[compile-overhead]	0.3185ms	0.1697ms	5.8910 KOps/s	5.7977 KOps/s	$\color{#35bf28}+1.61\%$
test_mod_wrap[eager]	0.3205ms	0.2388ms	4.1882 KOps/s	4.0474 KOps/s	$\color{#35bf28}+3.48\%$
test_mod_wrap[compile]	0.3483ms	0.2795ms	3.5779 KOps/s	3.4559 KOps/s	$\color{#35bf28}+3.53\%$
test_mod_wrap[compile-overhead]	7.1909ms	3.6595ms	273.2603 Ops/s	274.5701 Ops/s	$\color{#d91a1a}-0.48\%$
test_mod_wrap_and_backward[eager]	1.6099ms	1.3477ms	742.0096 Ops/s	724.4654 Ops/s	$\color{#35bf28}+2.42\%$
test_mod_wrap_and_backward[compile]	1.3452ms	1.2255ms	816.0079 Ops/s	799.9640 Ops/s	$\color{#35bf28}+2.01\%$
test_mod_wrap_and_backward[compile-overhead]	1.3437ms	0.9103ms	1.0986 KOps/s	1.0770 KOps/s	$\color{#35bf28}+2.00\%$
test_seq_add[eager]	0.2098ms	0.1062ms	9.4150 KOps/s	8.8469 KOps/s	$\textbf{\color{#35bf28}+6.42\%}$
test_seq_add[compile]	0.1836ms	85.3461μs	11.7170 KOps/s	11.6339 KOps/s	$\color{#35bf28}+0.71\%$
test_seq_add[compile-overhead]	0.2349ms	0.1317ms	7.5933 KOps/s	7.9350 KOps/s	$\color{#d91a1a}-4.31\%$
test_seq_wrap[eager]	0.5209ms	0.4112ms	2.4319 KOps/s	2.4195 KOps/s	$\color{#35bf28}+0.52\%$
test_seq_wrap[compile]	0.4052ms	0.2928ms	3.4150 KOps/s	3.4236 KOps/s	$\color{#d91a1a}-0.25\%$
test_seq_wrap[compile-overhead]	0.3050ms	0.2187ms	4.5721 KOps/s	4.5785 KOps/s	$\color{#d91a1a}-0.14\%$
test_func_call_runtime[False-eager]	0.7837ms	0.7138ms	1.4009 KOps/s	1.3911 KOps/s	$\color{#35bf28}+0.70\%$
test_func_call_runtime[False-compile]	0.8063ms	0.7179ms	1.3929 KOps/s	1.3970 KOps/s	$\color{#d91a1a}-0.29\%$
test_func_call_runtime[False-compile-overhead]	0.4048ms	0.3526ms	2.8359 KOps/s	2.8504 KOps/s	$\color{#d91a1a}-0.51\%$
test_func_call_runtime[True-eager]	0.9561ms	0.8805ms	1.1357 KOps/s	1.1346 KOps/s	$\color{#35bf28}+0.10\%$
test_func_call_runtime[True-compile]	0.8135ms	0.7367ms	1.3574 KOps/s	1.3538 KOps/s	$\color{#35bf28}+0.27\%$
test_func_call_runtime[True-compile-overhead]	0.4468ms	0.3747ms	2.6690 KOps/s	2.6843 KOps/s	$\color{#d91a1a}-0.57\%$
test_func_call_cm_runtime[False-eager]	0.7687ms	0.7150ms	1.3985 KOps/s	1.4007 KOps/s	$\color{#d91a1a}-0.16\%$
test_func_call_cm_runtime[False-compile]	0.8367ms	0.7216ms	1.3858 KOps/s	1.3925 KOps/s	$\color{#d91a1a}-0.48\%$
test_func_call_cm_runtime[False-compile-overhead]	0.3960ms	0.3562ms	2.8072 KOps/s	2.8370 KOps/s	$\color{#d91a1a}-1.05\%$
test_func_call_cm_runtime[True-eager]	1.0663ms	0.9767ms	1.0238 KOps/s	1.0076 KOps/s	$\color{#35bf28}+1.61\%$
test_func_call_cm_runtime[True-compile]	0.8037ms	0.7644ms	1.3083 KOps/s	1.3083 KOps/s	$-0.01\%$
test_func_call_cm_runtime[True-compile-overhead]	0.4470ms	0.3987ms	2.5081 KOps/s	2.4996 KOps/s	$\color{#35bf28}+0.34\%$
test_vmap_func_call_cm_runtime[eager]	2.4867ms	2.0437ms	489.3026 Ops/s	482.8714 Ops/s	$\color{#35bf28}+1.33\%$
test_vmap_func_call_cm_runtime[compile]	0.8365ms	0.7822ms	1.2784 KOps/s	1.2793 KOps/s	$\color{#d91a1a}-0.07\%$
test_vmap_func_call_cm_runtime[compile-overhead]	0.4524ms	0.4029ms	2.4818 KOps/s	2.4782 KOps/s	$\color{#35bf28}+0.14\%$
test_distributed	3.1187ms	0.2071ms	4.8277 KOps/s	7.8828 KOps/s	$\textbf{\color{#d91a1a}-38.76\%}$
test_tdmodule	0.3577ms	18.9214μs	52.8503 KOps/s	51.4240 KOps/s	$\color{#35bf28}+2.77\%$
test_tdmodule_dispatch	55.2110μs	32.8150μs	30.4739 KOps/s	28.1731 KOps/s	$\textbf{\color{#35bf28}+8.17\%}$
test_tdseq	27.5000μs	18.9159μs	52.8654 KOps/s	47.4550 KOps/s	$\textbf{\color{#35bf28}+11.40\%}$
test_tdseq_dispatch	62.8410μs	35.2202μs	28.3928 KOps/s	25.6071 KOps/s	$\textbf{\color{#35bf28}+10.88\%}$
test_instantiation_functorch	1.6245ms	1.5125ms	661.1401 Ops/s	669.1552 Ops/s	$\color{#d91a1a}-1.20\%$
test_exec_functorch	0.1809ms	0.1391ms	7.1872 KOps/s	7.1912 KOps/s	$\color{#d91a1a}-0.05\%$
test_exec_functional_call	0.1734ms	0.1335ms	7.4919 KOps/s	7.6330 KOps/s	$\color{#d91a1a}-1.85\%$
test_exec_td_decorator	0.4056ms	0.1815ms	5.5093 KOps/s	5.5332 KOps/s	$\color{#d91a1a}-0.43\%$
test_vmap_mlp_speed_decorator[True-True]	0.7601ms	0.6697ms	1.4931 KOps/s	1.4824 KOps/s	$\color{#35bf28}+0.72\%$
test_vmap_mlp_speed_decorator[True-False]	0.8448ms	0.6711ms	1.4902 KOps/s	1.4802 KOps/s	$\color{#35bf28}+0.67\%$
test_vmap_mlp_speed_decorator[False-True]	0.7030ms	0.5836ms	1.7136 KOps/s	1.7081 KOps/s	$\color{#35bf28}+0.32\%$
test_vmap_mlp_speed_decorator[False-False]	0.7018ms	0.5818ms	1.7188 KOps/s	1.7034 KOps/s	$\color{#35bf28}+0.90\%$
test_vmap_transformer_speed_decorator[True-True]	19.2270ms	18.8864ms	52.9482 Ops/s	52.6929 Ops/s	$\color{#35bf28}+0.48\%$
test_vmap_transformer_speed_decorator[True-False]	19.0696ms	18.9503ms	52.7697 Ops/s	52.5462 Ops/s	$\color{#35bf28}+0.43\%$
test_vmap_transformer_speed_decorator[False-True]	18.9485ms	18.8230ms	53.1264 Ops/s	53.0754 Ops/s	$\color{#35bf28}+0.10\%$
test_vmap_transformer_speed_decorator[False-False]	18.8633ms	18.7853ms	53.2331 Ops/s	52.9574 Ops/s	$\color{#35bf28}+0.52\%$
test_to_module_speed[True]	1.0687ms	0.9605ms	1.0412 KOps/s	1.0509 KOps/s	$\color{#d91a1a}-0.93\%$
test_to_module_speed[False]	1.3370ms	0.9553ms	1.0468 KOps/s	1.0759 KOps/s	$\color{#d91a1a}-2.70\%$
test_tc_init	64.1310μs	32.3728μs	30.8901 KOps/s	29.0539 KOps/s	$\textbf{\color{#35bf28}+6.32\%}$
test_tc_init_nested	0.1137ms	66.1733μs	15.1118 KOps/s	14.2634 KOps/s	$\textbf{\color{#35bf28}+5.95\%}$
test_tc_first_layer_tensor	5.0843μs	0.7028μs	1.4228 MOps/s	1.4667 MOps/s	$\color{#d91a1a}-2.99\%$
test_tc_first_layer_nontensor	23.8200μs	2.2436μs	445.7060 KOps/s	454.9922 KOps/s	$\color{#d91a1a}-2.04\%$
test_tc_second_layer_tensor	9.5533μs	1.4446μs	692.2125 KOps/s	710.6785 KOps/s	$\color{#d91a1a}-2.60\%$
test_tc_second_layer_nontensor	0.1388ms	2.9962μs	333.7574 KOps/s	337.1868 KOps/s	$\color{#d91a1a}-1.02\%$
test_unbind	0.2217s	10.1712ms	98.3170 Ops/s	146.2863 Ops/s	$\textbf{\color{#d91a1a}-32.79\%}$
test_full_like	9.8135ms	9.1927ms	108.7818 Ops/s	107.6308 Ops/s	$\color{#35bf28}+1.07\%$
test_zeros_like	4.8998ms	4.3290ms	231.0023 Ops/s	234.6626 Ops/s	$\color{#d91a1a}-1.56\%$
test_ones_like	4.6579ms	4.3359ms	230.6325 Ops/s	230.7104 Ops/s	$\color{#d91a1a}-0.03\%$
test_clone	6.7027ms	6.4046ms	156.1377 Ops/s	109.5115 Ops/s	$\textbf{\color{#35bf28}+42.58\%}$
test_squeeze	57.8410μs	9.5536μs	104.6730 KOps/s	110.5279 KOps/s	$\textbf{\color{#d91a1a}-5.30\%}$
test_unsqueeze	5.0006ms	71.4757μs	13.9908 KOps/s	14.9839 KOps/s	$\textbf{\color{#d91a1a}-6.63\%}$
test_split	0.2526ms	0.1566ms	6.3858 KOps/s	6.6421 KOps/s	$\color{#d91a1a}-3.86\%$
test_permute	0.2174ms	0.1751ms	5.7120 KOps/s	5.8879 KOps/s	$\color{#d91a1a}-2.99\%$
test_stack	51.2982ms	50.7511ms	19.7040 Ops/s	19.5978 Ops/s	$\color{#35bf28}+0.54\%$
test_cat	51.0116ms	50.6818ms	19.7309 Ops/s	19.7718 Ops/s	$\color{#d91a1a}-0.21\%$

[vmoens]

The following behavior is deprecated as part of this PR:

td0 = TensorDict(batch_size=[3, 4])
td1 = TensorDict(batch_size=[3, 4, 2])
td0 == td1

Previously, __eq__ and other unitary methods didn't look at the batch size of the TDs. Now, the TD shapes will be broadcast - in this example no compatible shape exists and an error is raised.

This change is necessary as this would be the expected behaviour:

td0 = TensorDict(a=torch.randn(3, 4), batch_size=[3, 4])
td1 = TensorDict(a=torch.randn(4), batch_size=[4])
td0 == td1 # works because td1 is broadcast to (3, 4)