[Refactor] use from_file instead of mmap+from_buffer for readonly files

[vmoens]

However, this fails on our tests
For a minimal reprod https://gist.github.com/vmoens/83cdb5b9059e319829607c22927f0383

cc @teopir @albanD @mikaylagawarecki

[github-actions]

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

Total Benchmarks: 144. Improved: $\large\color{#35bf28}33$. Worsened: $\large\color{#d91a1a}7$.

Expand to view detailed results

Name	Max	Mean	Ops	Ops on Repo HEAD	Change
test_plain_set_nested	0.2727ms	18.5146μs	54.0116 KOps/s	57.9460 KOps/s	$\textbf{\color{#d91a1a}-6.79\%}$
test_plain_set_stack_nested	0.2785ms	17.5949μs	56.8346 KOps/s	57.1607 KOps/s	$\color{#d91a1a}-0.57\%$
test_plain_set_nested_inplace	52.9380μs	18.2429μs	54.8157 KOps/s	50.4429 KOps/s	$\textbf{\color{#35bf28}+8.67\%}$
test_plain_set_stack_nested_inplace	45.6950μs	18.3260μs	54.5672 KOps/s	50.5923 KOps/s	$\textbf{\color{#35bf28}+7.86\%}$
test_items	16.1500μs	2.6635μs	375.4445 KOps/s	384.7576 KOps/s	$\color{#d91a1a}-2.42\%$
test_items_nested	0.4016ms	0.2655ms	3.7669 KOps/s	3.7647 KOps/s	$\color{#35bf28}+0.06\%$
test_items_nested_locked	1.1236ms	0.2680ms	3.7313 KOps/s	3.7364 KOps/s	$\color{#d91a1a}-0.14\%$
test_items_nested_leaf	0.1480ms	76.1620μs	13.1299 KOps/s	12.8894 KOps/s	$\color{#35bf28}+1.87\%$
test_items_stack_nested	0.3405ms	0.2699ms	3.7051 KOps/s	3.6961 KOps/s	$\color{#35bf28}+0.24\%$
test_items_stack_nested_leaf	0.1917ms	79.0886μs	12.6441 KOps/s	11.9649 KOps/s	$\textbf{\color{#35bf28}+5.68\%}$
test_items_stack_nested_locked	1.0170ms	0.2692ms	3.7154 KOps/s	3.7118 KOps/s	$\color{#35bf28}+0.09\%$
test_keys	24.4350μs	3.8029μs	262.9571 KOps/s	255.2959 KOps/s	$\color{#35bf28}+3.00\%$
test_keys_nested	0.2529ms	0.1362ms	7.3441 KOps/s	7.1271 KOps/s	$\color{#35bf28}+3.05\%$
test_keys_nested_locked	0.7401ms	0.1415ms	7.0681 KOps/s	6.9209 KOps/s	$\color{#35bf28}+2.13\%$
test_keys_nested_leaf	0.2185ms	0.1166ms	8.5769 KOps/s	8.3734 KOps/s	$\color{#35bf28}+2.43\%$
test_keys_stack_nested	0.2869ms	0.1369ms	7.3050 KOps/s	7.1770 KOps/s	$\color{#35bf28}+1.78\%$
test_keys_stack_nested_leaf	0.2070ms	0.1149ms	8.7005 KOps/s	8.4141 KOps/s	$\color{#35bf28}+3.40\%$
test_keys_stack_nested_locked	0.2008ms	0.1414ms	7.0722 KOps/s	6.9730 KOps/s	$\color{#35bf28}+1.42\%$
test_values	7.7795μs	1.1804μs	847.1778 KOps/s	861.0168 KOps/s	$\color{#d91a1a}-1.61\%$
test_values_nested	0.1860ms	51.9600μs	19.2456 KOps/s	19.6810 KOps/s	$\color{#d91a1a}-2.21\%$
test_values_nested_locked	0.3054ms	51.9235μs	19.2591 KOps/s	19.7050 KOps/s	$\color{#d91a1a}-2.26\%$
test_values_nested_leaf	0.1040ms	46.9658μs	21.2921 KOps/s	21.5650 KOps/s	$\color{#d91a1a}-1.27\%$
test_values_stack_nested	97.9220μs	52.7980μs	18.9401 KOps/s	19.3199 KOps/s	$\color{#d91a1a}-1.97\%$
test_values_stack_nested_leaf	77.7950μs	46.5575μs	21.4788 KOps/s	21.7479 KOps/s	$\color{#d91a1a}-1.24\%$
test_values_stack_nested_locked	86.8210μs	52.7520μs	18.9566 KOps/s	19.4860 KOps/s	$\color{#d91a1a}-2.72\%$
test_membership	16.0300μs	1.3555μs	737.7584 KOps/s	735.2388 KOps/s	$\color{#35bf28}+0.34\%$
test_membership_nested	32.1200μs	3.5112μs	284.8028 KOps/s	288.7505 KOps/s	$\color{#d91a1a}-1.37\%$
test_membership_nested_leaf	22.4920μs	3.5336μs	282.9990 KOps/s	292.3248 KOps/s	$\color{#d91a1a}-3.19\%$
test_membership_stacked_nested	25.9490μs	3.4788μs	287.4529 KOps/s	258.5084 KOps/s	$\textbf{\color{#35bf28}+11.20\%}$
test_membership_stacked_nested_leaf	0.1193ms	3.5389μs	282.5765 KOps/s	294.6031 KOps/s	$\color{#d91a1a}-4.08\%$
test_membership_nested_last	0.2241ms	4.2577μs	234.8708 KOps/s	241.9792 KOps/s	$\color{#d91a1a}-2.94\%$
test_membership_nested_leaf_last	30.8580μs	4.2638μs	234.5310 KOps/s	238.9135 KOps/s	$\color{#d91a1a}-1.83\%$
test_membership_stacked_nested_last	28.1220μs	4.8239μs	207.3001 KOps/s	208.1500 KOps/s	$\color{#d91a1a}-0.41\%$
test_membership_stacked_nested_leaf_last	21.8300μs	4.8551μs	205.9678 KOps/s	208.3851 KOps/s	$\color{#d91a1a}-1.16\%$
test_nested_getleaf	52.3760μs	10.8712μs	91.9861 KOps/s	92.9227 KOps/s	$\color{#d91a1a}-1.01\%$
test_nested_get	50.7850μs	10.2488μs	97.5729 KOps/s	98.6171 KOps/s	$\color{#d91a1a}-1.06\%$
test_stacked_getleaf	0.2615ms	11.2197μs	89.1293 KOps/s	94.2741 KOps/s	$\textbf{\color{#d91a1a}-5.46\%}$
test_stacked_get	57.3580μs	10.2121μs	97.9230 KOps/s	99.5787 KOps/s	$\color{#d91a1a}-1.66\%$
test_nested_getitemleaf	45.1540μs	11.2492μs	88.8951 KOps/s	87.3725 KOps/s	$\color{#35bf28}+1.74\%$
test_nested_getitem	30.6970μs	10.7002μs	93.4562 KOps/s	89.7092 KOps/s	$\color{#35bf28}+4.18\%$
test_stacked_getitemleaf	49.9930μs	11.3728μs	87.9293 KOps/s	88.0735 KOps/s	$\color{#d91a1a}-0.16\%$
test_stacked_getitem	31.6090μs	10.5192μs	95.0645 KOps/s	96.0376 KOps/s	$\color{#d91a1a}-1.01\%$
test_lock_nested	60.1212ms	0.4054ms	2.4667 KOps/s	2.8716 KOps/s	$\textbf{\color{#d91a1a}-14.10\%}$
test_lock_stack_nested	0.3618ms	0.3070ms	3.2577 KOps/s	3.1984 KOps/s	$\color{#35bf28}+1.85\%$
test_unlock_nested	0.7674ms	0.3526ms	2.8360 KOps/s	2.4406 KOps/s	$\textbf{\color{#35bf28}+16.20\%}$
test_unlock_stack_nested	0.6053ms	0.3155ms	3.1694 KOps/s	3.1344 KOps/s	$\color{#35bf28}+1.12\%$
test_flatten_speed	0.2329ms	94.4436μs	10.5883 KOps/s	10.3727 KOps/s	$\color{#35bf28}+2.08\%$
test_unflatten_speed	0.7386ms	0.4138ms	2.4165 KOps/s	2.4702 KOps/s	$\color{#d91a1a}-2.17\%$
test_common_ops	4.1928ms	0.6737ms	1.4843 KOps/s	1.3485 KOps/s	$\textbf{\color{#35bf28}+10.07\%}$
test_creation	25.5680μs	1.9351μs	516.7730 KOps/s	532.5191 KOps/s	$\color{#d91a1a}-2.96\%$
test_creation_empty	26.5000μs	8.0864μs	123.6651 KOps/s	78.9303 KOps/s	$\textbf{\color{#35bf28}+56.68\%}$
test_creation_nested_1	31.1580μs	10.9291μs	91.4984 KOps/s	67.4244 KOps/s	$\textbf{\color{#35bf28}+35.71\%}$
test_creation_nested_2	43.0700μs	14.1709μs	70.5671 KOps/s	57.0033 KOps/s	$\textbf{\color{#35bf28}+23.79\%}$
test_clone	0.1474ms	13.4029μs	74.6107 KOps/s	71.4109 KOps/s	$\color{#35bf28}+4.48\%$
test_getitem[int]	35.4860μs	11.5314μs	86.7198 KOps/s	86.3591 KOps/s	$\color{#35bf28}+0.42\%$
test_getitem[slice_int]	53.5600μs	22.6025μs	44.2428 KOps/s	43.1458 KOps/s	$\color{#35bf28}+2.54\%$
test_getitem[range]	81.9330μs	59.4488μs	16.8212 KOps/s	13.7243 KOps/s	$\textbf{\color{#35bf28}+22.56\%}$
test_getitem[tuple]	68.2270μs	19.0092μs	52.6060 KOps/s	51.7861 KOps/s	$\color{#35bf28}+1.58\%$
test_getitem[list]	0.1499ms	39.7885μs	25.1329 KOps/s	24.0321 KOps/s	$\color{#35bf28}+4.58\%$
test_setitem_dim[int]	71.6030μs	32.2757μs	30.9831 KOps/s	27.1151 KOps/s	$\textbf{\color{#35bf28}+14.26\%}$
test_setitem_dim[slice_int]	88.7860μs	57.5912μs	17.3638 KOps/s	15.8124 KOps/s	$\textbf{\color{#35bf28}+9.81\%}$
test_setitem_dim[range]	0.1375ms	79.6849μs	12.5494 KOps/s	11.5485 KOps/s	$\textbf{\color{#35bf28}+8.67\%}$
test_setitem_dim[tuple]	0.1096ms	47.1099μs	21.2270 KOps/s	17.9798 KOps/s	$\textbf{\color{#35bf28}+18.06\%}$
test_setitem	72.5450μs	18.5574μs	53.8868 KOps/s	47.4037 KOps/s	$\textbf{\color{#35bf28}+13.68\%}$
test_set	0.2609ms	17.9965μs	55.5664 KOps/s	48.6647 KOps/s	$\textbf{\color{#35bf28}+14.18\%}$
test_set_shared	3.4649ms	0.1435ms	6.9667 KOps/s	6.8637 KOps/s	$\color{#35bf28}+1.50\%$
test_update	0.2668ms	18.6893μs	53.5065 KOps/s	42.8358 KOps/s	$\textbf{\color{#35bf28}+24.91\%}$
test_update_nested	75.6810μs	26.4448μs	37.8146 KOps/s	32.4332 KOps/s	$\textbf{\color{#35bf28}+16.59\%}$
test_update__nested	76.6530μs	25.2000μs	39.6825 KOps/s	39.7929 KOps/s	$\color{#d91a1a}-0.28\%$
test_set_nested	0.2510ms	20.0234μs	49.9415 KOps/s	43.6842 KOps/s	$\textbf{\color{#35bf28}+14.32\%}$
test_set_nested_new	83.6050μs	23.9420μs	41.7676 KOps/s	36.8316 KOps/s	$\textbf{\color{#35bf28}+13.40\%}$
test_select	0.1065ms	38.9602μs	25.6672 KOps/s	23.7784 KOps/s	$\textbf{\color{#35bf28}+7.94\%}$
test_select_nested	0.1535ms	59.5686μs	16.7874 KOps/s	16.5713 KOps/s	$\color{#35bf28}+1.30\%$
test_exclude_nested	0.2666ms	0.1232ms	8.1170 KOps/s	8.4582 KOps/s	$\color{#d91a1a}-4.03\%$
test_empty[True]	0.6725ms	0.3969ms	2.5197 KOps/s	2.5412 KOps/s	$\color{#d91a1a}-0.85\%$
test_empty[False]	10.3568μs	1.1657μs	857.8258 KOps/s	847.4869 KOps/s	$\color{#35bf28}+1.22\%$
test_unbind_speed	0.4370ms	0.2563ms	3.9015 KOps/s	3.8614 KOps/s	$\color{#35bf28}+1.04\%$
test_unbind_speed_stack0	0.4567ms	0.2497ms	4.0048 KOps/s	3.9179 KOps/s	$\color{#35bf28}+2.22\%$
test_unbind_speed_stack1	91.4027ms	0.7486ms	1.3358 KOps/s	1.2658 KOps/s	$\textbf{\color{#35bf28}+5.54\%}$
test_split	74.7153ms	1.5943ms	627.2510 Ops/s	607.2579 Ops/s	$\color{#35bf28}+3.29\%$
test_chunk	75.4501ms	1.6065ms	622.4547 Ops/s	610.0858 Ops/s	$\color{#35bf28}+2.03\%$
test_creation[device0]	3.6787ms	90.7313μs	11.0215 KOps/s	11.5324 KOps/s	$\color{#d91a1a}-4.43\%$
test_creation_from_tensor	0.2741ms	88.0888μs	11.3522 KOps/s	11.5337 KOps/s	$\color{#d91a1a}-1.57\%$
test_add_one[memmap_tensor0]	0.1026ms	5.1638μs	193.6554 KOps/s	180.2155 KOps/s	$\textbf{\color{#35bf28}+7.46\%}$
test_contiguous[memmap_tensor0]	14.5470μs	0.6470μs	1.5456 MOps/s	1.5280 MOps/s	$\color{#35bf28}+1.15\%$
test_stack[memmap_tensor0]	27.0510μs	3.5061μs	285.2151 KOps/s	276.8686 KOps/s	$\color{#35bf28}+3.01\%$
test_memmaptd_index	0.9813ms	0.2575ms	3.8835 KOps/s	3.5727 KOps/s	$\textbf{\color{#35bf28}+8.70\%}$
test_memmaptd_index_astensor	0.7616ms	0.3325ms	3.0074 KOps/s	2.9703 KOps/s	$\color{#35bf28}+1.25\%$
test_memmaptd_index_op	0.9619ms	0.5702ms	1.7536 KOps/s	1.5604 KOps/s	$\textbf{\color{#35bf28}+12.38\%}$
test_serialize_model	0.1830s	0.1172s	8.5322 Ops/s	8.5727 Ops/s	$\color{#d91a1a}-0.47\%$
test_serialize_model_pickle	0.4496s	0.3783s	2.6433 Ops/s	2.6097 Ops/s	$\color{#35bf28}+1.29\%$
test_serialize_weights	0.1818s	0.1154s	8.6652 Ops/s	8.6607 Ops/s	$\color{#35bf28}+0.05\%$
test_serialize_weights_returnearly	0.2049s	0.1401s	7.1376 Ops/s	7.7251 Ops/s	$\textbf{\color{#d91a1a}-7.60\%}$
test_serialize_weights_pickle	0.6565s	0.4709s	2.1238 Ops/s	2.3445 Ops/s	$\textbf{\color{#d91a1a}-9.42\%}$
test_serialize_weights_filesystem	0.1043s	96.0515ms	10.4111 Ops/s	10.2154 Ops/s	$\color{#35bf28}+1.92\%$
test_serialize_model_filesystem	97.1398ms	95.7112ms	10.4481 Ops/s	10.4403 Ops/s	$\color{#35bf28}+0.08\%$
test_reshape_pytree	85.1080μs	25.6878μs	38.9290 KOps/s	38.2473 KOps/s	$\color{#35bf28}+1.78\%$
test_reshape_td	0.1228ms	34.1973μs	29.2421 KOps/s	28.6946 KOps/s	$\color{#35bf28}+1.91\%$
test_view_pytree	72.2350μs	25.7987μs	38.7616 KOps/s	39.5824 KOps/s	$\color{#d91a1a}-2.07\%$
test_view_td	0.2911ms	38.3156μs	26.0990 KOps/s	25.5811 KOps/s	$\color{#35bf28}+2.02\%$
test_unbind_pytree	69.1790μs	29.2518μs	34.1859 KOps/s	34.0998 KOps/s	$\color{#35bf28}+0.25\%$
test_unbind_td	0.3954ms	37.1197μs	26.9399 KOps/s	25.9088 KOps/s	$\color{#35bf28}+3.98\%$
test_split_pytree	66.2440μs	29.6138μs	33.7681 KOps/s	33.7791 KOps/s	$\color{#d91a1a}-0.03\%$
test_split_td	0.1438ms	40.1568μs	24.9024 KOps/s	24.2583 KOps/s	$\color{#35bf28}+2.66\%$
test_add_pytree	97.8020μs	34.7723μs	28.7585 KOps/s	28.2876 KOps/s	$\color{#35bf28}+1.66\%$
test_add_td	0.2394ms	50.9015μs	19.6458 KOps/s	17.1104 KOps/s	$\textbf{\color{#35bf28}+14.82\%}$
test_distributed	0.2613ms	0.1022ms	9.7876 KOps/s	9.7104 KOps/s	$\color{#35bf28}+0.79\%$
test_tdmodule	0.1167ms	16.7109μs	59.8412 KOps/s	53.9485 KOps/s	$\textbf{\color{#35bf28}+10.92\%}$
test_tdmodule_dispatch	53.0890μs	31.9296μs	31.3189 KOps/s	26.9609 KOps/s	$\textbf{\color{#35bf28}+16.16\%}$
test_tdseq	40.9260μs	19.1277μs	52.2803 KOps/s	46.1922 KOps/s	$\textbf{\color{#35bf28}+13.18\%}$
test_tdseq_dispatch	56.5560μs	36.5074μs	27.3917 KOps/s	23.2326 KOps/s	$\textbf{\color{#35bf28}+17.90\%}$
test_instantiation_functorch	1.6304ms	1.3385ms	747.0866 Ops/s	760.2517 Ops/s	$\color{#d91a1a}-1.73\%$
test_instantiation_td	1.7355ms	1.0292ms	971.6668 Ops/s	979.2481 Ops/s	$\color{#d91a1a}-0.77\%$
test_exec_functorch	0.3541ms	0.1602ms	6.2405 KOps/s	6.2446 KOps/s	$\color{#d91a1a}-0.07\%$
test_exec_functional_call	0.2843ms	0.1511ms	6.6164 KOps/s	6.7112 KOps/s	$\color{#d91a1a}-1.41\%$
test_exec_td	0.3508ms	0.1443ms	6.9281 KOps/s	6.9466 KOps/s	$\color{#d91a1a}-0.27\%$
test_exec_td_decorator	1.0686ms	0.2242ms	4.4611 KOps/s	4.4832 KOps/s	$\color{#d91a1a}-0.49\%$
test_vmap_mlp_speed[True-True]	0.8384ms	0.4844ms	2.0645 KOps/s	2.0342 KOps/s	$\color{#35bf28}+1.49\%$
test_vmap_mlp_speed[True-False]	0.7777ms	0.4825ms	2.0725 KOps/s	1.9294 KOps/s	$\textbf{\color{#35bf28}+7.42\%}$
test_vmap_mlp_speed[False-True]	1.8596ms	0.4092ms	2.4437 KOps/s	2.5014 KOps/s	$\color{#d91a1a}-2.31\%$
test_vmap_mlp_speed[False-False]	0.5857ms	0.3947ms	2.5333 KOps/s	2.5091 KOps/s	$\color{#35bf28}+0.96\%$
test_vmap_mlp_speed_decorator[True-True]	1.3080ms	0.5540ms	1.8052 KOps/s	1.7648 KOps/s	$\color{#35bf28}+2.28\%$
test_vmap_mlp_speed_decorator[True-False]	0.8860ms	0.5522ms	1.8108 KOps/s	1.7740 KOps/s	$\color{#35bf28}+2.07\%$
test_vmap_mlp_speed_decorator[False-True]	0.7375ms	0.4617ms	2.1658 KOps/s	2.1552 KOps/s	$\color{#35bf28}+0.49\%$
test_vmap_mlp_speed_decorator[False-False]	1.0619ms	0.4643ms	2.1539 KOps/s	2.1535 KOps/s	$\color{#35bf28}+0.02\%$
test_to_module_speed[True]	2.4842ms	1.7271ms	578.9984 Ops/s	591.8048 Ops/s	$\color{#d91a1a}-2.16\%$
test_to_module_speed[False]	2.3633ms	1.7065ms	585.9995 Ops/s	603.2827 Ops/s	$\color{#d91a1a}-2.86\%$
test_tc_init	54.8320μs	23.4081μs	42.7203 KOps/s	32.9424 KOps/s	$\textbf{\color{#35bf28}+29.68\%}$
test_tc_init_nested	0.1555ms	44.9665μs	22.2388 KOps/s	17.0740 KOps/s	$\textbf{\color{#35bf28}+30.25\%}$
test_tc_first_layer_tensor	4.9061μs	0.7001μs	1.4283 MOps/s	1.4565 MOps/s	$\color{#d91a1a}-1.93\%$
test_tc_first_layer_nontensor	1.8605μs	0.6673μs	1.4986 MOps/s	1.4757 MOps/s	$\color{#35bf28}+1.55\%$
test_tc_second_layer_tensor	26.3390μs	1.8322μs	545.7951 KOps/s	551.0102 KOps/s	$\color{#d91a1a}-0.95\%$
test_tc_second_layer_nontensor	9.5277μs	1.5123μs	661.2433 KOps/s	667.6953 KOps/s	$\color{#d91a1a}-0.97\%$
test_unbind	96.6495ms	8.6316ms	115.8528 Ops/s	118.7328 Ops/s	$\color{#d91a1a}-2.43\%$
test_full_like	17.2720ms	12.3558ms	80.9335 Ops/s	85.6447 Ops/s	$\textbf{\color{#d91a1a}-5.50\%}$
test_zeros_like	15.4765ms	6.3587ms	157.2643 Ops/s	155.8501 Ops/s	$\color{#35bf28}+0.91\%$
test_ones_like	11.3124ms	6.7885ms	147.3073 Ops/s	147.9961 Ops/s	$\color{#d91a1a}-0.47\%$
test_clone	16.1451ms	8.9150ms	112.1699 Ops/s	119.6761 Ops/s	$\textbf{\color{#d91a1a}-6.27\%}$
test_squeeze	72.9960μs	14.4515μs	69.1971 KOps/s	72.4818 KOps/s	$\color{#d91a1a}-4.53\%$
test_unsqueeze	0.1261ms	60.8614μs	16.4308 KOps/s	16.1226 KOps/s	$\color{#35bf28}+1.91\%$
test_split	0.2116ms	0.1128ms	8.8622 KOps/s	8.9370 KOps/s	$\color{#d91a1a}-0.84\%$
test_permute	0.3046ms	0.1286ms	7.7787 KOps/s	7.7788 KOps/s	$-0.00\%$
test_stack	30.5940ms	24.4430ms	40.9116 Ops/s	41.9292 Ops/s	$\color{#d91a1a}-2.43\%$
test_cat	31.6828ms	24.4839ms	40.8432 Ops/s	41.5337 Ops/s	$\color{#d91a1a}-1.66\%$

[github-actions]

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

Total Benchmarks: 152. Improved: $\large\color{#35bf28}25$. Worsened: $\large\color{#d91a1a}3$.

Expand to view detailed results

Name	Max	Mean	Ops	Ops on Repo HEAD	Change
test_plain_set_nested	0.1004ms	12.9904μs	76.9797 KOps/s	77.1153 KOps/s	$\color{#d91a1a}-0.18\%$
test_plain_set_stack_nested	31.4810μs	13.1357μs	76.1287 KOps/s	76.3613 KOps/s	$\color{#d91a1a}-0.30\%$
test_plain_set_nested_inplace	0.2067ms	14.2610μs	70.1214 KOps/s	70.1881 KOps/s	$\color{#d91a1a}-0.09\%$
test_plain_set_stack_nested_inplace	0.1980ms	14.4189μs	69.3533 KOps/s	69.3531 KOps/s	$+0.00\%$
test_items	0.1894ms	4.6691μs	214.1728 KOps/s	211.6821 KOps/s	$\color{#35bf28}+1.18\%$
test_items_nested	0.5318ms	0.3421ms	2.9235 KOps/s	2.9362 KOps/s	$\color{#d91a1a}-0.43\%$
test_items_nested_locked	0.5413ms	0.3489ms	2.8658 KOps/s	2.9243 KOps/s	$\color{#d91a1a}-2.00\%$
test_items_nested_leaf	0.1133ms	84.1317μs	11.8861 KOps/s	11.9609 KOps/s	$\color{#d91a1a}-0.62\%$
test_items_stack_nested	0.3739ms	0.3407ms	2.9349 KOps/s	2.8897 KOps/s	$\color{#35bf28}+1.56\%$
test_items_stack_nested_leaf	0.1754ms	86.2343μs	11.5963 KOps/s	11.8780 KOps/s	$\color{#d91a1a}-2.37\%$
test_items_stack_nested_locked	0.4000ms	0.3454ms	2.8950 KOps/s	2.9028 KOps/s	$\color{#d91a1a}-0.27\%$
test_keys	21.6510μs	4.3779μs	228.4208 KOps/s	227.6078 KOps/s	$\color{#35bf28}+0.36\%$
test_keys_nested	85.0320μs	67.8008μs	14.7491 KOps/s	14.8178 KOps/s	$\color{#d91a1a}-0.46\%$
test_keys_nested_locked	0.7827ms	73.8347μs	13.5438 KOps/s	13.7844 KOps/s	$\color{#d91a1a}-1.75\%$
test_keys_nested_leaf	93.4620μs	58.4386μs	17.1120 KOps/s	17.3024 KOps/s	$\color{#d91a1a}-1.10\%$
test_keys_stack_nested	0.1028ms	68.7897μs	14.5371 KOps/s	14.8697 KOps/s	$\color{#d91a1a}-2.24\%$
test_keys_stack_nested_leaf	82.3510μs	58.8872μs	16.9816 KOps/s	17.1654 KOps/s	$\color{#d91a1a}-1.07\%$
test_keys_stack_nested_locked	0.1062ms	74.0430μs	13.5057 KOps/s	13.9233 KOps/s	$\color{#d91a1a}-3.00\%$
test_values	9.3737μs	1.8346μs	545.0735 KOps/s	545.8520 KOps/s	$\color{#d91a1a}-0.14\%$
test_values_nested	0.1566ms	35.3028μs	28.3263 KOps/s	28.4248 KOps/s	$\color{#d91a1a}-0.35\%$
test_values_nested_locked	56.6310μs	37.2735μs	26.8287 KOps/s	26.5308 KOps/s	$\color{#35bf28}+1.12\%$
test_values_nested_leaf	47.4310μs	31.3232μs	31.9252 KOps/s	32.1625 KOps/s	$\color{#d91a1a}-0.74\%$
test_values_stack_nested	60.3410μs	35.8326μs	27.9075 KOps/s	27.5492 KOps/s	$\color{#35bf28}+1.30\%$
test_values_stack_nested_leaf	47.8810μs	32.1079μs	31.1450 KOps/s	31.1228 KOps/s	$\color{#35bf28}+0.07\%$
test_values_stack_nested_locked	52.8810μs	37.7996μs	26.4553 KOps/s	25.7928 KOps/s	$\color{#35bf28}+2.57\%$
test_membership	3.3057μs	0.7381μs	1.3548 MOps/s	1.3562 MOps/s	$\color{#d91a1a}-0.11\%$
test_membership_nested	61.5210μs	2.5739μs	388.5148 KOps/s	383.8238 KOps/s	$\color{#35bf28}+1.22\%$
test_membership_nested_leaf	18.9410μs	2.5973μs	385.0203 KOps/s	386.0992 KOps/s	$\color{#d91a1a}-0.28\%$
test_membership_stacked_nested	21.1310μs	2.5914μs	385.8970 KOps/s	384.3342 KOps/s	$\color{#35bf28}+0.41\%$
test_membership_stacked_nested_leaf	19.4400μs	2.5859μs	386.7177 KOps/s	388.8727 KOps/s	$\color{#d91a1a}-0.55\%$
test_membership_nested_last	18.5210μs	3.1158μs	320.9443 KOps/s	319.8583 KOps/s	$\color{#35bf28}+0.34\%$
test_membership_nested_leaf_last	19.8200μs	3.1096μs	321.5804 KOps/s	320.4185 KOps/s	$\color{#35bf28}+0.36\%$
test_membership_stacked_nested_last	24.5100μs	3.0984μs	322.7506 KOps/s	281.9671 KOps/s	$\textbf{\color{#35bf28}+14.46\%}$
test_membership_stacked_nested_leaf_last	16.0500μs	3.1138μs	321.1544 KOps/s	281.0003 KOps/s	$\textbf{\color{#35bf28}+14.29\%}$
test_nested_getleaf	53.6510μs	8.3491μs	119.7741 KOps/s	119.2427 KOps/s	$\color{#35bf28}+0.45\%$
test_nested_get	23.2000μs	7.8833μs	126.8512 KOps/s	126.6812 KOps/s	$\color{#35bf28}+0.13\%$
test_stacked_getleaf	30.1600μs	8.4057μs	118.9671 KOps/s	118.3994 KOps/s	$\color{#35bf28}+0.48\%$
test_stacked_get	24.1910μs	7.9066μs	126.4767 KOps/s	125.1162 KOps/s	$\color{#35bf28}+1.09\%$
test_nested_getitemleaf	0.1440ms	8.5540μs	116.9050 KOps/s	116.9990 KOps/s	$\color{#d91a1a}-0.08\%$
test_nested_getitem	22.7300μs	8.0628μs	124.0258 KOps/s	124.0704 KOps/s	$\color{#d91a1a}-0.04\%$
test_stacked_getitemleaf	22.6200μs	8.5978μs	116.3091 KOps/s	115.4514 KOps/s	$\color{#35bf28}+0.74\%$
test_stacked_getitem	29.9600μs	8.0682μs	123.9440 KOps/s	122.8375 KOps/s	$\color{#35bf28}+0.90\%$
test_lock_nested	57.5754ms	0.4063ms	2.4612 KOps/s	2.3387 KOps/s	$\textbf{\color{#35bf28}+5.24\%}$
test_lock_stack_nested	0.3777ms	0.3063ms	3.2642 KOps/s	3.1988 KOps/s	$\color{#35bf28}+2.04\%$
test_unlock_nested	59.2340ms	0.4101ms	2.4384 KOps/s	2.7459 KOps/s	$\textbf{\color{#d91a1a}-11.20\%}$
test_unlock_stack_nested	0.4163ms	0.3134ms	3.1907 KOps/s	3.1077 KOps/s	$\color{#35bf28}+2.67\%$
test_flatten_speed	0.2738ms	0.1032ms	9.6867 KOps/s	9.7332 KOps/s	$\color{#d91a1a}-0.48\%$
test_unflatten_speed	0.4909ms	0.2978ms	3.3575 KOps/s	3.4481 KOps/s	$\color{#d91a1a}-2.63\%$
test_common_ops	1.1765ms	0.6031ms	1.6580 KOps/s	1.5943 KOps/s	$\color{#35bf28}+4.00\%$
test_creation	16.9000μs	1.6863μs	593.0283 KOps/s	592.7644 KOps/s	$\color{#35bf28}+0.04\%$
test_creation_empty	0.2043ms	9.2940μs	107.5960 KOps/s	108.6429 KOps/s	$\color{#d91a1a}-0.96\%$
test_creation_nested_1	37.8510μs	11.0453μs	90.5364 KOps/s	91.4645 KOps/s	$\color{#d91a1a}-1.01\%$
test_creation_nested_2	31.1310μs	13.3503μs	74.9047 KOps/s	76.4294 KOps/s	$\color{#d91a1a}-1.99\%$
test_clone	68.6310μs	11.6809μs	85.6096 KOps/s	78.2995 KOps/s	$\textbf{\color{#35bf28}+9.34\%}$
test_getitem[int]	33.3110μs	10.9589μs	91.2499 KOps/s	87.7883 KOps/s	$\color{#35bf28}+3.94\%$
test_getitem[slice_int]	43.1310μs	21.0283μs	47.5551 KOps/s	45.5897 KOps/s	$\color{#35bf28}+4.31\%$
test_getitem[range]	67.5420μs	47.8410μs	20.9026 KOps/s	20.8156 KOps/s	$\color{#35bf28}+0.42\%$
test_getitem[tuple]	48.5010μs	18.7960μs	53.2027 KOps/s	51.5726 KOps/s	$\color{#35bf28}+3.16\%$
test_getitem[list]	0.1600ms	34.3182μs	29.1391 KOps/s	27.1697 KOps/s	$\textbf{\color{#35bf28}+7.25\%}$
test_setitem_dim[int]	75.8110μs	30.1852μs	33.1288 KOps/s	31.4886 KOps/s	$\textbf{\color{#35bf28}+5.21\%}$
test_setitem_dim[slice_int]	0.1757ms	50.7379μs	19.7091 KOps/s	19.0531 KOps/s	$\color{#35bf28}+3.44\%$
test_setitem_dim[range]	88.3810μs	68.9637μs	14.5004 KOps/s	14.1992 KOps/s	$\color{#35bf28}+2.12\%$
test_setitem_dim[tuple]	68.8210μs	45.2917μs	22.0791 KOps/s	21.6357 KOps/s	$\color{#35bf28}+2.05\%$
test_setitem	0.1101ms	16.7773μs	59.6042 KOps/s	55.6846 KOps/s	$\textbf{\color{#35bf28}+7.04\%}$
test_set	69.8310μs	16.3561μs	61.1392 KOps/s	57.2855 KOps/s	$\textbf{\color{#35bf28}+6.73\%}$
test_set_shared	1.0396ms	99.4263μs	10.0577 KOps/s	9.3514 KOps/s	$\textbf{\color{#35bf28}+7.55\%}$
test_update	83.2420μs	18.4937μs	54.0724 KOps/s	46.4774 KOps/s	$\textbf{\color{#35bf28}+16.34\%}$
test_update_nested	68.3710μs	23.8762μs	41.8827 KOps/s	39.1844 KOps/s	$\textbf{\color{#35bf28}+6.89\%}$
test_update__nested	0.1208ms	22.0631μs	45.3246 KOps/s	40.6638 KOps/s	$\textbf{\color{#35bf28}+11.46\%}$
test_set_nested	54.8710μs	17.2856μs	57.8515 KOps/s	53.9638 KOps/s	$\textbf{\color{#35bf28}+7.20\%}$
test_set_nested_new	59.5310μs	20.3965μs	49.0281 KOps/s	45.5441 KOps/s	$\textbf{\color{#35bf28}+7.65\%}$
test_select	87.8720μs	33.2514μs	30.0739 KOps/s	26.9332 KOps/s	$\textbf{\color{#35bf28}+11.66\%}$
test_select_nested	86.4310μs	55.6407μs	17.9724 KOps/s	18.2330 KOps/s	$\color{#d91a1a}-1.43\%$
test_exclude_nested	0.1887ms	0.1102ms	9.0754 KOps/s	8.9946 KOps/s	$\color{#35bf28}+0.90\%$
test_empty[True]	0.4136ms	0.3494ms	2.8620 KOps/s	2.8491 KOps/s	$\color{#35bf28}+0.45\%$
test_empty[False]	2.0786μs	0.9240μs	1.0822 MOps/s	1.0711 MOps/s	$\color{#35bf28}+1.04\%$
test_to	0.1032ms	78.1064μs	12.8031 KOps/s	12.7346 KOps/s	$\color{#35bf28}+0.54\%$
test_to_nonblocking	0.2141ms	63.3908μs	15.7752 KOps/s	15.1690 KOps/s	$\color{#35bf28}+4.00\%$
test_unbind_speed	0.3302ms	0.2654ms	3.7676 KOps/s	3.6315 KOps/s	$\color{#35bf28}+3.75\%$
test_unbind_speed_stack0	0.3960ms	0.2663ms	3.7554 KOps/s	3.6596 KOps/s	$\color{#35bf28}+2.62\%$
test_unbind_speed_stack1	75.0457ms	0.8380ms	1.1934 KOps/s	1.1579 KOps/s	$\color{#35bf28}+3.06\%$
test_split	75.2094ms	1.6997ms	588.3245 Ops/s	640.8934 Ops/s	$\textbf{\color{#d91a1a}-8.20\%}$
test_chunk	1.6080ms	1.5685ms	637.5373 Ops/s	594.8318 Ops/s	$\textbf{\color{#35bf28}+7.18\%}$
test_creation[device0]	0.1992ms	59.1409μs	16.9088 KOps/s	15.6778 KOps/s	$\textbf{\color{#35bf28}+7.85\%}$
test_creation_from_tensor	0.2042ms	56.3072μs	17.7597 KOps/s	16.6699 KOps/s	$\textbf{\color{#35bf28}+6.54\%}$
test_add_one[memmap_tensor0]	0.1366ms	7.1421μs	140.0158 KOps/s	131.4054 KOps/s	$\textbf{\color{#35bf28}+6.55\%}$
test_contiguous[memmap_tensor0]	20.3610μs	0.7185μs	1.3918 MOps/s	1.4124 MOps/s	$\color{#d91a1a}-1.46\%$
test_stack[memmap_tensor0]	42.2410μs	5.0880μs	196.5407 KOps/s	195.0689 KOps/s	$\color{#35bf28}+0.75\%$
test_memmaptd_index	1.1502ms	0.2953ms	3.3866 KOps/s	3.2982 KOps/s	$\color{#35bf28}+2.68\%$
test_memmaptd_index_astensor	0.6388ms	0.3658ms	2.7334 KOps/s	2.6595 KOps/s	$\color{#35bf28}+2.78\%$
test_memmaptd_index_op	1.1584ms	0.6796ms	1.4715 KOps/s	1.3986 KOps/s	$\textbf{\color{#35bf28}+5.21\%}$
test_serialize_model	0.1064s	0.1034s	9.6732 Ops/s	8.5488 Ops/s	$\textbf{\color{#35bf28}+13.15\%}$
test_serialize_model_pickle	1.3661s	1.2391s	0.8071 Ops/s	0.8074 Ops/s	$\color{#d91a1a}-0.04\%$
test_serialize_weights	0.1802s	0.1104s	9.0557 Ops/s	8.7058 Ops/s	$\color{#35bf28}+4.02\%$
test_serialize_weights_returnearly	0.2423s	0.1019s	9.8128 Ops/s	10.1879 Ops/s	$\color{#d91a1a}-3.68\%$
test_serialize_weights_pickle	1.3533s	1.2365s	0.8087 Ops/s	0.8009 Ops/s	$\color{#35bf28}+0.98\%$
test_reshape_pytree	93.3810μs	26.4478μs	37.8104 KOps/s	35.4348 KOps/s	$\textbf{\color{#35bf28}+6.70\%}$
test_reshape_td	90.5820μs	31.9686μs	31.2807 KOps/s	31.0524 KOps/s	$\color{#35bf28}+0.74\%$
test_view_pytree	0.1705ms	26.0873μs	38.3328 KOps/s	37.2699 KOps/s	$\color{#35bf28}+2.85\%$
test_view_td	66.6510μs	36.1114μs	27.6921 KOps/s	26.4914 KOps/s	$\color{#35bf28}+4.53\%$
test_unbind_pytree	64.7310μs	32.0318μs	31.2190 KOps/s	30.2990 KOps/s	$\color{#35bf28}+3.04\%$
test_unbind_td	0.4014ms	41.2026μs	24.2703 KOps/s	23.7051 KOps/s	$\color{#35bf28}+2.38\%$
test_split_pytree	57.6710μs	34.9665μs	28.5988 KOps/s	28.0048 KOps/s	$\color{#35bf28}+2.12\%$
test_split_td	0.1073ms	40.5883μs	24.6376 KOps/s	24.7891 KOps/s	$\color{#d91a1a}-0.61\%$
test_add_pytree	0.1834ms	39.5709μs	25.2711 KOps/s	23.7788 KOps/s	$\textbf{\color{#35bf28}+6.28\%}$
test_add_td	0.2255ms	54.2830μs	18.4220 KOps/s	18.2438 KOps/s	$\color{#35bf28}+0.98\%$
test_distributed	4.1594ms	80.7307μs	12.3869 KOps/s	10.4972 KOps/s	$\textbf{\color{#35bf28}+18.00\%}$
test_tdmodule	0.1422ms	15.6800μs	63.7753 KOps/s	66.3276 KOps/s	$\color{#d91a1a}-3.85\%$
test_tdmodule_dispatch	50.9010μs	30.1538μs	33.1633 KOps/s	33.3456 KOps/s	$\color{#d91a1a}-0.55\%$
test_tdseq	34.0200μs	17.4082μs	57.4442 KOps/s	57.8409 KOps/s	$\color{#d91a1a}-0.69\%$
test_tdseq_dispatch	50.1910μs	33.7372μs	29.6409 KOps/s	29.6784 KOps/s	$\color{#d91a1a}-0.13\%$
test_instantiation_functorch	1.7200ms	1.5706ms	636.7174 Ops/s	629.2309 Ops/s	$\color{#35bf28}+1.19\%$
test_instantiation_td	1.5387ms	1.0766ms	928.8247 Ops/s	851.7520 Ops/s	$\textbf{\color{#35bf28}+9.05\%}$
test_exec_functorch	0.2239ms	0.1563ms	6.3975 KOps/s	6.2070 KOps/s	$\color{#35bf28}+3.07\%$
test_exec_functional_call	0.3137ms	0.1435ms	6.9674 KOps/s	6.8144 KOps/s	$\color{#35bf28}+2.24\%$
test_exec_td	0.1707ms	0.1387ms	7.2111 KOps/s	6.8823 KOps/s	$\color{#35bf28}+4.78\%$
test_exec_td_decorator	0.4949ms	0.2188ms	4.5694 KOps/s	4.5753 KOps/s	$\color{#d91a1a}-0.13\%$
test_vmap_mlp_speed[True-True]	0.8193ms	0.6203ms	1.6120 KOps/s	1.5972 KOps/s	$\color{#35bf28}+0.92\%$
test_vmap_mlp_speed[True-False]	0.7819ms	0.6185ms	1.6167 KOps/s	1.5993 KOps/s	$\color{#35bf28}+1.09\%$
test_vmap_mlp_speed[False-True]	0.7264ms	0.5688ms	1.7580 KOps/s	1.8068 KOps/s	$\color{#d91a1a}-2.70\%$
test_vmap_mlp_speed[False-False]	0.7062ms	0.5494ms	1.8203 KOps/s	1.8127 KOps/s	$\color{#35bf28}+0.42\%$
test_vmap_mlp_speed_decorator[True-True]	1.1395ms	0.6870ms	1.4557 KOps/s	1.4396 KOps/s	$\color{#35bf28}+1.12\%$
test_vmap_mlp_speed_decorator[True-False]	0.8485ms	0.6849ms	1.4601 KOps/s	1.4479 KOps/s	$\color{#35bf28}+0.84\%$
test_vmap_mlp_speed_decorator[False-True]	0.7765ms	0.6086ms	1.6431 KOps/s	1.6346 KOps/s	$\color{#35bf28}+0.52\%$
test_vmap_mlp_speed_decorator[False-False]	0.7993ms	0.6092ms	1.6416 KOps/s	1.6390 KOps/s	$\color{#35bf28}+0.16\%$
test_vmap_transformer_speed[True-True]	8.4705ms	8.1799ms	122.2509 Ops/s	120.1829 Ops/s	$\color{#35bf28}+1.72\%$
test_vmap_transformer_speed[True-False]	8.6357ms	8.2111ms	121.7867 Ops/s	120.4085 Ops/s	$\color{#35bf28}+1.14\%$
test_vmap_transformer_speed[False-True]	8.3459ms	8.1078ms	123.3381 Ops/s	121.4150 Ops/s	$\color{#35bf28}+1.58\%$
test_vmap_transformer_speed[False-False]	8.8301ms	8.1633ms	122.5000 Ops/s	121.5368 Ops/s	$\color{#35bf28}+0.79\%$
test_vmap_transformer_speed_decorator[True-True]	20.2955ms	19.9154ms	50.2124 Ops/s	49.6955 Ops/s	$\color{#35bf28}+1.04\%$
test_vmap_transformer_speed_decorator[True-False]	20.4971ms	19.9208ms	50.1989 Ops/s	49.7917 Ops/s	$\color{#35bf28}+0.82\%$
test_vmap_transformer_speed_decorator[False-True]	20.2115ms	19.7983ms	50.5094 Ops/s	49.9736 Ops/s	$\color{#35bf28}+1.07\%$
test_vmap_transformer_speed_decorator[False-False]	20.0132ms	19.7769ms	50.5642 Ops/s	50.0538 Ops/s	$\color{#35bf28}+1.02\%$
test_to_module_speed[True]	2.1141ms	1.5723ms	635.9973 Ops/s	642.4245 Ops/s	$\color{#d91a1a}-1.00\%$
test_to_module_speed[False]	1.6535ms	1.5365ms	650.8329 Ops/s	647.7895 Ops/s	$\color{#35bf28}+0.47\%$
test_tc_init	85.8310μs	25.3702μs	39.4163 KOps/s	39.3794 KOps/s	$\color{#35bf28}+0.09\%$
test_tc_init_nested	93.7910μs	50.7250μs	19.7141 KOps/s	19.2576 KOps/s	$\color{#35bf28}+2.37\%$
test_tc_first_layer_tensor	0.7558μs	0.3745μs	2.6700 MOps/s	2.6742 MOps/s	$\color{#d91a1a}-0.16\%$
test_tc_first_layer_nontensor	4.0462μs	0.4038μs	2.4762 MOps/s	2.5066 MOps/s	$\color{#d91a1a}-1.21\%$
test_tc_second_layer_tensor	4.2720μs	0.9975μs	1.0025 MOps/s	999.4594 KOps/s	$\color{#35bf28}+0.31\%$
test_tc_second_layer_nontensor	4.4250μs	0.8523μs	1.1733 MOps/s	1.1656 MOps/s	$\color{#35bf28}+0.66\%$
test_unbind	92.7126ms	6.6425ms	150.5448 Ops/s	188.2249 Ops/s	$\textbf{\color{#d91a1a}-20.02\%}$
test_full_like	14.2627ms	13.4837ms	74.1633 Ops/s	73.1956 Ops/s	$\color{#35bf28}+1.32\%$
test_zeros_like	8.2594ms	7.8389ms	127.5697 Ops/s	126.6101 Ops/s	$\color{#35bf28}+0.76\%$
test_ones_like	8.3005ms	7.8327ms	127.6701 Ops/s	128.1007 Ops/s	$\color{#d91a1a}-0.34\%$
test_clone	9.9532ms	9.4858ms	105.4208 Ops/s	105.0317 Ops/s	$\color{#35bf28}+0.37\%$
test_squeeze	0.1438ms	11.3147μs	88.3808 KOps/s	89.4251 KOps/s	$\color{#d91a1a}-1.17\%$
test_unsqueeze	0.1828ms	53.8997μs	18.5530 KOps/s	18.4146 KOps/s	$\color{#35bf28}+0.75\%$
test_split	0.2280ms	0.1017ms	9.8323 KOps/s	9.7586 KOps/s	$\color{#35bf28}+0.75\%$
test_permute	0.2351ms	0.1145ms	8.7370 KOps/s	8.7716 KOps/s	$\color{#d91a1a}-0.39\%$
test_stack	28.8452ms	27.8106ms	35.9576 Ops/s	36.0777 Ops/s	$\color{#d91a1a}-0.33\%$
test_cat	28.3947ms	27.7569ms	36.0271 Ops/s	36.3131 Ops/s	$\color{#d91a1a}-0.79\%$