[Refactor] Put values, lengths and offsets of NJTs together in storage

[vmoens]

Stack from ghstack (oldest at bottom):

• [Performance] Faster clone #1040
• -> [Refactor] Put values, lengths and offsets of NJTs together in storage #1023
• [Feature] TD+NJT to(device) support #1022

[github-actions]

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

Total Benchmarks: 216. Improved: $\large\color{#35bf28}45$. Worsened: $\large\color{#d91a1a}10$.

Expand to view detailed results

Name	Max	Mean	Ops	Ops on Repo HEAD	Change
test_plain_set_nested	49.0610μs	23.5053μs	42.5436 KOps/s	39.1417 KOps/s	$\textbf{\color{#35bf28}+8.69\%}$
test_plain_set_stack_nested	52.0170μs	23.6152μs	42.3457 KOps/s	38.4767 KOps/s	$\textbf{\color{#35bf28}+10.06\%}$
test_plain_set_nested_inplace	75.7610μs	25.7347μs	38.8580 KOps/s	35.3896 KOps/s	$\textbf{\color{#35bf28}+9.80\%}$
test_plain_set_stack_nested_inplace	66.6240μs	25.6589μs	38.9728 KOps/s	35.4439 KOps/s	$\textbf{\color{#35bf28}+9.96\%}$
test_items	23.1530μs	4.2308μs	236.3605 KOps/s	243.3109 KOps/s	$\color{#d91a1a}-2.86\%$
test_items_nested	0.5120ms	0.3846ms	2.6001 KOps/s	2.4514 KOps/s	$\textbf{\color{#35bf28}+6.07\%}$
test_items_nested_locked	0.5626ms	0.3830ms	2.6113 KOps/s	2.5881 KOps/s	$\color{#35bf28}+0.90\%$
test_items_nested_leaf	0.1561ms	80.2421μs	12.4623 KOps/s	12.2595 KOps/s	$\color{#35bf28}+1.65\%$
test_items_stack_nested	0.5271ms	0.3849ms	2.5983 KOps/s	2.5406 KOps/s	$\color{#35bf28}+2.27\%$
test_items_stack_nested_leaf	0.1499ms	83.1392μs	12.0280 KOps/s	11.7947 KOps/s	$\color{#35bf28}+1.98\%$
test_items_stack_nested_locked	0.5837ms	0.3837ms	2.6061 KOps/s	2.5818 KOps/s	$\color{#35bf28}+0.94\%$
test_keys	36.3170μs	3.5698μs	280.1312 KOps/s	283.5856 KOps/s	$\color{#d91a1a}-1.22\%$
test_keys_nested	0.2309ms	0.1336ms	7.4870 KOps/s	7.1516 KOps/s	$\color{#35bf28}+4.69\%$
test_keys_nested_locked	0.7723ms	0.1386ms	7.2163 KOps/s	6.9005 KOps/s	$\color{#35bf28}+4.58\%$
test_keys_nested_leaf	0.2032ms	0.1167ms	8.5726 KOps/s	8.1743 KOps/s	$\color{#35bf28}+4.87\%$
test_keys_stack_nested	0.2267ms	0.1346ms	7.4292 KOps/s	7.1854 KOps/s	$\color{#35bf28}+3.39\%$
test_keys_stack_nested_leaf	0.2002ms	0.1172ms	8.5342 KOps/s	8.4332 KOps/s	$\color{#35bf28}+1.20\%$
test_keys_stack_nested_locked	0.2335ms	0.1387ms	7.2113 KOps/s	6.9625 KOps/s	$\color{#35bf28}+3.57\%$
test_values	7.8486μs	1.0476μs	954.5256 KOps/s	939.6585 KOps/s	$\color{#35bf28}+1.58\%$
test_values_nested	0.1545ms	93.4116μs	10.7053 KOps/s	10.3933 KOps/s	$\color{#35bf28}+3.00\%$
test_values_nested_locked	0.1571ms	93.3294μs	10.7147 KOps/s	10.3669 KOps/s	$\color{#35bf28}+3.36\%$
test_values_nested_leaf	0.1441ms	80.2601μs	12.4595 KOps/s	12.0563 KOps/s	$\color{#35bf28}+3.34\%$
test_values_stack_nested	0.1582ms	92.6340μs	10.7952 KOps/s	10.1839 KOps/s	$\textbf{\color{#35bf28}+6.00\%}$
test_values_stack_nested_leaf	0.1459ms	79.6106μs	12.5611 KOps/s	12.0492 KOps/s	$\color{#35bf28}+4.25\%$
test_values_stack_nested_locked	0.1616ms	93.8749μs	10.6525 KOps/s	10.4612 KOps/s	$\color{#35bf28}+1.83\%$
test_membership	5.4416μs	0.7386μs	1.3539 MOps/s	1.3163 MOps/s	$\color{#35bf28}+2.86\%$
test_membership_nested	37.7400μs	2.7543μs	363.0657 KOps/s	362.9317 KOps/s	$\color{#35bf28}+0.04\%$
test_membership_nested_leaf	25.7080μs	2.7826μs	359.3811 KOps/s	356.3370 KOps/s	$\color{#35bf28}+0.85\%$
test_membership_stacked_nested	28.2120μs	2.7079μs	369.2878 KOps/s	367.7906 KOps/s	$\color{#35bf28}+0.41\%$
test_membership_stacked_nested_leaf	17.8530μs	2.7418μs	364.7223 KOps/s	357.9529 KOps/s	$\color{#35bf28}+1.89\%$
test_membership_nested_last	40.4740μs	4.1380μs	241.6627 KOps/s	237.1864 KOps/s	$\color{#35bf28}+1.89\%$
test_membership_nested_leaf_last	27.5610μs	4.1981μs	238.2009 KOps/s	239.6246 KOps/s	$\color{#d91a1a}-0.59\%$
test_membership_stacked_nested_last	35.1760μs	4.8159μs	207.6446 KOps/s	131.6535 KOps/s	$\textbf{\color{#35bf28}+57.72\%}$
test_membership_stacked_nested_leaf_last	27.7710μs	4.9234μs	203.1109 KOps/s	132.1912 KOps/s	$\textbf{\color{#35bf28}+53.65\%}$
test_nested_getleaf	37.1590μs	10.9382μs	91.4228 KOps/s	91.5765 KOps/s	$\color{#d91a1a}-0.17\%$
test_nested_get	46.8700μs	9.9937μs	100.0632 KOps/s	96.6923 KOps/s	$\color{#35bf28}+3.49\%$
test_stacked_getleaf	30.9170μs	10.5633μs	94.6678 KOps/s	92.3649 KOps/s	$\color{#35bf28}+2.49\%$
test_stacked_get	30.6480μs	10.0350μs	99.6512 KOps/s	87.4757 KOps/s	$\textbf{\color{#35bf28}+13.92\%}$
test_nested_getitemleaf	39.1630μs	10.9390μs	91.4164 KOps/s	88.6413 KOps/s	$\color{#35bf28}+3.13\%$
test_nested_getitem	45.1940μs	10.2140μs	97.9051 KOps/s	91.2810 KOps/s	$\textbf{\color{#35bf28}+7.26\%}$
test_stacked_getitemleaf	43.0300μs	10.9572μs	91.2644 KOps/s	90.1011 KOps/s	$\color{#35bf28}+1.29\%$
test_stacked_getitem	42.4790μs	10.2504μs	97.5571 KOps/s	95.8118 KOps/s	$\color{#35bf28}+1.82\%$
test_lock_nested	83.7694ms	0.5897ms	1.6957 KOps/s	1.9658 KOps/s	$\textbf{\color{#d91a1a}-13.74\%}$
test_lock_stack_nested	0.8638ms	0.4741ms	2.1094 KOps/s	2.1409 KOps/s	$\color{#d91a1a}-1.47\%$
test_unlock_nested	88.3304ms	0.5127ms	1.9504 KOps/s	2.3529 KOps/s	$\textbf{\color{#d91a1a}-17.11\%}$
test_unlock_stack_nested	0.6036ms	0.3875ms	2.5804 KOps/s	2.5931 KOps/s	$\color{#d91a1a}-0.49\%$
test_flatten_speed	0.1699ms	0.1009ms	9.9127 KOps/s	9.8866 KOps/s	$\color{#35bf28}+0.26\%$
test_unflatten_speed	1.0662ms	0.5114ms	1.9553 KOps/s	1.8977 KOps/s	$\color{#35bf28}+3.04\%$
test_common_ops	5.0005ms	1.1439ms	874.2084 Ops/s	849.5130 Ops/s	$\color{#35bf28}+2.91\%$
test_creation	83.2550μs	2.0554μs	486.5146 KOps/s	466.4653 KOps/s	$\color{#35bf28}+4.30\%$
test_creation_empty	63.1480μs	17.3041μs	57.7898 KOps/s	47.6935 KOps/s	$\textbf{\color{#35bf28}+21.17\%}$
test_creation_nested_1	93.9050μs	20.7416μs	48.2122 KOps/s	41.5624 KOps/s	$\textbf{\color{#35bf28}+16.00\%}$
test_creation_nested_2	69.0590μs	25.4009μs	39.3687 KOps/s	34.7097 KOps/s	$\textbf{\color{#35bf28}+13.42\%}$
test_clone	0.2069ms	16.8872μs	59.2165 KOps/s	56.8080 KOps/s	$\color{#35bf28}+4.24\%$
test_getitem[int]	1.0482ms	17.1227μs	58.4021 KOps/s	57.6856 KOps/s	$\color{#35bf28}+1.24\%$
test_getitem[slice_int]	0.1320ms	30.6116μs	32.6674 KOps/s	31.5211 KOps/s	$\color{#35bf28}+3.64\%$
test_getitem[range]	0.1785ms	57.7666μs	17.3110 KOps/s	17.2966 KOps/s	$\color{#35bf28}+0.08\%$
test_getitem[tuple]	0.1281ms	25.4129μs	39.3501 KOps/s	38.1120 KOps/s	$\color{#35bf28}+3.25\%$
test_getitem[list]	0.1795ms	54.1270μs	18.4751 KOps/s	18.8048 KOps/s	$\color{#d91a1a}-1.75\%$
test_setitem_dim[int]	75.9520μs	31.6939μs	31.5518 KOps/s	30.5202 KOps/s	$\color{#35bf28}+3.38\%$
test_setitem_dim[slice_int]	0.1109ms	59.4150μs	16.8308 KOps/s	16.1975 KOps/s	$\color{#35bf28}+3.91\%$
test_setitem_dim[range]	0.1265ms	82.5450μs	12.1146 KOps/s	11.7560 KOps/s	$\color{#35bf28}+3.05\%$
test_setitem_dim[tuple]	92.8430μs	47.6918μs	20.9680 KOps/s	20.3218 KOps/s	$\color{#35bf28}+3.18\%$
test_setitem	66.8550μs	29.3897μs	34.0256 KOps/s	31.9962 KOps/s	$\textbf{\color{#35bf28}+6.34\%}$
test_set	82.7140μs	29.1163μs	34.3450 KOps/s	32.6248 KOps/s	$\textbf{\color{#35bf28}+5.27\%}$
test_set_shared	3.8988ms	0.2177ms	4.5940 KOps/s	4.4901 KOps/s	$\color{#35bf28}+2.32\%$
test_update	0.2293ms	37.5342μs	26.6423 KOps/s	24.3149 KOps/s	$\textbf{\color{#35bf28}+9.57\%}$
test_update_nested	0.2333ms	48.8261μs	20.4808 KOps/s	19.5792 KOps/s	$\color{#35bf28}+4.60\%$
test_update__nested	0.5303ms	45.0250μs	22.2099 KOps/s	22.0291 KOps/s	$\color{#35bf28}+0.82\%$
test_set_nested	0.2418ms	32.0446μs	31.2065 KOps/s	29.4335 KOps/s	$\textbf{\color{#35bf28}+6.02\%}$
test_set_nested_new	82.2430μs	37.2347μs	26.8567 KOps/s	25.9132 KOps/s	$\color{#35bf28}+3.64\%$
test_select	0.2799ms	54.2937μs	18.4183 KOps/s	17.6881 KOps/s	$\color{#35bf28}+4.13\%$
test_select_nested	0.1273ms	61.0514μs	16.3796 KOps/s	16.6846 KOps/s	$\color{#d91a1a}-1.83\%$
test_exclude_nested	0.1586ms	75.1628μs	13.3045 KOps/s	13.2250 KOps/s	$\color{#35bf28}+0.60\%$
test_empty[True]	0.6481ms	0.3540ms	2.8251 KOps/s	2.8052 KOps/s	$\color{#35bf28}+0.71\%$
test_empty[False]	8.6462μs	1.2748μs	784.4391 KOps/s	809.3043 KOps/s	$\color{#d91a1a}-3.07\%$
test_unbind_speed	0.6478ms	0.3019ms	3.3119 KOps/s	3.2610 KOps/s	$\color{#35bf28}+1.56\%$
test_unbind_speed_stack0	0.6184ms	0.3032ms	3.2977 KOps/s	3.3980 KOps/s	$\color{#d91a1a}-2.95\%$
test_unbind_speed_stack1	0.1002s	0.8241ms	1.2135 KOps/s	1.3627 KOps/s	$\textbf{\color{#d91a1a}-10.95\%}$
test_split	3.0370ms	2.0072ms	498.2038 Ops/s	447.2298 Ops/s	$\textbf{\color{#35bf28}+11.40\%}$
test_chunk	99.2087ms	2.4158ms	413.9353 Ops/s	441.4800 Ops/s	$\textbf{\color{#d91a1a}-6.24\%}$
test_creation[device0]	0.2424ms	0.1153ms	8.6727 KOps/s	8.6340 KOps/s	$\color{#35bf28}+0.45\%$
test_creation_from_tensor	4.0644ms	0.1186ms	8.4292 KOps/s	8.4827 KOps/s	$\color{#d91a1a}-0.63\%$
test_add_one[memmap_tensor0]	78.0560μs	6.9743μs	143.3827 KOps/s	140.7443 KOps/s	$\color{#35bf28}+1.87\%$
test_contiguous[memmap_tensor0]	16.9310μs	1.9563μs	511.1746 KOps/s	448.4936 KOps/s	$\textbf{\color{#35bf28}+13.98\%}$
test_stack[memmap_tensor0]	29.2240μs	5.4897μs	182.1588 KOps/s	144.1686 KOps/s	$\textbf{\color{#35bf28}+26.35\%}$
test_memmaptd_index	1.1261ms	0.4188ms	2.3875 KOps/s	2.3605 KOps/s	$\color{#35bf28}+1.14\%$
test_memmaptd_index_astensor	1.1792ms	0.5196ms	1.9247 KOps/s	1.8909 KOps/s	$\color{#35bf28}+1.79\%$
test_memmaptd_index_op	1.8109ms	1.0249ms	975.6852 Ops/s	898.5474 Ops/s	$\textbf{\color{#35bf28}+8.58\%}$
test_serialize_model	0.1228s	0.1174s	8.5182 Ops/s	8.5903 Ops/s	$\color{#d91a1a}-0.84\%$
test_serialize_model_pickle	0.4448s	0.3889s	2.5714 Ops/s	2.4938 Ops/s	$\color{#35bf28}+3.11\%$
test_serialize_weights	0.1259s	0.1184s	8.4445 Ops/s	8.5402 Ops/s	$\color{#d91a1a}-1.12\%$
test_serialize_weights_returnearly	0.2702s	0.1740s	5.7472 Ops/s	6.4035 Ops/s	$\textbf{\color{#d91a1a}-10.25\%}$
test_serialize_weights_pickle	0.4813s	0.3882s	2.5762 Ops/s	1.0960 Ops/s	$\textbf{\color{#35bf28}+135.06\%}$
test_serialize_weights_filesystem	0.1455s	0.1407s	7.1093 Ops/s	7.0708 Ops/s	$\color{#35bf28}+0.55\%$
test_serialize_model_filesystem	0.1603s	0.1510s	6.6245 Ops/s	6.9618 Ops/s	$\color{#d91a1a}-4.85\%$
test_reshape_pytree	94.5970μs	39.2474μs	25.4794 KOps/s	24.7041 KOps/s	$\color{#35bf28}+3.14\%$
test_reshape_td	95.8190μs	48.0832μs	20.7973 KOps/s	20.6583 KOps/s	$\color{#35bf28}+0.67\%$
test_view_pytree	81.6420μs	39.1785μs	25.5242 KOps/s	24.6678 KOps/s	$\color{#35bf28}+3.47\%$
test_view_td	0.1071ms	53.4581μs	18.7062 KOps/s	18.3162 KOps/s	$\color{#35bf28}+2.13\%$
test_unbind_pytree	89.5570μs	35.9759μs	27.7964 KOps/s	26.3323 KOps/s	$\textbf{\color{#35bf28}+5.56\%}$
test_unbind_td	0.2898ms	45.0976μs	22.1741 KOps/s	21.5521 KOps/s	$\color{#35bf28}+2.89\%$
test_split_pytree	78.8570μs	38.3060μs	26.1056 KOps/s	25.1049 KOps/s	$\color{#35bf28}+3.99\%$
test_split_td	0.1985ms	59.7583μs	16.7341 KOps/s	17.0388 KOps/s	$\color{#d91a1a}-1.79\%$
test_add_pytree	0.1083ms	44.5684μs	22.4374 KOps/s	21.9518 KOps/s	$\color{#35bf28}+2.21\%$
test_add_td	0.6759ms	84.8142μs	11.7905 KOps/s	10.5725 KOps/s	$\textbf{\color{#35bf28}+11.52\%}$
test_compile_add_one_nested[tensordict-compile]	0.1426ms	59.7438μs	16.7381 KOps/s	16.1710 KOps/s	$\color{#35bf28}+3.51\%$
test_compile_add_one_nested[tensordict-eager]	0.9754ms	0.1975ms	5.0623 KOps/s	5.0544 KOps/s	$\color{#35bf28}+0.16\%$
test_compile_add_one_nested[pytree-compile]	0.4995ms	59.1942μs	16.8936 KOps/s	17.3048 KOps/s	$\color{#d91a1a}-2.38\%$
test_compile_add_one_nested[pytree-eager]	0.2727ms	0.1395ms	7.1670 KOps/s	7.0769 KOps/s	$\color{#35bf28}+1.27\%$
test_compile_copy_nested[tensordict-compile]	72.6560μs	24.9649μs	40.0562 KOps/s	42.3826 KOps/s	$\textbf{\color{#d91a1a}-5.49\%}$
test_compile_copy_nested[tensordict-eager]	0.1392ms	73.5309μs	13.5997 KOps/s	12.9529 KOps/s	$\color{#35bf28}+4.99\%$
test_compile_copy_nested[pytree-compile]	0.1715ms	76.7046μs	13.0370 KOps/s	12.7214 KOps/s	$\color{#35bf28}+2.48\%$
test_compile_copy_nested[pytree-eager]	0.4148ms	68.7638μs	14.5425 KOps/s	14.0121 KOps/s	$\color{#35bf28}+3.79\%$
test_compile_add_one_flat[tensordict-compile]	0.3275ms	0.1813ms	5.5167 KOps/s	5.4814 KOps/s	$\color{#35bf28}+0.64\%$
test_compile_add_one_flat[tensordict-eager]	0.3840ms	0.2358ms	4.2413 KOps/s	4.0446 KOps/s	$\color{#35bf28}+4.86\%$
test_compile_add_one_flat[tensorclass-compile]	0.1111ms	48.3547μs	20.6805 KOps/s	21.1166 KOps/s	$\color{#d91a1a}-2.07\%$
test_compile_add_one_flat[tensorclass-eager]	0.4334ms	76.4686μs	13.0773 KOps/s	10.7469 KOps/s	$\textbf{\color{#35bf28}+21.68\%}$
test_compile_add_one_flat[pytree-compile]	0.3696ms	0.1778ms	5.6238 KOps/s	5.6469 KOps/s	$\color{#d91a1a}-0.41\%$
test_compile_add_one_flat[pytree-eager]	0.5953ms	0.2846ms	3.5131 KOps/s	3.4617 KOps/s	$\color{#35bf28}+1.49\%$
test_compile_add_self_flat[tensordict-eager]	0.4031ms	0.2742ms	3.6464 KOps/s	3.4868 KOps/s	$\color{#35bf28}+4.58\%$
test_compile_add_self_flat[tensordict-compile]	0.5476ms	0.1874ms	5.3370 KOps/s	5.4416 KOps/s	$\color{#d91a1a}-1.92\%$
test_compile_add_self_flat[tensorclass-eager]	0.1868ms	74.3900μs	13.4427 KOps/s	13.5028 KOps/s	$\color{#d91a1a}-0.45\%$
test_compile_add_self_flat[tensorclass-compile]	0.1058ms	47.8239μs	20.9100 KOps/s	20.6431 KOps/s	$\color{#35bf28}+1.29\%$
test_compile_add_self_flat[pytree-eager]	0.4071ms	0.2317ms	4.3162 KOps/s	4.3574 KOps/s	$\color{#d91a1a}-0.95\%$
test_compile_add_self_flat[pytree-compile]	0.2734ms	0.1741ms	5.7424 KOps/s	5.6310 KOps/s	$\color{#35bf28}+1.98\%$
test_compile_copy_flat[tensordict-compile]	0.2501ms	0.1114ms	8.9739 KOps/s	8.8710 KOps/s	$\color{#35bf28}+1.16\%$
test_compile_copy_flat[tensordict-eager]	0.1781ms	79.1425μs	12.6354 KOps/s	12.2842 KOps/s	$\color{#35bf28}+2.86\%$
test_compile_copy_flat[pytree-compile]	0.1481ms	78.7643μs	12.6961 KOps/s	12.1962 KOps/s	$\color{#35bf28}+4.10\%$
test_compile_copy_flat[pytree-eager]	0.1542ms	69.3490μs	14.4198 KOps/s	13.8430 KOps/s	$\color{#35bf28}+4.17\%$
test_compile_assign_and_add[tensordict-compile]	0.3817ms	0.1934ms	5.1701 KOps/s	5.2188 KOps/s	$\color{#d91a1a}-0.93\%$
test_compile_assign_and_add[tensordict-eager]	2.2561ms	1.7400ms	574.7133 Ops/s	561.8206 Ops/s	$\color{#35bf28}+2.29\%$
test_compile_assign_and_add[pytree-compile]	0.2841ms	0.1868ms	5.3520 KOps/s	5.1807 KOps/s	$\color{#35bf28}+3.31\%$
test_compile_assign_and_add[pytree-eager]	1.2326ms	1.0864ms	920.4682 Ops/s	883.2286 Ops/s	$\color{#35bf28}+4.22\%$
test_compile_assign_and_add_stack[compile]	0.7310ms	0.4163ms	2.4018 KOps/s	2.3944 KOps/s	$\color{#35bf28}+0.31\%$
test_compile_assign_and_add_stack[eager]	4.1494ms	3.9825ms	251.0975 Ops/s	236.1811 Ops/s	$\textbf{\color{#35bf28}+6.32\%}$
test_compile_indexing[tensor-tensordict-compile]	80.5400μs	33.0560μs	30.2517 KOps/s	28.6803 KOps/s	$\textbf{\color{#35bf28}+5.48\%}$
test_compile_indexing[tensor-tensordict-eager]	1.1363ms	47.3647μs	21.1128 KOps/s	20.2003 KOps/s	$\color{#35bf28}+4.52\%$
test_compile_indexing[tensor-tensorclass-compile]	78.2960μs	28.7367μs	34.7987 KOps/s	32.6825 KOps/s	$\textbf{\color{#35bf28}+6.48\%}$
test_compile_indexing[tensor-tensorclass-eager]	80.0800μs	29.3675μs	34.0513 KOps/s	33.2060 KOps/s	$\color{#35bf28}+2.55\%$
test_compile_indexing[tensor-pytree-compile]	72.6260μs	28.5836μs	34.9851 KOps/s	32.7506 KOps/s	$\textbf{\color{#35bf28}+6.82\%}$
test_compile_indexing[tensor-pytree-eager]	77.5640μs	28.8515μs	34.6602 KOps/s	33.3390 KOps/s	$\color{#35bf28}+3.96\%$
test_compile_indexing[slice-tensordict-compile]	0.1501ms	72.1994μs	13.8505 KOps/s	13.4752 KOps/s	$\color{#35bf28}+2.79\%$
test_compile_indexing[slice-tensordict-eager]	0.5581ms	27.9301μs	35.8036 KOps/s	35.6697 KOps/s	$\color{#35bf28}+0.38\%$
test_compile_indexing[slice-tensorclass-compile]	0.1889ms	67.6504μs	14.7819 KOps/s	14.5571 KOps/s	$\color{#35bf28}+1.54\%$
test_compile_indexing[slice-tensorclass-eager]	65.0920μs	23.5475μs	42.4674 KOps/s	40.7201 KOps/s	$\color{#35bf28}+4.29\%$
test_compile_indexing[slice-pytree-compile]	0.1449ms	67.3576μs	14.8461 KOps/s	14.6915 KOps/s	$\color{#35bf28}+1.05\%$
test_compile_indexing[slice-pytree-eager]	65.6920μs	23.2089μs	43.0870 KOps/s	41.1881 KOps/s	$\color{#35bf28}+4.61\%$
test_compile_indexing[int-tensordict-compile]	0.1630ms	72.2276μs	13.8451 KOps/s	13.4283 KOps/s	$\color{#35bf28}+3.10\%$
test_compile_indexing[int-tensordict-eager]	0.8570ms	27.7778μs	36.0000 KOps/s	36.2061 KOps/s	$\color{#d91a1a}-0.57\%$
test_compile_indexing[int-tensorclass-compile]	0.1523ms	67.4436μs	14.8272 KOps/s	14.8126 KOps/s	$\color{#35bf28}+0.10\%$
test_compile_indexing[int-tensorclass-eager]	61.3040μs	22.9795μs	43.5171 KOps/s	41.7125 KOps/s	$\color{#35bf28}+4.33\%$
test_compile_indexing[int-pytree-compile]	0.1561ms	66.5148μs	15.0342 KOps/s	14.6826 KOps/s	$\color{#35bf28}+2.39\%$
test_compile_indexing[int-pytree-eager]	69.6700μs	23.3166μs	42.8880 KOps/s	41.2940 KOps/s	$\color{#35bf28}+3.86\%$
test_mod_add[eager]	67.6760μs	23.6471μs	42.2884 KOps/s	36.0950 KOps/s	$\textbf{\color{#35bf28}+17.16\%}$
test_mod_add[compile]	92.9140μs	37.7610μs	26.4823 KOps/s	25.6845 KOps/s	$\color{#35bf28}+3.11\%$
test_mod_add[compile-overhead]	86.8420μs	37.6462μs	26.5631 KOps/s	26.1177 KOps/s	$\color{#35bf28}+1.71\%$
test_mod_wrap[eager]	0.3145ms	0.2022ms	4.9467 KOps/s	4.8563 KOps/s	$\color{#35bf28}+1.86\%$
test_mod_wrap[compile]	0.3854ms	0.2283ms	4.3796 KOps/s	4.3407 KOps/s	$\color{#35bf28}+0.90\%$
test_mod_wrap[compile-overhead]	0.3635ms	0.2288ms	4.3715 KOps/s	4.3411 KOps/s	$\color{#35bf28}+0.70\%$
test_mod_wrap_and_backward[eager]	12.0761ms	10.6616ms	93.7946 Ops/s	92.5318 Ops/s	$\color{#35bf28}+1.36\%$
test_mod_wrap_and_backward[compile]	11.6413ms	10.7579ms	92.9546 Ops/s	83.7756 Ops/s	$\textbf{\color{#35bf28}+10.96\%}$
test_mod_wrap_and_backward[compile-overhead]	12.0332ms	10.7137ms	93.3385 Ops/s	87.0019 Ops/s	$\textbf{\color{#35bf28}+7.28\%}$
test_seq_add[eager]	0.1712ms	89.5440μs	11.1677 KOps/s	10.4428 KOps/s	$\textbf{\color{#35bf28}+6.94\%}$
test_seq_add[compile]	0.1590ms	63.3212μs	15.7925 KOps/s	15.3861 KOps/s	$\color{#35bf28}+2.64\%$
test_seq_add[compile-overhead]	0.1377ms	63.1119μs	15.8449 KOps/s	15.7873 KOps/s	$\color{#35bf28}+0.36\%$
test_seq_wrap[eager]	0.7073ms	0.3699ms	2.7031 KOps/s	2.5554 KOps/s	$\textbf{\color{#35bf28}+5.78\%}$
test_seq_wrap[compile]	0.4813ms	0.2654ms	3.7684 KOps/s	3.7220 KOps/s	$\color{#35bf28}+1.24\%$
test_seq_wrap[compile-overhead]	0.4846ms	0.2670ms	3.7448 KOps/s	3.7036 KOps/s	$\color{#35bf28}+1.11\%$
test_func_call_runtime[False-eager]	0.7349ms	0.5136ms	1.9471 KOps/s	1.9021 KOps/s	$\color{#35bf28}+2.36\%$
test_func_call_runtime[False-compile]	0.5841ms	0.4979ms	2.0083 KOps/s	1.9760 KOps/s	$\color{#35bf28}+1.64\%$
test_func_call_runtime[False-compile-overhead]	0.7267ms	0.4985ms	2.0061 KOps/s	1.9881 KOps/s	$\color{#35bf28}+0.90\%$
test_func_call_runtime[True-eager]	2.3652ms	0.7640ms	1.3089 KOps/s	1.3414 KOps/s	$\color{#d91a1a}-2.42\%$
test_func_call_runtime[True-compile]	0.9092ms	0.5194ms	1.9254 KOps/s	1.9485 KOps/s	$\color{#d91a1a}-1.18\%$
test_func_call_runtime[True-compile-overhead]	0.6640ms	0.5158ms	1.9388 KOps/s	1.9513 KOps/s	$\color{#d91a1a}-0.64\%$
test_func_call_cm_runtime[False-eager]	1.1546ms	0.5138ms	1.9461 KOps/s	1.9337 KOps/s	$\color{#35bf28}+0.65\%$
test_func_call_cm_runtime[False-compile]	0.9346ms	0.5030ms	1.9883 KOps/s	1.9708 KOps/s	$\color{#35bf28}+0.89\%$
test_func_call_cm_runtime[False-compile-overhead]	0.7341ms	0.5017ms	1.9931 KOps/s	1.9884 KOps/s	$\color{#35bf28}+0.24\%$
test_func_call_cm_runtime[True-eager]	1.4229ms	0.8832ms	1.1322 KOps/s	1.1156 KOps/s	$\color{#35bf28}+1.49\%$
test_func_call_cm_runtime[True-compile]	0.8862ms	0.7269ms	1.3757 KOps/s	1.3609 KOps/s	$\color{#35bf28}+1.09\%$
test_func_call_cm_runtime[True-compile-overhead]	0.8427ms	0.7229ms	1.3834 KOps/s	1.3515 KOps/s	$\color{#35bf28}+2.36\%$
test_vmap_func_call_cm_runtime[eager]	3.3373ms	1.8798ms	531.9826 Ops/s	530.0901 Ops/s	$\color{#35bf28}+0.36\%$
test_vmap_func_call_cm_runtime[compile]	6.0250ms	1.9845ms	503.9058 Ops/s	512.7588 Ops/s	$\color{#d91a1a}-1.73\%$
test_vmap_func_call_cm_runtime[compile-overhead]	2.6732ms	1.9213ms	520.4714 Ops/s	513.4657 Ops/s	$\color{#35bf28}+1.36\%$
test_distributed	0.2295ms	0.1252ms	7.9867 KOps/s	7.7268 KOps/s	$\color{#35bf28}+3.36\%$
test_tdmodule	82.6340μs	17.6779μs	56.5677 KOps/s	51.9173 KOps/s	$\textbf{\color{#35bf28}+8.96\%}$
test_tdmodule_dispatch	55.1730μs	35.1623μs	28.4396 KOps/s	26.1455 KOps/s	$\textbf{\color{#35bf28}+8.77\%}$
test_tdseq	37.7000μs	19.9882μs	50.0294 KOps/s	45.1731 KOps/s	$\textbf{\color{#35bf28}+10.75\%}$
test_tdseq_dispatch	67.2650μs	40.5113μs	24.6845 KOps/s	22.7598 KOps/s	$\textbf{\color{#35bf28}+8.46\%}$
test_instantiation_functorch	1.8460ms	1.5567ms	642.3942 Ops/s	634.8262 Ops/s	$\color{#35bf28}+1.19\%$
test_exec_functorch	0.3755ms	0.1856ms	5.3878 KOps/s	5.2992 KOps/s	$\color{#35bf28}+1.67\%$
test_exec_functional_call	0.7564ms	0.1722ms	5.8069 KOps/s	5.7438 KOps/s	$\color{#35bf28}+1.10\%$
test_exec_td_decorator	0.5415ms	0.2314ms	4.3218 KOps/s	4.1385 KOps/s	$\color{#35bf28}+4.43\%$
test_vmap_mlp_speed_decorator[True-True]	1.0285ms	0.6299ms	1.5875 KOps/s	1.5426 KOps/s	$\color{#35bf28}+2.91\%$
test_vmap_mlp_speed_decorator[True-False]	0.9416ms	0.6284ms	1.5913 KOps/s	1.5488 KOps/s	$\color{#35bf28}+2.75\%$
test_vmap_mlp_speed_decorator[False-True]	0.8152ms	0.5220ms	1.9156 KOps/s	1.8434 KOps/s	$\color{#35bf28}+3.92\%$
test_vmap_mlp_speed_decorator[False-False]	0.7339ms	0.5192ms	1.9260 KOps/s	1.8951 KOps/s	$\color{#35bf28}+1.63\%$
test_to_module_speed[True]	2.3430ms	1.4281ms	700.2525 Ops/s	702.4027 Ops/s	$\color{#d91a1a}-0.31\%$
test_to_module_speed[False]	1.6341ms	1.3853ms	721.8777 Ops/s	723.2419 Ops/s	$\color{#d91a1a}-0.19\%$
test_tc_init	83.7870μs	44.5585μs	22.4424 KOps/s	19.7760 KOps/s	$\textbf{\color{#35bf28}+13.48\%}$
test_tc_init_nested	0.2009ms	90.3639μs	11.0664 KOps/s	10.0457 KOps/s	$\textbf{\color{#35bf28}+10.16\%}$
test_tc_first_layer_tensor	31.8800μs	1.4356μs	696.5690 KOps/s	659.5404 KOps/s	$\textbf{\color{#35bf28}+5.61\%}$
test_tc_first_layer_nontensor	29.4550μs	4.6366μs	215.6762 KOps/s	212.0546 KOps/s	$\color{#35bf28}+1.71\%$
test_tc_second_layer_tensor	28.1930μs	2.7470μs	364.0282 KOps/s	360.8835 KOps/s	$\color{#35bf28}+0.87\%$
test_tc_second_layer_nontensor	31.6590μs	5.8864μs	169.8840 KOps/s	165.6390 KOps/s	$\color{#35bf28}+2.56\%$
test_unbind	12.5131ms	7.5823ms	131.8854 Ops/s	75.4141 Ops/s	$\textbf{\color{#35bf28}+74.88\%}$
test_full_like	18.4921ms	12.4539ms	80.2964 Ops/s	136.7223 Ops/s	$\textbf{\color{#d91a1a}-41.27\%}$
test_zeros_like	17.1823ms	7.7189ms	129.5513 Ops/s	362.6966 Ops/s	$\textbf{\color{#d91a1a}-64.28\%}$
test_ones_like	15.9530ms	7.5213ms	132.9563 Ops/s	316.4179 Ops/s	$\textbf{\color{#d91a1a}-57.98\%}$
test_clone	18.8352ms	9.3441ms	107.0195 Ops/s	198.3276 Ops/s	$\textbf{\color{#d91a1a}-46.04\%}$
test_squeeze	59.5920μs	12.1084μs	82.5876 KOps/s	77.9240 KOps/s	$\textbf{\color{#35bf28}+5.98\%}$
test_unsqueeze	0.1538ms	89.2859μs	11.2000 KOps/s	10.6004 KOps/s	$\textbf{\color{#35bf28}+5.66\%}$
test_split	0.4312ms	0.1914ms	5.2247 KOps/s	5.0981 KOps/s	$\color{#35bf28}+2.48\%$
test_permute	0.3569ms	0.2219ms	4.5059 KOps/s	4.5453 KOps/s	$\color{#d91a1a}-0.87\%$
test_stack	28.5902ms	26.1015ms	38.3120 Ops/s	36.1989 Ops/s	$\textbf{\color{#35bf28}+5.84\%}$
test_cat	28.9102ms	26.0643ms	38.3666 Ops/s	38.7258 Ops/s	$\color{#d91a1a}-0.93\%$

[github-actions]

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

Total Benchmarks: 218. Improved: $\large\color{#35bf28}47$. Worsened: $\large\color{#d91a1a}1$.

Expand to view detailed results

Name	Max	Mean	Ops	Ops on Repo HEAD	Change
test_plain_set_nested	0.1517ms	17.1475μs	58.3176 KOps/s	55.0363 KOps/s	$\textbf{\color{#35bf28}+5.96\%}$
test_plain_set_stack_nested	40.7400μs	17.1156μs	58.4261 KOps/s	54.4979 KOps/s	$\textbf{\color{#35bf28}+7.21\%}$
test_plain_set_nested_inplace	47.4210μs	18.2672μs	54.7430 KOps/s	51.9111 KOps/s	$\textbf{\color{#35bf28}+5.46\%}$
test_plain_set_stack_nested_inplace	44.8910μs	18.0773μs	55.3180 KOps/s	51.6800 KOps/s	$\textbf{\color{#35bf28}+7.04\%}$
test_items	24.0310μs	2.8303μs	353.3226 KOps/s	345.1482 KOps/s	$\color{#35bf28}+2.37\%$
test_items_nested	0.3859ms	0.3399ms	2.9423 KOps/s	2.9549 KOps/s	$\color{#d91a1a}-0.43\%$
test_items_nested_locked	0.3956ms	0.3394ms	2.9467 KOps/s	2.9420 KOps/s	$\color{#35bf28}+0.16\%$
test_items_nested_leaf	87.3710μs	62.1304μs	16.0952 KOps/s	15.9758 KOps/s	$\color{#35bf28}+0.75\%$
test_items_stack_nested	0.3922ms	0.3449ms	2.8997 KOps/s	2.9127 KOps/s	$\color{#d91a1a}-0.44\%$
test_items_stack_nested_leaf	96.9420μs	65.0871μs	15.3640 KOps/s	15.4783 KOps/s	$\color{#d91a1a}-0.74\%$
test_items_stack_nested_locked	0.4929ms	0.3451ms	2.8979 KOps/s	2.9286 KOps/s	$\color{#d91a1a}-1.05\%$
test_keys	25.8600μs	3.3984μs	294.2552 KOps/s	289.3804 KOps/s	$\color{#35bf28}+1.68\%$
test_keys_nested	0.1032ms	70.4676μs	14.1909 KOps/s	14.4181 KOps/s	$\color{#d91a1a}-1.58\%$
test_keys_nested_locked	2.3615ms	76.6026μs	13.0544 KOps/s	13.0627 KOps/s	$\color{#d91a1a}-0.06\%$
test_keys_nested_leaf	0.1043ms	61.3138μs	16.3096 KOps/s	16.1535 KOps/s	$\color{#35bf28}+0.97\%$
test_keys_stack_nested	0.1134ms	71.6165μs	13.9633 KOps/s	13.7908 KOps/s	$\color{#35bf28}+1.25\%$
test_keys_stack_nested_leaf	94.1210μs	63.6168μs	15.7191 KOps/s	15.8691 KOps/s	$\color{#d91a1a}-0.95\%$
test_keys_stack_nested_locked	0.1061ms	76.6588μs	13.0448 KOps/s	12.8398 KOps/s	$\color{#35bf28}+1.60\%$
test_values	5.6152μs	0.8609μs	1.1615 MOps/s	1.1419 MOps/s	$\color{#35bf28}+1.72\%$
test_values_nested	84.6620μs	48.7275μs	20.5223 KOps/s	20.5075 KOps/s	$\color{#35bf28}+0.07\%$
test_values_nested_locked	84.0020μs	49.9683μs	20.0127 KOps/s	19.9963 KOps/s	$\color{#35bf28}+0.08\%$
test_values_nested_leaf	67.8910μs	42.4974μs	23.5308 KOps/s	23.5198 KOps/s	$\color{#35bf28}+0.05\%$
test_values_stack_nested	92.0120μs	50.4062μs	19.8388 KOps/s	19.8073 KOps/s	$\color{#35bf28}+0.16\%$
test_values_stack_nested_leaf	75.5120μs	43.8721μs	22.7936 KOps/s	22.6345 KOps/s	$\color{#35bf28}+0.70\%$
test_values_stack_nested_locked	0.1021ms	51.3167μs	19.4868 KOps/s	19.4054 KOps/s	$\color{#35bf28}+0.42\%$
test_membership	2.3016μs	0.5090μs	1.9646 MOps/s	1.9964 MOps/s	$\color{#d91a1a}-1.59\%$
test_membership_nested	15.4100μs	1.8727μs	533.9772 KOps/s	524.8455 KOps/s	$\color{#35bf28}+1.74\%$
test_membership_nested_leaf	10.1833μs	1.8494μs	540.7154 KOps/s	541.0576 KOps/s	$\color{#d91a1a}-0.06\%$
test_membership_stacked_nested	32.4710μs	1.9414μs	515.0924 KOps/s	514.5154 KOps/s	$\color{#35bf28}+0.11\%$
test_membership_stacked_nested_leaf	24.7910μs	1.9563μs	511.1659 KOps/s	511.4671 KOps/s	$\color{#d91a1a}-0.06\%$
test_membership_nested_last	31.1900μs	2.9817μs	335.3756 KOps/s	334.2882 KOps/s	$\color{#35bf28}+0.33\%$
test_membership_nested_leaf_last	33.9010μs	2.9618μs	337.6335 KOps/s	337.6237 KOps/s	$+0.00\%$
test_membership_stacked_nested_last	39.6610μs	2.9999μs	333.3495 KOps/s	287.3409 KOps/s	$\textbf{\color{#35bf28}+16.01\%}$
test_membership_stacked_nested_leaf_last	26.3910μs	2.9499μs	338.9966 KOps/s	287.5203 KOps/s	$\textbf{\color{#35bf28}+17.90\%}$
test_nested_getleaf	34.3210μs	6.0716μs	164.7015 KOps/s	162.4108 KOps/s	$\color{#35bf28}+1.41\%$
test_nested_get	35.2710μs	5.7023μs	175.3685 KOps/s	173.4923 KOps/s	$\color{#35bf28}+1.08\%$
test_stacked_getleaf	29.7400μs	6.0218μs	166.0647 KOps/s	163.7060 KOps/s	$\color{#35bf28}+1.44\%$
test_stacked_get	37.9000μs	5.6100μs	178.2530 KOps/s	175.2396 KOps/s	$\color{#35bf28}+1.72\%$
test_nested_getitemleaf	32.6900μs	6.0595μs	165.0313 KOps/s	163.5802 KOps/s	$\color{#35bf28}+0.89\%$
test_nested_getitem	36.5910μs	5.7529μs	173.8242 KOps/s	175.5089 KOps/s	$\color{#d91a1a}-0.96\%$
test_stacked_getitemleaf	37.2100μs	6.0221μs	166.0546 KOps/s	162.4451 KOps/s	$\color{#35bf28}+2.22\%$
test_stacked_getitem	32.5310μs	5.6681μs	176.4255 KOps/s	172.7817 KOps/s	$\color{#35bf28}+2.11\%$
test_lock_nested	5.1619ms	0.4273ms	2.3403 KOps/s	2.3246 KOps/s	$\color{#35bf28}+0.67\%$
test_lock_stack_nested	0.4405ms	0.3848ms	2.5987 KOps/s	2.5551 KOps/s	$\color{#35bf28}+1.70\%$
test_unlock_nested	0.7637ms	0.3601ms	2.7774 KOps/s	2.7592 KOps/s	$\color{#35bf28}+0.66\%$
test_unlock_stack_nested	0.3761ms	0.3241ms	3.0852 KOps/s	3.0567 KOps/s	$\color{#35bf28}+0.93\%$
test_flatten_speed	0.1561ms	76.1800μs	13.1268 KOps/s	13.0477 KOps/s	$\color{#35bf28}+0.61\%$
test_unflatten_speed	0.3566ms	0.3179ms	3.1454 KOps/s	3.0914 KOps/s	$\color{#35bf28}+1.75\%$
test_common_ops	1.5553ms	1.2444ms	803.6185 Ops/s	755.4151 Ops/s	$\textbf{\color{#35bf28}+6.38\%}$
test_creation	23.4100μs	1.4847μs	673.5264 KOps/s	671.1538 KOps/s	$\color{#35bf28}+0.35\%$
test_creation_empty	41.1010μs	15.9574μs	62.6669 KOps/s	53.8095 KOps/s	$\textbf{\color{#35bf28}+16.46\%}$
test_creation_nested_1	65.7110μs	17.5298μs	57.0456 KOps/s	49.7706 KOps/s	$\textbf{\color{#35bf28}+14.62\%}$
test_creation_nested_2	57.1620μs	20.0053μs	49.9868 KOps/s	43.4168 KOps/s	$\textbf{\color{#35bf28}+15.13\%}$
test_clone	65.6610μs	28.5231μs	35.0592 KOps/s	33.3449 KOps/s	$\textbf{\color{#35bf28}+5.14\%}$
test_getitem[int]	1.2499ms	15.7499μs	63.4924 KOps/s	61.5917 KOps/s	$\color{#35bf28}+3.09\%$
test_getitem[slice_int]	0.1240ms	27.2373μs	36.7144 KOps/s	36.4322 KOps/s	$\color{#35bf28}+0.77\%$
test_getitem[range]	0.2870ms	0.1088ms	9.1928 KOps/s	8.9759 KOps/s	$\color{#35bf28}+2.42\%$
test_getitem[tuple]	0.1158ms	23.4333μs	42.6743 KOps/s	41.9694 KOps/s	$\color{#35bf28}+1.68\%$
test_getitem[list]	0.1859ms	98.3322μs	10.1696 KOps/s	10.0307 KOps/s	$\color{#35bf28}+1.38\%$
test_setitem_dim[int]	67.9210μs	45.0706μs	22.1874 KOps/s	22.2760 KOps/s	$\color{#d91a1a}-0.40\%$
test_setitem_dim[slice_int]	0.1025ms	66.9515μs	14.9362 KOps/s	15.0777 KOps/s	$\color{#d91a1a}-0.94\%$
test_setitem_dim[range]	0.1638ms	0.1278ms	7.8258 KOps/s	7.8374 KOps/s	$\color{#d91a1a}-0.15\%$
test_setitem_dim[tuple]	86.6610μs	60.3174μs	16.5790 KOps/s	16.4583 KOps/s	$\color{#35bf28}+0.73\%$
test_setitem	93.4210μs	41.6709μs	23.9976 KOps/s	22.3965 KOps/s	$\textbf{\color{#35bf28}+7.15\%}$
test_set	69.5310μs	39.9387μs	25.0384 KOps/s	23.1146 KOps/s	$\textbf{\color{#35bf28}+8.32\%}$
test_set_shared	0.3546ms	52.6819μs	18.9819 KOps/s	18.4579 KOps/s	$\color{#35bf28}+2.84\%$
test_update	97.0120μs	49.8032μs	20.0790 KOps/s	18.5230 KOps/s	$\textbf{\color{#35bf28}+8.40\%}$
test_update_nested	98.2220μs	57.2448μs	17.4688 KOps/s	16.4095 KOps/s	$\textbf{\color{#35bf28}+6.46\%}$
test_update__nested	0.1739ms	61.4051μs	16.2853 KOps/s	16.1746 KOps/s	$\color{#35bf28}+0.68\%$
test_set_nested	82.2920μs	43.5732μs	22.9499 KOps/s	21.6484 KOps/s	$\textbf{\color{#35bf28}+6.01\%}$
test_set_nested_new	85.9410μs	47.4392μs	21.0796 KOps/s	20.4008 KOps/s	$\color{#35bf28}+3.33\%$
test_select	93.8010μs	60.3386μs	16.5731 KOps/s	15.9872 KOps/s	$\color{#35bf28}+3.67\%$
test_select_nested	0.4263ms	42.2588μs	23.6637 KOps/s	23.6528 KOps/s	$\color{#35bf28}+0.05\%$
test_exclude_nested	96.1120μs	59.0294μs	16.9407 KOps/s	16.0882 KOps/s	$\textbf{\color{#35bf28}+5.30\%}$
test_empty[True]	0.3096ms	0.2575ms	3.8829 KOps/s	3.8296 KOps/s	$\color{#35bf28}+1.39\%$
test_empty[False]	3.2620μs	0.7376μs	1.3558 MOps/s	1.3384 MOps/s	$\color{#35bf28}+1.30\%$
test_to	61.4410μs	27.1775μs	36.7952 KOps/s	36.3677 KOps/s	$\color{#35bf28}+1.18\%$
test_to_nonblocking	54.9210μs	25.6531μs	38.9817 KOps/s	38.2869 KOps/s	$\color{#35bf28}+1.81\%$
test_unbind_speed	0.3361ms	0.2801ms	3.5708 KOps/s	3.5519 KOps/s	$\color{#35bf28}+0.53\%$
test_unbind_speed_stack0	0.3249ms	0.2780ms	3.5968 KOps/s	3.5610 KOps/s	$\color{#35bf28}+1.01\%$
test_unbind_speed_stack1	92.7001ms	0.7050ms	1.4183 KOps/s	1.4153 KOps/s	$\color{#35bf28}+0.22\%$
test_split	95.2307ms	2.2037ms	453.7753 Ops/s	454.0773 Ops/s	$\color{#d91a1a}-0.07\%$
test_chunk	95.4556ms	2.2161ms	451.2427 Ops/s	452.6570 Ops/s	$\color{#d91a1a}-0.31\%$
test_creation[device0]	0.3428ms	0.1286ms	7.7779 KOps/s	7.7598 KOps/s	$\color{#35bf28}+0.23\%$
test_creation_from_tensor	0.3510ms	0.1296ms	7.7180 KOps/s	7.6752 KOps/s	$\color{#35bf28}+0.56\%$
test_add_one[memmap_tensor0]	0.2333ms	8.7528μs	114.2497 KOps/s	112.7858 KOps/s	$\color{#35bf28}+1.30\%$
test_contiguous[memmap_tensor0]	21.5710μs	2.2284μs	448.7479 KOps/s	442.3126 KOps/s	$\color{#35bf28}+1.45\%$
test_stack[memmap_tensor0]	37.6810μs	7.1204μs	140.4410 KOps/s	141.1180 KOps/s	$\color{#d91a1a}-0.48\%$
test_memmaptd_index	1.2196ms	0.4431ms	2.2569 KOps/s	2.2662 KOps/s	$\color{#d91a1a}-0.41\%$
test_memmaptd_index_astensor	0.7785ms	0.5137ms	1.9467 KOps/s	1.9276 KOps/s	$\color{#35bf28}+0.99\%$
test_memmaptd_index_op	1.4350ms	1.0412ms	960.4328 Ops/s	929.9012 Ops/s	$\color{#35bf28}+3.28\%$
test_serialize_model	0.1304s	0.1298s	7.7067 Ops/s	7.7258 Ops/s	$\color{#d91a1a}-0.25\%$
test_serialize_model_pickle	1.3477s	1.2121s	0.8250 Ops/s	0.8221 Ops/s	$\color{#35bf28}+0.35\%$
test_serialize_weights	0.1306s	0.1295s	7.7235 Ops/s	7.7620 Ops/s	$\color{#d91a1a}-0.50\%$
test_serialize_weights_returnearly	0.2405s	63.1841ms	15.8268 Ops/s	15.7426 Ops/s	$\color{#35bf28}+0.53\%$
test_serialize_weights_pickle	1.3472s	1.1853s	0.8437 Ops/s	0.8197 Ops/s	$\color{#35bf28}+2.92\%$
test_reshape_pytree	62.2210μs	36.6218μs	27.3061 KOps/s	27.0492 KOps/s	$\color{#35bf28}+0.95\%$
test_reshape_td	95.2420μs	43.1747μs	23.1617 KOps/s	22.5901 KOps/s	$\color{#35bf28}+2.53\%$
test_view_pytree	67.3510μs	35.6397μs	28.0586 KOps/s	27.3141 KOps/s	$\color{#35bf28}+2.73\%$
test_view_td	86.1810μs	46.7242μs	21.4022 KOps/s	20.8592 KOps/s	$\color{#35bf28}+2.60\%$
test_unbind_pytree	68.9010μs	34.7130μs	28.8077 KOps/s	29.0452 KOps/s	$\color{#d91a1a}-0.82\%$
test_unbind_td	0.5209ms	43.6801μs	22.8937 KOps/s	23.0665 KOps/s	$\color{#d91a1a}-0.75\%$
test_split_pytree	85.3420μs	46.0940μs	21.6948 KOps/s	21.1232 KOps/s	$\color{#35bf28}+2.71\%$
test_split_td	0.6926ms	55.2295μs	18.1062 KOps/s	17.5159 KOps/s	$\color{#35bf28}+3.37\%$
test_add_pytree	0.1080ms	56.7415μs	17.6238 KOps/s	16.7382 KOps/s	$\textbf{\color{#35bf28}+5.29\%}$
test_add_td	0.2288ms	97.1708μs	10.2912 KOps/s	10.1511 KOps/s	$\color{#35bf28}+1.38\%$
test_compile_add_one_nested[tensordict-compile]	0.2574ms	0.1612ms	6.2020 KOps/s	6.0281 KOps/s	$\color{#35bf28}+2.88\%$
test_compile_add_one_nested[tensordict-eager]	0.5605ms	0.1604ms	6.2341 KOps/s	6.3055 KOps/s	$\color{#d91a1a}-1.13\%$
test_compile_add_one_nested[pytree-compile]	0.2152ms	0.1534ms	6.5176 KOps/s	6.1371 KOps/s	$\textbf{\color{#35bf28}+6.20\%}$
test_compile_add_one_nested[pytree-eager]	0.5797ms	0.1865ms	5.3620 KOps/s	4.9255 KOps/s	$\textbf{\color{#35bf28}+8.86\%}$
test_compile_copy_nested[tensordict-compile]	0.4046ms	21.7103μs	46.0611 KOps/s	47.5989 KOps/s	$\color{#d91a1a}-3.23\%$
test_compile_copy_nested[tensordict-eager]	0.1276ms	47.7419μs	20.9459 KOps/s	20.2171 KOps/s	$\color{#35bf28}+3.61\%$
test_compile_copy_nested[pytree-compile]	0.4531ms	64.4790μs	15.5089 KOps/s	16.0400 KOps/s	$\color{#d91a1a}-3.31\%$
test_compile_copy_nested[pytree-eager]	0.4355ms	49.9122μs	20.0352 KOps/s	20.0599 KOps/s	$\color{#d91a1a}-0.12\%$
test_compile_add_one_flat[tensordict-compile]	0.4106ms	0.3172ms	3.1525 KOps/s	3.0922 KOps/s	$\color{#35bf28}+1.95\%$
test_compile_add_one_flat[tensordict-eager]	0.6309ms	0.2296ms	4.3558 KOps/s	4.3071 KOps/s	$\color{#35bf28}+1.13\%$
test_compile_add_one_flat[tensorclass-compile]	0.1774ms	0.1269ms	7.8791 KOps/s	7.7935 KOps/s	$\color{#35bf28}+1.10\%$
test_compile_add_one_flat[tensorclass-eager]	0.4749ms	64.8658μs	15.4164 KOps/s	15.3428 KOps/s	$\color{#35bf28}+0.48\%$
test_compile_add_one_flat[pytree-compile]	0.7312ms	0.3258ms	3.0690 KOps/s	3.0265 KOps/s	$\color{#35bf28}+1.40\%$
test_compile_add_one_flat[pytree-eager]	1.0231ms	0.6364ms	1.5713 KOps/s	1.4663 KOps/s	$\textbf{\color{#35bf28}+7.16\%}$
test_compile_add_self_flat[tensordict-eager]	0.6787ms	0.2806ms	3.5636 KOps/s	3.5245 KOps/s	$\color{#35bf28}+1.11\%$
test_compile_add_self_flat[tensordict-compile]	0.3602ms	0.3188ms	3.1372 KOps/s	3.0786 KOps/s	$\color{#35bf28}+1.90\%$
test_compile_add_self_flat[tensorclass-eager]	0.1612ms	76.5751μs	13.0591 KOps/s	12.9942 KOps/s	$\color{#35bf28}+0.50\%$
test_compile_add_self_flat[tensorclass-compile]	0.1931ms	0.1291ms	7.7446 KOps/s	7.7718 KOps/s	$\color{#d91a1a}-0.35\%$
test_compile_add_self_flat[pytree-eager]	0.6483ms	0.5344ms	1.8712 KOps/s	1.6820 KOps/s	$\textbf{\color{#35bf28}+11.24\%}$
test_compile_add_self_flat[pytree-compile]	0.3933ms	0.3258ms	3.0694 KOps/s	3.0107 KOps/s	$\color{#35bf28}+1.95\%$
test_compile_copy_flat[tensordict-compile]	86.9810μs	20.5835μs	48.5826 KOps/s	48.2196 KOps/s	$\color{#35bf28}+0.75\%$
test_compile_copy_flat[tensordict-eager]	94.6920μs	37.4553μs	26.6985 KOps/s	25.6605 KOps/s	$\color{#35bf28}+4.05\%$
test_compile_copy_flat[pytree-compile]	0.1541ms	68.5090μs	14.5966 KOps/s	14.4313 KOps/s	$\color{#35bf28}+1.15\%$
test_compile_copy_flat[pytree-eager]	0.4313ms	51.2940μs	19.4955 KOps/s	19.5822 KOps/s	$\color{#d91a1a}-0.44\%$
test_compile_assign_and_add[tensordict-compile]	2.4331ms	0.8431ms	1.1861 KOps/s	1.0845 KOps/s	$\textbf{\color{#35bf28}+9.37\%}$
test_compile_assign_and_add[tensordict-eager]	3.6000ms	3.2498ms	307.7135 Ops/s	294.0750 Ops/s	$\color{#35bf28}+4.64\%$
test_compile_assign_and_add[pytree-compile]	2.4478ms	0.8494ms	1.1773 KOps/s	1.0736 KOps/s	$\textbf{\color{#35bf28}+9.66\%}$
test_compile_assign_and_add[pytree-eager]	3.3706ms	3.2046ms	312.0560 Ops/s	283.9896 Ops/s	$\textbf{\color{#35bf28}+9.88\%}$
test_compile_indexing[tensor-tensordict-compile]	0.1566ms	0.1176ms	8.5070 KOps/s	8.0505 KOps/s	$\textbf{\color{#35bf28}+5.67\%}$
test_compile_indexing[tensor-tensordict-eager]	0.1826ms	62.5291μs	15.9926 KOps/s	15.7138 KOps/s	$\color{#35bf28}+1.77\%$
test_compile_indexing[tensor-tensorclass-compile]	0.1642ms	0.1130ms	8.8475 KOps/s	8.4767 KOps/s	$\color{#35bf28}+4.37\%$
test_compile_indexing[tensor-tensorclass-eager]	0.1056ms	44.3390μs	22.5535 KOps/s	21.2692 KOps/s	$\textbf{\color{#35bf28}+6.04\%}$
test_compile_indexing[tensor-pytree-compile]	0.1758ms	0.1196ms	8.3643 KOps/s	8.2274 KOps/s	$\color{#35bf28}+1.66\%$
test_compile_indexing[tensor-pytree-eager]	85.8920μs	43.4603μs	23.0095 KOps/s	21.1474 KOps/s	$\textbf{\color{#35bf28}+8.81\%}$
test_compile_indexing[slice-tensordict-compile]	0.1874ms	0.1447ms	6.9090 KOps/s	6.6865 KOps/s	$\color{#35bf28}+3.33\%$
test_compile_indexing[slice-tensordict-eager]	0.1527ms	25.5059μs	39.2066 KOps/s	37.6781 KOps/s	$\color{#35bf28}+4.06\%$
test_compile_indexing[slice-tensorclass-compile]	0.1818ms	0.1443ms	6.9302 KOps/s	6.9680 KOps/s	$\color{#d91a1a}-0.54\%$
test_compile_indexing[slice-tensorclass-eager]	62.6710μs	21.6055μs	46.2844 KOps/s	48.2264 KOps/s	$\color{#d91a1a}-4.03\%$
test_compile_indexing[slice-pytree-compile]	0.1913ms	0.1414ms	7.0735 KOps/s	6.8211 KOps/s	$\color{#35bf28}+3.70\%$
test_compile_indexing[slice-pytree-eager]	55.5510μs	21.0783μs	47.4422 KOps/s	48.7111 KOps/s	$\color{#d91a1a}-2.61\%$
test_compile_indexing[int-tensordict-compile]	0.2740ms	0.1451ms	6.8931 KOps/s	6.5633 KOps/s	$\textbf{\color{#35bf28}+5.03\%}$
test_compile_indexing[int-tensordict-eager]	0.4872ms	24.3249μs	41.1101 KOps/s	38.1537 KOps/s	$\textbf{\color{#35bf28}+7.75\%}$
test_compile_indexing[int-tensorclass-compile]	0.2046ms	0.1394ms	7.1718 KOps/s	6.7676 KOps/s	$\textbf{\color{#35bf28}+5.97\%}$
test_compile_indexing[int-tensorclass-eager]	58.4210μs	20.6386μs	48.4530 KOps/s	48.0781 KOps/s	$\color{#35bf28}+0.78\%$
test_compile_indexing[int-pytree-compile]	0.1818ms	0.1394ms	7.1734 KOps/s	6.7964 KOps/s	$\textbf{\color{#35bf28}+5.55\%}$
test_compile_indexing[int-pytree-eager]	52.7310μs	21.2432μs	47.0740 KOps/s	48.4152 KOps/s	$\color{#d91a1a}-2.77\%$
test_mod_add[eager]	65.6420μs	32.3338μs	30.9274 KOps/s	27.6327 KOps/s	$\textbf{\color{#35bf28}+11.92\%}$
test_mod_add[compile]	0.1205ms	80.6134μs	12.4049 KOps/s	11.4541 KOps/s	$\textbf{\color{#35bf28}+8.30\%}$
test_mod_add[compile-overhead]	0.3052ms	0.1525ms	6.5557 KOps/s	6.1021 KOps/s	$\textbf{\color{#35bf28}+7.43\%}$
test_mod_wrap[eager]	0.3452ms	0.2393ms	4.1785 KOps/s	3.8287 KOps/s	$\textbf{\color{#35bf28}+9.13\%}$
test_mod_wrap[compile]	1.5572ms	0.3069ms	3.2581 KOps/s	3.1779 KOps/s	$\color{#35bf28}+2.52\%$
test_mod_wrap[compile-overhead]	7.6448ms	4.0528ms	246.7436 Ops/s	246.2503 Ops/s	$\color{#35bf28}+0.20\%$
test_mod_wrap_and_backward[eager]	1.4795ms	1.3117ms	762.3685 Ops/s	696.9353 Ops/s	$\textbf{\color{#35bf28}+9.39\%}$
test_mod_wrap_and_backward[compile]	1.5539ms	1.3182ms	758.6131 Ops/s	690.9030 Ops/s	$\textbf{\color{#35bf28}+9.80\%}$
test_mod_wrap_and_backward[compile-overhead]	1.3355ms	0.8937ms	1.1190 KOps/s	977.1837 Ops/s	$\textbf{\color{#35bf28}+14.51\%}$
test_seq_add[eager]	0.1633ms	0.1043ms	9.5921 KOps/s	9.4509 KOps/s	$\color{#35bf28}+1.49\%$
test_seq_add[compile]	0.3339ms	95.9950μs	10.4172 KOps/s	10.9596 KOps/s	$\color{#d91a1a}-4.95\%$
test_seq_add[compile-overhead]	0.1764ms	0.1243ms	8.0471 KOps/s	8.0774 KOps/s	$\color{#d91a1a}-0.38\%$
test_seq_wrap[eager]	0.4952ms	0.3935ms	2.5415 KOps/s	2.4247 KOps/s	$\color{#35bf28}+4.82\%$
test_seq_wrap[compile]	0.3492ms	0.3128ms	3.1969 KOps/s	3.1037 KOps/s	$\color{#35bf28}+3.00\%$
test_seq_wrap[compile-overhead]	0.2694ms	0.2188ms	4.5697 KOps/s	4.4759 KOps/s	$\color{#35bf28}+2.10\%$
test_func_call_runtime[False-eager]	0.8018ms	0.7218ms	1.3854 KOps/s	1.2778 KOps/s	$\textbf{\color{#35bf28}+8.43\%}$
test_func_call_runtime[False-compile]	0.8300ms	0.7794ms	1.2831 KOps/s	1.2383 KOps/s	$\color{#35bf28}+3.62\%$
test_func_call_runtime[False-compile-overhead]	0.4249ms	0.3589ms	2.7862 KOps/s	2.7545 KOps/s	$\color{#35bf28}+1.15\%$
test_func_call_runtime[True-eager]	0.9579ms	0.8762ms	1.1413 KOps/s	1.1093 KOps/s	$\color{#35bf28}+2.88\%$
test_func_call_runtime[True-compile]	0.9467ms	0.8156ms	1.2260 KOps/s	1.2150 KOps/s	$\color{#35bf28}+0.91\%$
test_func_call_runtime[True-compile-overhead]	0.4316ms	0.3806ms	2.6276 KOps/s	2.5965 KOps/s	$\color{#35bf28}+1.20\%$
test_func_call_cm_runtime[False-eager]	0.7677ms	0.7095ms	1.4094 KOps/s	1.2975 KOps/s	$\textbf{\color{#35bf28}+8.63\%}$
test_func_call_cm_runtime[False-compile]	0.8649ms	0.7964ms	1.2557 KOps/s	1.2426 KOps/s	$\color{#35bf28}+1.05\%$
test_func_call_cm_runtime[False-compile-overhead]	0.4159ms	0.3593ms	2.7832 KOps/s	2.7377 KOps/s	$\color{#35bf28}+1.66\%$
test_func_call_cm_runtime[True-eager]	1.1181ms	0.9906ms	1.0095 KOps/s	999.6554 Ops/s	$\color{#35bf28}+0.99\%$
test_func_call_cm_runtime[True-compile]	0.8999ms	0.8412ms	1.1888 KOps/s	1.1489 KOps/s	$\color{#35bf28}+3.48\%$
test_func_call_cm_runtime[True-compile-overhead]	0.4484ms	0.4025ms	2.4848 KOps/s	2.4169 KOps/s	$\color{#35bf28}+2.81\%$
test_vmap_func_call_cm_runtime[eager]	2.5408ms	2.0429ms	489.5014 Ops/s	482.6524 Ops/s	$\color{#35bf28}+1.42\%$
test_vmap_func_call_cm_runtime[compile]	0.9299ms	0.8566ms	1.1674 KOps/s	1.1409 KOps/s	$\color{#35bf28}+2.32\%$
test_vmap_func_call_cm_runtime[compile-overhead]	0.4674ms	0.4065ms	2.4601 KOps/s	2.4160 KOps/s	$\color{#35bf28}+1.82\%$
test_distributed	5.4632ms	0.2290ms	4.3671 KOps/s	8.5867 KOps/s	$\textbf{\color{#d91a1a}-49.14\%}$
test_tdmodule	48.6110μs	15.0568μs	66.4152 KOps/s	58.6941 KOps/s	$\textbf{\color{#35bf28}+13.15\%}$
test_tdmodule_dispatch	42.0010μs	29.1315μs	34.3271 KOps/s	31.0034 KOps/s	$\textbf{\color{#35bf28}+10.72\%}$
test_tdseq	34.3210μs	16.0410μs	62.3404 KOps/s	57.3394 KOps/s	$\textbf{\color{#35bf28}+8.72\%}$
test_tdseq_dispatch	52.3710μs	32.0821μs	31.1701 KOps/s	28.3368 KOps/s	$\textbf{\color{#35bf28}+10.00\%}$
test_instantiation_functorch	2.0309ms	1.8671ms	535.5892 Ops/s	528.4144 Ops/s	$\color{#35bf28}+1.36\%$
test_exec_functorch	0.2556ms	0.2074ms	4.8217 KOps/s	4.7783 KOps/s	$\color{#35bf28}+0.91\%$
test_exec_functional_call	0.3090ms	0.2010ms	4.9764 KOps/s	4.8287 KOps/s	$\color{#35bf28}+3.06\%$
test_exec_td_decorator	0.4416ms	0.2608ms	3.8337 KOps/s	3.8678 KOps/s	$\color{#d91a1a}-0.88\%$
test_vmap_mlp_speed_decorator[True-True]	0.7976ms	0.6663ms	1.5008 KOps/s	1.4660 KOps/s	$\color{#35bf28}+2.38\%$
test_vmap_mlp_speed_decorator[True-False]	0.7960ms	0.6673ms	1.4987 KOps/s	1.4748 KOps/s	$\color{#35bf28}+1.62\%$
test_vmap_mlp_speed_decorator[False-True]	0.6858ms	0.5834ms	1.7140 KOps/s	1.6910 KOps/s	$\color{#35bf28}+1.36\%$
test_vmap_mlp_speed_decorator[False-False]	0.6866ms	0.5810ms	1.7211 KOps/s	1.6921 KOps/s	$\color{#35bf28}+1.71\%$
test_vmap_transformer_speed_decorator[True-True]	19.7858ms	19.1357ms	52.2583 Ops/s	51.6777 Ops/s	$\color{#35bf28}+1.12\%$
test_vmap_transformer_speed_decorator[True-False]	19.2965ms	19.1114ms	52.3249 Ops/s	51.6290 Ops/s	$\color{#35bf28}+1.35\%$
test_vmap_transformer_speed_decorator[False-True]	19.9821ms	19.0215ms	52.5720 Ops/s	52.1584 Ops/s	$\color{#35bf28}+0.79\%$
test_vmap_transformer_speed_decorator[False-False]	19.0894ms	18.9667ms	52.7240 Ops/s	52.0173 Ops/s	$\color{#35bf28}+1.36\%$
test_to_module_speed[True]	1.5217ms	0.9990ms	1.0010 KOps/s	988.5761 Ops/s	$\color{#35bf28}+1.26\%$
test_to_module_speed[False]	1.3945ms	0.9792ms	1.0212 KOps/s	1.0280 KOps/s	$\color{#d91a1a}-0.66\%$
test_tc_init	61.7110μs	33.5859μs	29.7744 KOps/s	27.3196 KOps/s	$\textbf{\color{#35bf28}+8.99\%}$
test_tc_init_nested	0.1068ms	70.7227μs	14.1397 KOps/s	13.5939 KOps/s	$\color{#35bf28}+4.02\%$
test_tc_first_layer_tensor	4.8043μs	0.6847μs	1.4606 MOps/s	1.4682 MOps/s	$\color{#d91a1a}-0.52\%$
test_tc_first_layer_nontensor	40.6310μs	2.2506μs	444.3221 KOps/s	445.4248 KOps/s	$\color{#d91a1a}-0.25\%$
test_tc_second_layer_tensor	10.5133μs	1.3935μs	717.6019 KOps/s	735.2682 KOps/s	$\color{#d91a1a}-2.40\%$
test_tc_second_layer_nontensor	24.0710μs	2.9746μs	336.1808 KOps/s	338.8458 KOps/s	$\color{#d91a1a}-0.79\%$
test_unbind	0.1909s	9.4946ms	105.3233 Ops/s	91.6300 Ops/s	$\textbf{\color{#35bf28}+14.94\%}$
test_full_like	0.6554ms	0.5756ms	1.7373 KOps/s	1.7538 KOps/s	$\color{#d91a1a}-0.94\%$
test_zeros_like	0.2753ms	0.1979ms	5.0531 KOps/s	5.0498 KOps/s	$\color{#35bf28}+0.07\%$
test_ones_like	0.2323ms	0.1978ms	5.0567 KOps/s	5.0555 KOps/s	$\color{#35bf28}+0.02\%$
test_clone	0.4552ms	0.4146ms	2.4120 KOps/s	2.4104 KOps/s	$\color{#35bf28}+0.07\%$
test_squeeze	36.1510μs	9.6750μs	103.3588 KOps/s	100.4966 KOps/s	$\color{#35bf28}+2.85\%$
test_unsqueeze	0.2460ms	75.2998μs	13.2802 KOps/s	12.9466 KOps/s	$\color{#35bf28}+2.58\%$
test_split	0.4227ms	0.1555ms	6.4327 KOps/s	6.3061 KOps/s	$\color{#35bf28}+2.01\%$
test_permute	0.2292ms	0.1787ms	5.5958 KOps/s	5.4530 KOps/s	$\color{#35bf28}+2.62\%$
test_stack	1.2574ms	0.8571ms	1.1668 KOps/s	1.1621 KOps/s	$\color{#35bf28}+0.41\%$
test_cat	1.2619ms	1.2311ms	812.2899 Ops/s	812.0360 Ops/s	$\color{#35bf28}+0.03\%$