[BugFix] Better tensor allocation in memmap_like

[vmoens]

The memmap_like function does not use _set_str, therefore a to(...) can be called on memmap tensors, leading them to be regular tensors.
This PR solves this issue

[github-actions]

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

Total Benchmarks: 105. Improved: $\large\color{#35bf28}13$. Worsened: $\large\color{#d91a1a}29$.

Expand to view detailed results

Name	Max	Mean	Ops	Ops on Repo HEAD	Change
test_plain_set_nested	0.1253ms	20.3075μs	49.2428 KOps/s	45.9230 KOps/s	$\textbf{\color{#35bf28}+7.23\%}$
test_plain_set_stack_nested	1.5014ms	0.2062ms	4.8497 KOps/s	4.3702 KOps/s	$\textbf{\color{#35bf28}+10.97\%}$
test_plain_set_nested_inplace	2.0826ms	29.1581μs	34.2958 KOps/s	33.7351 KOps/s	$\color{#35bf28}+1.66\%$
test_plain_set_stack_nested_inplace	5.1725ms	0.2842ms	3.5192 KOps/s	3.8326 KOps/s	$\textbf{\color{#d91a1a}-8.18\%}$
test_items	2.1288ms	3.9267μs	254.6648 KOps/s	256.3579 KOps/s	$\color{#d91a1a}-0.66\%$
test_items_nested	2.6436ms	0.3958ms	2.5268 KOps/s	2.5080 KOps/s	$\color{#35bf28}+0.75\%$
test_items_nested_locked	6.5170ms	0.3962ms	2.5241 KOps/s	2.4876 KOps/s	$\color{#35bf28}+1.47\%$
test_items_nested_leaf	3.4541ms	0.2445ms	4.0900 KOps/s	4.0189 KOps/s	$\color{#35bf28}+1.77\%$
test_items_stack_nested	4.8019ms	2.4640ms	405.8515 Ops/s	392.1884 Ops/s	$\color{#35bf28}+3.48\%$
test_items_stack_nested_leaf	7.4461ms	2.3652ms	422.7905 Ops/s	447.0910 Ops/s	$\textbf{\color{#d91a1a}-5.44\%}$
test_items_stack_nested_locked	6.9820ms	1.3643ms	732.9740 Ops/s	782.3738 Ops/s	$\textbf{\color{#d91a1a}-6.31\%}$
test_keys	1.3312ms	5.3127μs	188.2290 KOps/s	192.7224 KOps/s	$\color{#d91a1a}-2.33\%$
test_keys_nested	2.3374ms	0.1892ms	5.2854 KOps/s	4.9317 KOps/s	$\textbf{\color{#35bf28}+7.17\%}$
test_keys_nested_locked	9.1093ms	0.1867ms	5.3563 KOps/s	5.2806 KOps/s	$\color{#35bf28}+1.43\%$
test_keys_nested_leaf	2.8880ms	0.1853ms	5.3960 KOps/s	5.3787 KOps/s	$\color{#35bf28}+0.32\%$
test_keys_stack_nested	7.6089ms	2.3127ms	432.3881 Ops/s	417.1650 Ops/s	$\color{#35bf28}+3.65\%$
test_keys_stack_nested_leaf	5.3414ms	2.4231ms	412.6955 Ops/s	417.2773 Ops/s	$\color{#d91a1a}-1.10\%$
test_keys_stack_nested_locked	2.6028ms	1.1112ms	899.8917 Ops/s	883.0447 Ops/s	$\color{#35bf28}+1.91\%$
test_values	0.3191ms	1.4503μs	689.5179 KOps/s	694.7675 KOps/s	$\color{#d91a1a}-0.76\%$
test_values_nested	2.4649ms	69.4485μs	14.3992 KOps/s	13.9199 KOps/s	$\color{#35bf28}+3.44\%$
test_values_nested_locked	1.2254ms	66.6422μs	15.0055 KOps/s	14.1805 KOps/s	$\textbf{\color{#35bf28}+5.82\%}$
test_values_nested_leaf	4.9540ms	60.7980μs	16.4479 KOps/s	15.6836 KOps/s	$\color{#35bf28}+4.87\%$
test_values_stack_nested	6.2717ms	1.9920ms	502.0035 Ops/s	508.0448 Ops/s	$\color{#d91a1a}-1.19\%$
test_values_stack_nested_leaf	3.9445ms	1.9755ms	506.2131 Ops/s	494.4853 Ops/s	$\color{#35bf28}+2.37\%$
test_values_stack_nested_locked	2.5155ms	0.9026ms	1.1079 KOps/s	1.1423 KOps/s	$\color{#d91a1a}-3.02\%$
test_membership	1.4753ms	2.1855μs	457.5635 KOps/s	558.9156 KOps/s	$\textbf{\color{#d91a1a}-18.13\%}$
test_membership_nested	1.2034ms	4.3104μs	231.9959 KOps/s	248.5675 KOps/s	$\textbf{\color{#d91a1a}-6.67\%}$
test_membership_nested_leaf	0.9206ms	4.3520μs	229.7783 KOps/s	235.5521 KOps/s	$\color{#d91a1a}-2.45\%$
test_membership_stacked_nested	1.1331ms	16.2057μs	61.7066 KOps/s	61.8999 KOps/s	$\color{#d91a1a}-0.31\%$
test_membership_stacked_nested_leaf	0.4260ms	15.9529μs	62.6846 KOps/s	58.8257 KOps/s	$\textbf{\color{#35bf28}+6.56\%}$
test_membership_nested_last	4.5696ms	8.7793μs	113.9049 KOps/s	118.6792 KOps/s	$\color{#d91a1a}-4.02\%$
test_membership_nested_leaf_last	1.2039ms	8.7112μs	114.7946 KOps/s	114.0584 KOps/s	$\color{#35bf28}+0.65\%$
test_membership_stacked_nested_last	2.9628ms	0.2751ms	3.6356 KOps/s	3.6095 KOps/s	$\color{#35bf28}+0.72\%$
test_membership_stacked_nested_leaf_last	1.2531ms	20.0917μs	49.7717 KOps/s	49.3476 KOps/s	$\color{#35bf28}+0.86\%$
test_nested_getleaf	1.4224ms	17.0221μs	58.7472 KOps/s	55.0062 KOps/s	$\textbf{\color{#35bf28}+6.80\%}$
test_nested_get	0.3420ms	16.0486μs	62.3105 KOps/s	59.8196 KOps/s	$\color{#35bf28}+4.16\%$
test_stacked_getleaf	3.2808ms	1.0902ms	917.2474 Ops/s	924.4855 Ops/s	$\color{#d91a1a}-0.78\%$
test_stacked_get	6.9611ms	1.0563ms	946.6659 Ops/s	973.6335 Ops/s	$\color{#d91a1a}-2.77\%$
test_nested_getitemleaf	1.4626ms	17.1725μs	58.2326 KOps/s	56.3858 KOps/s	$\color{#35bf28}+3.28\%$
test_nested_getitem	1.1629ms	16.3148μs	61.2942 KOps/s	60.5595 KOps/s	$\color{#35bf28}+1.21\%$
test_stacked_getitemleaf	6.8877ms	1.0980ms	910.7560 Ops/s	893.2261 Ops/s	$\color{#35bf28}+1.96\%$
test_stacked_getitem	3.1988ms	1.0213ms	979.1509 Ops/s	1.0193 KOps/s	$\color{#d91a1a}-3.94\%$
test_lock_nested	83.5075ms	1.9299ms	518.1509 Ops/s	541.1430 Ops/s	$\color{#d91a1a}-4.25\%$
test_lock_stack_nested	0.1116s	24.4534ms	40.8941 Ops/s	39.1593 Ops/s	$\color{#35bf28}+4.43\%$
test_unlock_nested	79.7144ms	1.8756ms	533.1647 Ops/s	511.8164 Ops/s	$\color{#35bf28}+4.17\%$
test_unlock_stack_nested	0.1217s	25.2719ms	39.5696 Ops/s	38.7493 Ops/s	$\color{#35bf28}+2.12\%$
test_flatten_speed	6.2838ms	1.1203ms	892.6516 Ops/s	849.1616 Ops/s	$\textbf{\color{#35bf28}+5.12\%}$
test_unflatten_speed	4.8611ms	2.0570ms	486.1485 Ops/s	477.8297 Ops/s	$\color{#35bf28}+1.74\%$
test_common_ops	3.7408ms	1.6600ms	602.4168 Ops/s	627.6735 Ops/s	$\color{#d91a1a}-4.02\%$
test_creation	0.8265ms	6.5975μs	151.5727 KOps/s	141.8989 KOps/s	$\textbf{\color{#35bf28}+6.82\%}$
test_creation_empty	1.2935ms	15.5652μs	64.2458 KOps/s	60.5570 KOps/s	$\textbf{\color{#35bf28}+6.09\%}$
test_creation_nested_1	1.5741ms	28.8176μs	34.7010 KOps/s	30.7995 KOps/s	$\textbf{\color{#35bf28}+12.67\%}$
test_creation_nested_2	1.0136ms	32.2518μs	31.0060 KOps/s	28.4411 KOps/s	$\textbf{\color{#35bf28}+9.02\%}$
test_clone	0.9406ms	33.1811μs	30.1376 KOps/s	30.4864 KOps/s	$\color{#d91a1a}-1.14\%$
test_getitem[int]	4.9104ms	38.1325μs	26.2243 KOps/s	26.6930 KOps/s	$\color{#d91a1a}-1.76\%$
test_getitem[slice_int]	1.6649ms	79.1019μs	12.6419 KOps/s	12.8225 KOps/s	$\color{#d91a1a}-1.41\%$
test_getitem[range]	1.2555ms	0.1380ms	7.2479 KOps/s	8.0378 KOps/s	$\textbf{\color{#d91a1a}-9.83\%}$
test_getitem[tuple]	2.5602ms	70.2189μs	14.2412 KOps/s	15.8705 KOps/s	$\textbf{\color{#d91a1a}-10.27\%}$
test_getitem[list]	1.4524ms	0.1341ms	7.4592 KOps/s	7.7725 KOps/s	$\color{#d91a1a}-4.03\%$
test_setitem_dim[int]	0.8447ms	61.4634μs	16.2698 KOps/s	16.5410 KOps/s	$\color{#d91a1a}-1.64\%$
test_setitem_dim[slice_int]	2.6030ms	0.1170ms	8.5487 KOps/s	10.4188 KOps/s	$\textbf{\color{#d91a1a}-17.95\%}$
test_setitem_dim[range]	1.7557ms	0.1504ms	6.6478 KOps/s	7.1310 KOps/s	$\textbf{\color{#d91a1a}-6.78\%}$
test_setitem_dim[tuple]	2.7720ms	83.3356μs	11.9997 KOps/s	12.1379 KOps/s	$\color{#d91a1a}-1.14\%$
test_setitem	2.8571ms	49.2693μs	20.2966 KOps/s	21.8481 KOps/s	$\textbf{\color{#d91a1a}-7.10\%}$
test_set	1.9870ms	47.0715μs	21.2443 KOps/s	22.4526 KOps/s	$\textbf{\color{#d91a1a}-5.38\%}$
test_set_shared	2.9542ms	0.3275ms	3.0532 KOps/s	2.9161 KOps/s	$\color{#35bf28}+4.70\%$
test_update	4.8988ms	55.7000μs	17.9533 KOps/s	19.2858 KOps/s	$\textbf{\color{#d91a1a}-6.91\%}$
test_update_nested	3.0663ms	73.0738μs	13.6848 KOps/s	12.8469 KOps/s	$\textbf{\color{#35bf28}+6.52\%}$
test_set_nested	8.3263ms	53.5936μs	18.6590 KOps/s	20.1380 KOps/s	$\textbf{\color{#d91a1a}-7.34\%}$
test_set_nested_new	3.2019ms	76.6758μs	13.0419 KOps/s	13.8633 KOps/s	$\textbf{\color{#d91a1a}-5.92\%}$
test_select	4.1147ms	0.1399ms	7.1479 KOps/s	7.7206 KOps/s	$\textbf{\color{#d91a1a}-7.42\%}$
test_unbind_speed	3.7550ms	0.7821ms	1.2786 KOps/s	1.3348 KOps/s	$\color{#d91a1a}-4.21\%$
test_unbind_speed_stack0	88.3994ms	10.4818ms	95.4036 Ops/s	96.8713 Ops/s	$\color{#d91a1a}-1.52\%$
test_unbind_speed_stack1	4.6152ms	1.2389μs	807.1704 KOps/s	1.0472 MOps/s	$\textbf{\color{#d91a1a}-22.92\%}$
test_creation[device0]	6.0431ms	0.6073ms	1.6467 KOps/s	1.6446 KOps/s	$\color{#35bf28}+0.13\%$
test_creation_from_tensor	7.9095ms	0.6778ms	1.4755 KOps/s	1.6545 KOps/s	$\textbf{\color{#d91a1a}-10.82\%}$
test_add_one[memmap_tensor0]	3.9124ms	72.7928μs	13.7376 KOps/s	15.8867 KOps/s	$\textbf{\color{#d91a1a}-13.53\%}$
test_contiguous[memmap_tensor0]	1.0459ms	13.5892μs	73.5880 KOps/s	80.8446 KOps/s	$\textbf{\color{#d91a1a}-8.98\%}$
test_stack[memmap_tensor0]	2.2030ms	47.8967μs	20.8783 KOps/s	21.9940 KOps/s	$\textbf{\color{#d91a1a}-5.07\%}$
test_memmaptd_index	5.1818ms	0.4187ms	2.3881 KOps/s	2.5861 KOps/s	$\textbf{\color{#d91a1a}-7.66\%}$
test_memmaptd_index_astensor	7.4053ms	1.9253ms	519.3883 Ops/s	488.3396 Ops/s	$\textbf{\color{#35bf28}+6.36\%}$
test_memmaptd_index_op	7.4785ms	5.1341ms	194.7777 Ops/s	190.9614 Ops/s	$\color{#35bf28}+2.00\%$
test_reshape_pytree	2.6013ms	41.5962μs	24.0407 KOps/s	25.0614 KOps/s	$\color{#d91a1a}-4.07\%$
test_reshape_td	1.4945ms	52.9956μs	18.8695 KOps/s	18.6939 KOps/s	$\color{#35bf28}+0.94\%$
test_view_pytree	2.6258ms	38.8592μs	25.7339 KOps/s	26.5494 KOps/s	$\color{#d91a1a}-3.07\%$
test_view_td	0.8918ms	10.0651μs	99.3529 KOps/s	101.9492 KOps/s	$\color{#d91a1a}-2.55\%$
test_unbind_pytree	1.0609ms	48.0864μs	20.7959 KOps/s	21.6465 KOps/s	$\color{#d91a1a}-3.93\%$
test_unbind_td	2.3601ms	0.1310ms	7.6343 KOps/s	8.3943 KOps/s	$\textbf{\color{#d91a1a}-9.05\%}$
test_split_pytree	2.3407ms	51.1323μs	19.5571 KOps/s	22.8530 KOps/s	$\textbf{\color{#d91a1a}-14.42\%}$
test_split_td	3.0771ms	0.1546ms	6.4679 KOps/s	6.8749 KOps/s	$\textbf{\color{#d91a1a}-5.92\%}$
test_add_pytree	2.8924ms	76.7669μs	13.0264 KOps/s	14.6333 KOps/s	$\textbf{\color{#d91a1a}-10.98\%}$
test_add_td	2.6833ms	0.1383ms	7.2291 KOps/s	7.2776 KOps/s	$\color{#d91a1a}-0.67\%$
test_distributed	0.7993ms	9.4978μs	105.2871 KOps/s	111.5351 KOps/s	$\textbf{\color{#d91a1a}-5.60\%}$
test_tdmodule	0.5962ms	40.5425μs	24.6655 KOps/s	25.6911 KOps/s	$\color{#d91a1a}-3.99\%$
test_tdmodule_dispatch	0.8099ms	78.0050μs	12.8197 KOps/s	13.0100 KOps/s	$\color{#d91a1a}-1.46\%$
test_tdseq	0.4077ms	48.3397μs	20.6869 KOps/s	22.2486 KOps/s	$\textbf{\color{#d91a1a}-7.02\%}$
test_tdseq_dispatch	0.3341ms	95.1994μs	10.5043 KOps/s	10.0374 KOps/s	$\color{#35bf28}+4.65\%$
test_instantiation_functorch	6.3699ms	2.2143ms	451.6078 Ops/s	495.9035 Ops/s	$\textbf{\color{#d91a1a}-8.93\%}$
test_instantiation_td	4.9764ms	1.6871ms	592.7490 Ops/s	612.6192 Ops/s	$\color{#d91a1a}-3.24\%$
test_exec_functorch	1.5201ms	0.3077ms	3.2501 KOps/s	3.3403 KOps/s	$\color{#d91a1a}-2.70\%$
test_exec_td	2.7371ms	0.3219ms	3.1065 KOps/s	3.2950 KOps/s	$\textbf{\color{#d91a1a}-5.72\%}$
test_vmap_mlp_speed[True-True]	16.3034ms	2.0686ms	483.4082 Ops/s	495.9900 Ops/s	$\color{#d91a1a}-2.54\%$
test_vmap_mlp_speed[True-False]	5.8715ms	1.1130ms	898.5097 Ops/s	911.0957 Ops/s	$\color{#d91a1a}-1.38\%$
test_vmap_mlp_speed[False-True]	13.7816ms	1.8334ms	545.4367 Ops/s	553.5525 Ops/s	$\color{#d91a1a}-1.47\%$
test_vmap_mlp_speed[False-False]	12.0079ms	0.8848ms	1.1302 KOps/s	1.1749 KOps/s	$\color{#d91a1a}-3.80\%$

[matteobettini]

LGTM