[Feature] Cudagraphs

[vmoens]

Added CudaGraphModule class to provide a user-friendly interface to CUDA graphs for PyTorch callables.

This class enables fast, CPU-overhead-free execution of operations on GPU while ensuring essential checks for input functions. Includes documentation and example usage.

cc @mikaylagawarecki @albanD @eellison @BoyuanFeng @Chillee

[github-actions]

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

Total Benchmarks: 222. Improved: $\large\color{#35bf28}30$. Worsened: $\large\color{#d91a1a}10$.

Expand to view detailed results

Name	Max	Mean	Ops	Ops on Repo HEAD	Change
test_plain_set_nested	44.9230μs	20.2935μs	49.2769 KOps/s	48.1950 KOps/s	$\color{#35bf28}+2.24\%$
test_plain_set_stack_nested	40.4560μs	20.2741μs	49.3240 KOps/s	46.7545 KOps/s	$\textbf{\color{#35bf28}+5.50\%}$
test_plain_set_nested_inplace	64.9810μs	22.1341μs	45.1791 KOps/s	44.2036 KOps/s	$\color{#35bf28}+2.21\%$
test_plain_set_stack_nested_inplace	69.3900μs	22.3665μs	44.7098 KOps/s	44.7502 KOps/s	$\color{#d91a1a}-0.09\%$
test_items	18.0340μs	4.1504μs	240.9414 KOps/s	236.3517 KOps/s	$\color{#35bf28}+1.94\%$
test_items_nested	0.5622ms	0.3298ms	3.0324 KOps/s	3.0677 KOps/s	$\color{#d91a1a}-1.15\%$
test_items_nested_locked	0.6390ms	0.3299ms	3.0314 KOps/s	3.0491 KOps/s	$\color{#d91a1a}-0.58\%$
test_items_nested_leaf	0.1573ms	84.9869μs	11.7665 KOps/s	11.6020 KOps/s	$\color{#35bf28}+1.42\%$
test_items_stack_nested	0.6346ms	0.3315ms	3.0166 KOps/s	3.0437 KOps/s	$\color{#d91a1a}-0.89\%$
test_items_stack_nested_leaf	0.1584ms	86.6158μs	11.5452 KOps/s	11.5263 KOps/s	$\color{#35bf28}+0.16\%$
test_items_stack_nested_locked	0.6486ms	0.3379ms	2.9593 KOps/s	3.0109 KOps/s	$\color{#d91a1a}-1.72\%$
test_keys	23.0130μs	3.6910μs	270.9295 KOps/s	277.0112 KOps/s	$\color{#d91a1a}-2.20\%$
test_keys_nested	0.2683ms	97.6110μs	10.2448 KOps/s	10.5283 KOps/s	$\color{#d91a1a}-2.69\%$
test_keys_nested_locked	0.7958ms	0.1025ms	9.7570 KOps/s	10.1924 KOps/s	$\color{#d91a1a}-4.27\%$
test_keys_nested_leaf	0.1583ms	82.6219μs	12.1033 KOps/s	12.7868 KOps/s	$\textbf{\color{#d91a1a}-5.35\%}$
test_keys_stack_nested	0.2019ms	97.6924μs	10.2362 KOps/s	10.4511 KOps/s	$\color{#d91a1a}-2.06\%$
test_keys_stack_nested_leaf	0.1796ms	82.4706μs	12.1255 KOps/s	12.5887 KOps/s	$\color{#d91a1a}-3.68\%$
test_keys_stack_nested_locked	0.3090ms	0.1025ms	9.7589 KOps/s	10.0299 KOps/s	$\color{#d91a1a}-2.70\%$
test_values	10.8282μs	1.0767μs	928.7664 KOps/s	1.0048 MOps/s	$\textbf{\color{#d91a1a}-7.56\%}$
test_values_nested	96.7000μs	47.7035μs	20.9628 KOps/s	21.1891 KOps/s	$\color{#d91a1a}-1.07\%$
test_values_nested_locked	0.1220ms	47.3255μs	21.1303 KOps/s	21.0452 KOps/s	$\color{#35bf28}+0.40\%$
test_values_nested_leaf	97.7920μs	42.0639μs	23.7733 KOps/s	23.7393 KOps/s	$\color{#35bf28}+0.14\%$
test_values_stack_nested	98.8570μs	47.6410μs	20.9903 KOps/s	20.8766 KOps/s	$\color{#35bf28}+0.54\%$
test_values_stack_nested_leaf	0.1074ms	42.4419μs	23.5616 KOps/s	23.9206 KOps/s	$\color{#d91a1a}-1.50\%$
test_values_stack_nested_locked	95.8890μs	48.1760μs	20.7572 KOps/s	21.0088 KOps/s	$\color{#d91a1a}-1.20\%$
test_membership	17.4620μs	0.9270μs	1.0788 MOps/s	1.1799 MOps/s	$\textbf{\color{#d91a1a}-8.57\%}$
test_membership_nested	49.2120μs	2.5806μs	387.4999 KOps/s	379.7985 KOps/s	$\color{#35bf28}+2.03\%$
test_membership_nested_leaf	20.3480μs	2.5677μs	389.4509 KOps/s	365.7579 KOps/s	$\textbf{\color{#35bf28}+6.48\%}$
test_membership_stacked_nested	50.0130μs	2.5734μs	388.5879 KOps/s	386.2551 KOps/s	$\color{#35bf28}+0.60\%$
test_membership_stacked_nested_leaf	17.6330μs	2.5907μs	385.9894 KOps/s	381.2843 KOps/s	$\color{#35bf28}+1.23\%$
test_membership_nested_last	50.9050μs	3.8850μs	257.4010 KOps/s	263.9171 KOps/s	$\color{#d91a1a}-2.47\%$
test_membership_nested_leaf_last	40.5360μs	3.8082μs	262.5890 KOps/s	261.4246 KOps/s	$\color{#35bf28}+0.45\%$
test_membership_stacked_nested_last	44.7830μs	3.8342μs	260.8085 KOps/s	187.8979 KOps/s	$\textbf{\color{#35bf28}+38.80\%}$
test_membership_stacked_nested_leaf_last	29.1440μs	3.8692μs	258.4527 KOps/s	186.9683 KOps/s	$\textbf{\color{#35bf28}+38.23\%}$
test_nested_getleaf	52.3370μs	11.2425μs	88.9478 KOps/s	93.7320 KOps/s	$\textbf{\color{#d91a1a}-5.10\%}$
test_nested_get	65.3420μs	10.0614μs	99.3897 KOps/s	97.6181 KOps/s	$\color{#35bf28}+1.81\%$
test_stacked_getleaf	36.0280μs	10.6302μs	94.0716 KOps/s	92.5325 KOps/s	$\color{#35bf28}+1.66\%$
test_stacked_get	54.0810μs	10.0832μs	99.1753 KOps/s	98.6730 KOps/s	$\color{#35bf28}+0.51\%$
test_nested_getitemleaf	59.5710μs	10.9185μs	91.5873 KOps/s	90.1808 KOps/s	$\color{#35bf28}+1.56\%$
test_nested_getitem	34.6640μs	10.1537μs	98.4863 KOps/s	96.3202 KOps/s	$\color{#35bf28}+2.25\%$
test_stacked_getitemleaf	55.7540μs	10.9515μs	91.3116 KOps/s	91.5176 KOps/s	$\color{#d91a1a}-0.23\%$
test_stacked_getitem	31.0270μs	10.1550μs	98.4733 KOps/s	98.7140 KOps/s	$\color{#d91a1a}-0.24\%$
test_lock_nested	1.1537ms	0.4853ms	2.0608 KOps/s	2.0636 KOps/s	$\color{#d91a1a}-0.14\%$
test_lock_stack_nested	0.8272ms	0.4533ms	2.2062 KOps/s	2.2181 KOps/s	$\color{#d91a1a}-0.54\%$
test_unlock_nested	95.0731ms	0.4981ms	2.0076 KOps/s	2.4515 KOps/s	$\textbf{\color{#d91a1a}-18.11\%}$
test_unlock_stack_nested	0.7198ms	0.3687ms	2.7121 KOps/s	2.6688 KOps/s	$\color{#35bf28}+1.62\%$
test_flatten_speed	0.6353ms	0.1096ms	9.1200 KOps/s	9.5276 KOps/s	$\color{#d91a1a}-4.28\%$
test_unflatten_speed	0.6418ms	0.4692ms	2.1312 KOps/s	2.1521 KOps/s	$\color{#d91a1a}-0.97\%$
test_common_ops	5.2125ms	1.1294ms	885.4351 Ops/s	886.7956 Ops/s	$\color{#d91a1a}-0.15\%$
test_creation	29.6960μs	2.0851μs	479.5972 KOps/s	478.8909 KOps/s	$\color{#35bf28}+0.15\%$
test_creation_empty	79.5810μs	17.4802μs	57.2077 KOps/s	54.3249 KOps/s	$\textbf{\color{#35bf28}+5.31\%}$
test_creation_nested_1	46.8570μs	20.2265μs	49.4402 KOps/s	46.6593 KOps/s	$\textbf{\color{#35bf28}+5.96\%}$
test_creation_nested_2	94.4130μs	24.3210μs	41.1167 KOps/s	37.7734 KOps/s	$\textbf{\color{#35bf28}+8.85\%}$
test_clone	0.1906ms	17.3746μs	57.5553 KOps/s	55.1652 KOps/s	$\color{#35bf28}+4.33\%$
test_getitem[int]	0.7693ms	16.4333μs	60.8519 KOps/s	58.6557 KOps/s	$\color{#35bf28}+3.74\%$
test_getitem[slice_int]	0.1361ms	30.0574μs	33.2697 KOps/s	31.8464 KOps/s	$\color{#35bf28}+4.47\%$
test_getitem[range]	0.1920ms	58.0061μs	17.2396 KOps/s	17.2734 KOps/s	$\color{#d91a1a}-0.20\%$
test_getitem[tuple]	0.1291ms	25.0450μs	39.9281 KOps/s	38.9486 KOps/s	$\color{#35bf28}+2.51\%$
test_getitem[list]	0.3352ms	53.3772μs	18.7346 KOps/s	18.8283 KOps/s	$\color{#d91a1a}-0.50\%$
test_setitem_dim[int]	54.5210μs	33.3627μs	29.9736 KOps/s	29.0004 KOps/s	$\color{#35bf28}+3.36\%$
test_setitem_dim[slice_int]	0.1061ms	61.2532μs	16.3257 KOps/s	15.8716 KOps/s	$\color{#35bf28}+2.86\%$
test_setitem_dim[range]	0.1560ms	84.1818μs	11.8790 KOps/s	11.4403 KOps/s	$\color{#35bf28}+3.83\%$
test_setitem_dim[tuple]	0.1299ms	51.2740μs	19.5031 KOps/s	19.6915 KOps/s	$\color{#d91a1a}-0.96\%$
test_setitem	0.1974ms	30.3870μs	32.9088 KOps/s	32.2092 KOps/s	$\color{#35bf28}+2.17\%$
test_set	0.1545ms	29.9652μs	33.3721 KOps/s	32.7659 KOps/s	$\color{#35bf28}+1.85\%$
test_set_shared	1.3080ms	0.2217ms	4.5113 KOps/s	4.7228 KOps/s	$\color{#d91a1a}-4.48\%$
test_update	0.1823ms	36.6645μs	27.2743 KOps/s	26.7648 KOps/s	$\color{#35bf28}+1.90\%$
test_update_nested	0.1850ms	47.1423μs	21.2124 KOps/s	20.5825 KOps/s	$\color{#35bf28}+3.06\%$
test_update__nested	0.1902ms	35.7440μs	27.9767 KOps/s	27.1634 KOps/s	$\color{#35bf28}+2.99\%$
test_set_nested	0.1955ms	32.3786μs	30.8846 KOps/s	29.9505 KOps/s	$\color{#35bf28}+3.12\%$
test_set_nested_new	0.1753ms	37.7761μs	26.4718 KOps/s	25.7835 KOps/s	$\color{#35bf28}+2.67\%$
test_select	0.2235ms	55.0188μs	18.1756 KOps/s	17.4702 KOps/s	$\color{#35bf28}+4.04\%$
test_select_nested	0.1162ms	59.4482μs	16.8214 KOps/s	15.5991 KOps/s	$\textbf{\color{#35bf28}+7.84\%}$
test_exclude_nested	0.1409ms	75.4135μs	13.2602 KOps/s	12.5679 KOps/s	$\textbf{\color{#35bf28}+5.51\%}$
test_empty[True]	0.5865ms	0.3153ms	3.1714 KOps/s	3.1584 KOps/s	$\color{#35bf28}+0.41\%$
test_empty[False]	13.1043μs	1.1972μs	835.3161 KOps/s	815.3780 KOps/s	$\color{#35bf28}+2.45\%$
test_unbind_speed	0.5586ms	0.2950ms	3.3895 KOps/s	3.2168 KOps/s	$\textbf{\color{#35bf28}+5.37\%}$
test_unbind_speed_stack0	0.5148ms	0.2951ms	3.3885 KOps/s	3.3218 KOps/s	$\color{#35bf28}+2.01\%$
test_unbind_speed_stack1	0.1002s	0.8245ms	1.2129 KOps/s	1.3490 KOps/s	$\textbf{\color{#d91a1a}-10.09\%}$
test_split	2.2368ms	1.9894ms	502.6717 Ops/s	442.3783 Ops/s	$\textbf{\color{#35bf28}+13.63\%}$
test_chunk	0.1004s	2.3722ms	421.5580 Ops/s	437.1694 Ops/s	$\color{#d91a1a}-3.57\%$
test_creation[device0]	0.2404ms	0.1180ms	8.4764 KOps/s	8.3648 KOps/s	$\color{#35bf28}+1.33\%$
test_creation_from_tensor	3.5688ms	0.1190ms	8.4058 KOps/s	8.6127 KOps/s	$\color{#d91a1a}-2.40\%$
test_add_one[memmap_tensor0]	0.1924ms	7.5459μs	132.5227 KOps/s	131.8736 KOps/s	$\color{#35bf28}+0.49\%$
test_contiguous[memmap_tensor0]	28.4940μs	1.8931μs	528.2217 KOps/s	519.2956 KOps/s	$\color{#35bf28}+1.72\%$
test_stack[memmap_tensor0]	51.8160μs	5.5005μs	181.8027 KOps/s	173.6110 KOps/s	$\color{#35bf28}+4.72\%$
test_memmaptd_index	1.0791ms	0.3992ms	2.5048 KOps/s	2.4579 KOps/s	$\color{#35bf28}+1.91\%$
test_memmaptd_index_astensor	0.9725ms	0.4803ms	2.0818 KOps/s	2.0242 KOps/s	$\color{#35bf28}+2.85\%$
test_memmaptd_index_op	1.5788ms	1.0090ms	991.0662 Ops/s	961.3237 Ops/s	$\color{#35bf28}+3.09\%$
test_serialize_model	0.1286s	0.1187s	8.4239 Ops/s	8.2945 Ops/s	$\color{#35bf28}+1.56\%$
test_serialize_model_pickle	0.4501s	0.3885s	2.5743 Ops/s	2.5043 Ops/s	$\color{#35bf28}+2.80\%$
test_serialize_weights	0.1253s	0.1166s	8.5768 Ops/s	8.4700 Ops/s	$\color{#35bf28}+1.26\%$
test_serialize_weights_returnearly	0.2714s	0.1739s	5.7518 Ops/s	6.3371 Ops/s	$\textbf{\color{#d91a1a}-9.24\%}$
test_serialize_weights_pickle	0.4633s	0.4071s	2.4563 Ops/s	1.0700 Ops/s	$\textbf{\color{#35bf28}+129.57\%}$
test_serialize_weights_filesystem	0.1454s	0.1393s	7.1805 Ops/s	6.9728 Ops/s	$\color{#35bf28}+2.98\%$
test_serialize_model_filesystem	0.1592s	0.1494s	6.6917 Ops/s	6.1864 Ops/s	$\textbf{\color{#35bf28}+8.17\%}$
test_reshape_pytree	87.9230μs	38.1607μs	26.2050 KOps/s	26.0186 KOps/s	$\color{#35bf28}+0.72\%$
test_reshape_td	96.0180μs	45.4522μs	22.0011 KOps/s	20.0203 KOps/s	$\textbf{\color{#35bf28}+9.89\%}$
test_view_pytree	0.1150ms	37.7418μs	26.4958 KOps/s	25.9513 KOps/s	$\color{#35bf28}+2.10\%$
test_view_td	0.1323ms	51.1945μs	19.5333 KOps/s	18.1305 KOps/s	$\textbf{\color{#35bf28}+7.74\%}$
test_unbind_pytree	0.1091ms	35.3821μs	28.2629 KOps/s	28.0645 KOps/s	$\color{#35bf28}+0.71\%$
test_unbind_td	0.3179ms	44.0500μs	22.7015 KOps/s	22.1488 KOps/s	$\color{#35bf28}+2.50\%$
test_split_pytree	0.1150ms	37.8935μs	26.3898 KOps/s	26.7752 KOps/s	$\color{#d91a1a}-1.44\%$
test_split_td	0.4719ms	56.7056μs	17.6350 KOps/s	16.9400 KOps/s	$\color{#35bf28}+4.10\%$
test_add_pytree	0.1046ms	44.2861μs	22.5805 KOps/s	22.6278 KOps/s	$\color{#d91a1a}-0.21\%$
test_add_td	0.1751ms	80.8281μs	12.3719 KOps/s	11.5184 KOps/s	$\textbf{\color{#35bf28}+7.41\%}$
test_compile_add_one_nested[tensordict-compile]	0.1230ms	55.9847μs	17.8620 KOps/s	17.5972 KOps/s	$\color{#35bf28}+1.51\%$
test_compile_add_one_nested[tensordict-eager]	0.4491ms	0.1913ms	5.2285 KOps/s	5.1628 KOps/s	$\color{#35bf28}+1.27\%$
test_compile_add_one_nested[pytree-compile]	0.1092ms	55.6432μs	17.9717 KOps/s	17.5278 KOps/s	$\color{#35bf28}+2.53\%$
test_compile_add_one_nested[pytree-eager]	0.3122ms	0.1432ms	6.9844 KOps/s	7.1369 KOps/s	$\color{#d91a1a}-2.14\%$
test_compile_copy_nested[tensordict-compile]	0.1330ms	20.5550μs	48.6501 KOps/s	47.5267 KOps/s	$\color{#35bf28}+2.36\%$
test_compile_copy_nested[tensordict-eager]	0.1475ms	67.8650μs	14.7351 KOps/s	13.9720 KOps/s	$\textbf{\color{#35bf28}+5.46\%}$
test_compile_copy_nested[pytree-compile]	0.1517ms	77.0162μs	12.9843 KOps/s	13.3580 KOps/s	$\color{#d91a1a}-2.80\%$
test_compile_copy_nested[pytree-eager]	0.1437ms	69.6883μs	14.3496 KOps/s	14.8160 KOps/s	$\color{#d91a1a}-3.15\%$
test_compile_add_one_flat[tensordict-compile]	0.3579ms	0.1709ms	5.8531 KOps/s	5.7329 KOps/s	$\color{#35bf28}+2.10\%$
test_compile_add_one_flat[tensordict-eager]	0.3771ms	0.1887ms	5.3006 KOps/s	4.9766 KOps/s	$\textbf{\color{#35bf28}+6.51\%}$
test_compile_add_one_flat[tensorclass-compile]	0.1056ms	45.5960μs	21.9318 KOps/s	20.9762 KOps/s	$\color{#35bf28}+4.56\%$
test_compile_add_one_flat[tensorclass-eager]	0.6287ms	70.1591μs	14.2533 KOps/s	14.0212 KOps/s	$\color{#35bf28}+1.66\%$
test_compile_add_one_flat[pytree-compile]	0.5138ms	0.1782ms	5.6127 KOps/s	5.7211 KOps/s	$\color{#d91a1a}-1.90\%$
test_compile_add_one_flat[pytree-eager]	0.5411ms	0.2925ms	3.4194 KOps/s	3.4210 KOps/s	$\color{#d91a1a}-0.05\%$
test_compile_add_self_flat[tensordict-eager]	0.3260ms	0.2009ms	4.9766 KOps/s	4.7272 KOps/s	$\textbf{\color{#35bf28}+5.28\%}$
test_compile_add_self_flat[tensordict-compile]	0.3645ms	0.1729ms	5.7848 KOps/s	5.7723 KOps/s	$\color{#35bf28}+0.22\%$
test_compile_add_self_flat[tensorclass-eager]	0.2768ms	63.3542μs	15.7843 KOps/s	15.3514 KOps/s	$\color{#35bf28}+2.82\%$
test_compile_add_self_flat[tensorclass-compile]	0.1187ms	46.9298μs	21.3084 KOps/s	20.8859 KOps/s	$\color{#35bf28}+2.02\%$
test_compile_add_self_flat[pytree-eager]	0.4222ms	0.2295ms	4.3568 KOps/s	4.1970 KOps/s	$\color{#35bf28}+3.81\%$
test_compile_add_self_flat[pytree-compile]	0.2851ms	0.1744ms	5.7324 KOps/s	5.6980 KOps/s	$\color{#35bf28}+0.60\%$
test_compile_copy_flat[tensordict-compile]	0.2272ms	0.1020ms	9.8051 KOps/s	9.6413 KOps/s	$\color{#35bf28}+1.70\%$
test_compile_copy_flat[tensordict-eager]	0.1297ms	56.5739μs	17.6760 KOps/s	17.2323 KOps/s	$\color{#35bf28}+2.57\%$
test_compile_copy_flat[pytree-compile]	0.1783ms	78.0534μs	12.8117 KOps/s	13.1015 KOps/s	$\color{#d91a1a}-2.21\%$
test_compile_copy_flat[pytree-eager]	0.1296ms	69.1689μs	14.4574 KOps/s	14.0662 KOps/s	$\color{#35bf28}+2.78\%$
test_compile_assign_and_add[tensordict-compile]	0.3072ms	0.1981ms	5.0482 KOps/s	5.0185 KOps/s	$\color{#35bf28}+0.59\%$
test_compile_assign_and_add[tensordict-eager]	1.9909ms	1.6202ms	617.2111 Ops/s	581.5811 Ops/s	$\textbf{\color{#35bf28}+6.13\%}$
test_compile_assign_and_add[pytree-compile]	0.3910ms	0.1961ms	5.0991 KOps/s	5.2101 KOps/s	$\color{#d91a1a}-2.13\%$
test_compile_assign_and_add[pytree-eager]	1.3513ms	1.0979ms	910.8126 Ops/s	894.0810 Ops/s	$\color{#35bf28}+1.87\%$
test_compile_assign_and_add_stack[compile]	0.7460ms	0.4207ms	2.3772 KOps/s	2.3856 KOps/s	$\color{#d91a1a}-0.35\%$
test_compile_assign_and_add_stack[eager]	3.9110ms	3.6417ms	274.5944 Ops/s	263.9524 Ops/s	$\color{#35bf28}+4.03\%$
test_compile_indexing[tensor-tensordict-compile]	0.1221ms	32.8905μs	30.4039 KOps/s	28.2334 KOps/s	$\textbf{\color{#35bf28}+7.69\%}$
test_compile_indexing[tensor-tensordict-eager]	1.0511ms	49.0853μs	20.3727 KOps/s	20.3076 KOps/s	$\color{#35bf28}+0.32\%$
test_compile_indexing[tensor-tensorclass-compile]	0.1111ms	29.4633μs	33.9406 KOps/s	32.8467 KOps/s	$\color{#35bf28}+3.33\%$
test_compile_indexing[tensor-tensorclass-eager]	75.5410μs	29.0467μs	34.4274 KOps/s	35.3385 KOps/s	$\color{#d91a1a}-2.58\%$
test_compile_indexing[tensor-pytree-compile]	95.7290μs	28.9491μs	34.5433 KOps/s	32.9601 KOps/s	$\color{#35bf28}+4.80\%$
test_compile_indexing[tensor-pytree-eager]	73.3270μs	28.6086μs	34.9546 KOps/s	35.0490 KOps/s	$\color{#d91a1a}-0.27\%$
test_compile_indexing[slice-tensordict-compile]	0.1638ms	74.3798μs	13.4445 KOps/s	13.4693 KOps/s	$\color{#d91a1a}-0.18\%$
test_compile_indexing[slice-tensordict-eager]	0.5501ms	27.3785μs	36.5250 KOps/s	34.2444 KOps/s	$\textbf{\color{#35bf28}+6.66\%}$
test_compile_indexing[slice-tensorclass-compile]	0.1512ms	69.3353μs	14.4227 KOps/s	14.7557 KOps/s	$\color{#d91a1a}-2.26\%$
test_compile_indexing[slice-tensorclass-eager]	95.7880μs	22.3375μs	44.7678 KOps/s	43.6262 KOps/s	$\color{#35bf28}+2.62\%$
test_compile_indexing[slice-pytree-compile]	0.1462ms	69.0134μs	14.4899 KOps/s	14.6067 KOps/s	$\color{#d91a1a}-0.80\%$
test_compile_indexing[slice-pytree-eager]	74.4690μs	22.4651μs	44.5136 KOps/s	43.3276 KOps/s	$\color{#35bf28}+2.74\%$
test_compile_indexing[int-tensordict-compile]	0.1459ms	74.0870μs	13.4976 KOps/s	13.5066 KOps/s	$\color{#d91a1a}-0.07\%$
test_compile_indexing[int-tensordict-eager]	1.1046ms	27.5795μs	36.2589 KOps/s	35.0932 KOps/s	$\color{#35bf28}+3.32\%$
test_compile_indexing[int-tensorclass-compile]	0.1622ms	69.1238μs	14.4668 KOps/s	14.8093 KOps/s	$\color{#d91a1a}-2.31\%$
test_compile_indexing[int-tensorclass-eager]	0.2910ms	22.4763μs	44.4913 KOps/s	43.9466 KOps/s	$\color{#35bf28}+1.24\%$
test_compile_indexing[int-pytree-compile]	0.1705ms	69.8266μs	14.3212 KOps/s	14.6290 KOps/s	$\color{#d91a1a}-2.10\%$
test_compile_indexing[int-pytree-eager]	64.2700μs	22.1624μs	45.1215 KOps/s	44.0837 KOps/s	$\color{#35bf28}+2.35\%$
test_mod_add[eager]	86.5940μs	24.3472μs	41.0724 KOps/s	38.9623 KOps/s	$\textbf{\color{#35bf28}+5.42\%}$
test_mod_add[compile]	0.1124ms	36.7599μs	27.2035 KOps/s	25.4371 KOps/s	$\textbf{\color{#35bf28}+6.94\%}$
test_mod_add[compile-overhead]	0.1103ms	37.4027μs	26.7360 KOps/s	25.5667 KOps/s	$\color{#35bf28}+4.57\%$
test_mod_wrap[eager]	0.3999ms	0.2059ms	4.8574 KOps/s	4.7879 KOps/s	$\color{#35bf28}+1.45\%$
test_mod_wrap[compile]	0.4364ms	0.2286ms	4.3743 KOps/s	4.2372 KOps/s	$\color{#35bf28}+3.23\%$
test_mod_wrap[compile-overhead]	0.6142ms	0.2303ms	4.3425 KOps/s	4.3084 KOps/s	$\color{#35bf28}+0.79\%$
test_mod_wrap_and_backward[eager]	11.6447ms	10.7584ms	92.9502 Ops/s	83.0654 Ops/s	$\textbf{\color{#35bf28}+11.90\%}$
test_mod_wrap_and_backward[compile]	12.6682ms	11.2616ms	88.7974 Ops/s	80.9293 Ops/s	$\textbf{\color{#35bf28}+9.72\%}$
test_mod_wrap_and_backward[compile-overhead]	12.8029ms	11.7793ms	84.8949 Ops/s	76.8817 Ops/s	$\textbf{\color{#35bf28}+10.42\%}$
test_seq_add[eager]	0.2220ms	91.2945μs	10.9536 KOps/s	10.8818 KOps/s	$\color{#35bf28}+0.66\%$
test_seq_add[compile]	0.1412ms	62.2768μs	16.0573 KOps/s	15.8304 KOps/s	$\color{#35bf28}+1.43\%$
test_seq_add[compile-overhead]	0.1345ms	60.3700μs	16.5645 KOps/s	16.0084 KOps/s	$\color{#35bf28}+3.47\%$
test_seq_wrap[eager]	0.9736ms	0.3842ms	2.6028 KOps/s	2.6324 KOps/s	$\color{#d91a1a}-1.13\%$
test_seq_wrap[compile]	1.3953ms	0.2676ms	3.7376 KOps/s	3.7310 KOps/s	$\color{#35bf28}+0.18\%$
test_seq_wrap[compile-overhead]	1.3553ms	0.2638ms	3.7912 KOps/s	3.7460 KOps/s	$\color{#35bf28}+1.21\%$
test_func_call_runtime[False-eager]	0.9140ms	0.5100ms	1.9607 KOps/s	1.9093 KOps/s	$\color{#35bf28}+2.70\%$
test_func_call_runtime[False-compile]	1.0596ms	0.5069ms	1.9729 KOps/s	1.9354 KOps/s	$\color{#35bf28}+1.94\%$
test_func_call_runtime[False-compile-overhead]	0.6384ms	0.4962ms	2.0153 KOps/s	1.9451 KOps/s	$\color{#35bf28}+3.61\%$
test_func_call_runtime[True-eager]	1.1820ms	0.7278ms	1.3740 KOps/s	1.3395 KOps/s	$\color{#35bf28}+2.58\%$
test_func_call_runtime[True-compile]	0.8449ms	0.5120ms	1.9533 KOps/s	1.8909 KOps/s	$\color{#35bf28}+3.30\%$
test_func_call_runtime[True-compile-overhead]	0.8746ms	0.5095ms	1.9627 KOps/s	1.9130 KOps/s	$\color{#35bf28}+2.60\%$
test_func_call_cm_runtime[False-eager]	0.8985ms	0.5053ms	1.9789 KOps/s	1.9385 KOps/s	$\color{#35bf28}+2.08\%$
test_func_call_cm_runtime[False-compile]	0.6190ms	0.4984ms	2.0063 KOps/s	1.9294 KOps/s	$\color{#35bf28}+3.99\%$
test_func_call_cm_runtime[False-compile-overhead]	0.6502ms	0.4988ms	2.0049 KOps/s	1.9384 KOps/s	$\color{#35bf28}+3.43\%$
test_func_call_cm_runtime[True-eager]	0.9924ms	0.8487ms	1.1782 KOps/s	1.1234 KOps/s	$\color{#35bf28}+4.88\%$
test_func_call_cm_runtime[True-compile]	1.0836ms	0.7246ms	1.3801 KOps/s	1.3542 KOps/s	$\color{#35bf28}+1.92\%$
test_func_call_cm_runtime[True-compile-overhead]	1.2257ms	0.7350ms	1.3606 KOps/s	1.3427 KOps/s	$\color{#35bf28}+1.33\%$
test_vmap_func_call_cm_runtime[eager]	2.4160ms	1.8641ms	536.4582 Ops/s	519.7422 Ops/s	$\color{#35bf28}+3.22\%$
test_vmap_func_call_cm_runtime[compile]	2.6960ms	1.9084ms	524.0116 Ops/s	505.5672 Ops/s	$\color{#35bf28}+3.65\%$
test_vmap_func_call_cm_runtime[compile-overhead]	2.6632ms	1.9053ms	524.8545 Ops/s	506.0725 Ops/s	$\color{#35bf28}+3.71\%$
test_distributed	0.2607ms	0.1242ms	8.0518 KOps/s	7.7921 KOps/s	$\color{#35bf28}+3.33\%$
test_tdmodule	38.1920μs	16.9065μs	59.1490 KOps/s	53.2644 KOps/s	$\textbf{\color{#35bf28}+11.05\%}$
test_tdmodule_dispatch	54.3320μs	35.3376μs	28.2985 KOps/s	27.4899 KOps/s	$\color{#35bf28}+2.94\%$
test_tdseq	42.1180μs	20.7290μs	48.2415 KOps/s	45.8338 KOps/s	$\textbf{\color{#35bf28}+5.25\%}$
test_tdseq_dispatch	71.5230μs	40.8272μs	24.4935 KOps/s	23.7227 KOps/s	$\color{#35bf28}+3.25\%$
test_instantiation_functorch	1.7908ms	1.5856ms	630.6907 Ops/s	613.4312 Ops/s	$\color{#35bf28}+2.81\%$
test_instantiation_td	1.9742ms	1.1625ms	860.1888 Ops/s	843.1359 Ops/s	$\color{#35bf28}+2.02\%$
test_exec_functorch	0.4194ms	0.1826ms	5.4777 KOps/s	5.4454 KOps/s	$\color{#35bf28}+0.59\%$
test_exec_functional_call	0.2801ms	0.1719ms	5.8173 KOps/s	5.7499 KOps/s	$\color{#35bf28}+1.17\%$
test_exec_td	0.2430ms	0.1670ms	5.9872 KOps/s	5.8888 KOps/s	$\color{#35bf28}+1.67\%$
test_exec_td_decorator	1.0469ms	0.2199ms	4.5471 KOps/s	4.3907 KOps/s	$\color{#35bf28}+3.56\%$
test_vmap_mlp_speed[True-True]	0.9610ms	0.6419ms	1.5580 KOps/s	1.5537 KOps/s	$\color{#35bf28}+0.27\%$
test_vmap_mlp_speed[True-False]	0.8668ms	0.6456ms	1.5490 KOps/s	1.5673 KOps/s	$\color{#d91a1a}-1.17\%$
test_vmap_mlp_speed[False-True]	0.8030ms	0.4988ms	2.0049 KOps/s	2.0044 KOps/s	$\color{#35bf28}+0.03\%$
test_vmap_mlp_speed[False-False]	0.7794ms	0.4966ms	2.0138 KOps/s	2.0322 KOps/s	$\color{#d91a1a}-0.91\%$
test_vmap_mlp_speed_decorator[True-True]	1.4049ms	0.6247ms	1.6008 KOps/s	1.5974 KOps/s	$\color{#35bf28}+0.21\%$
test_vmap_mlp_speed_decorator[True-False]	1.0215ms	0.6189ms	1.6159 KOps/s	1.5905 KOps/s	$\color{#35bf28}+1.60\%$
test_vmap_mlp_speed_decorator[False-True]	0.7300ms	0.5062ms	1.9757 KOps/s	1.8883 KOps/s	$\color{#35bf28}+4.63\%$
test_vmap_mlp_speed_decorator[False-False]	0.8666ms	0.5063ms	1.9750 KOps/s	1.9434 KOps/s	$\color{#35bf28}+1.63\%$
test_to_module_speed[True]	1.5307ms	1.2838ms	778.9516 Ops/s	755.1510 Ops/s	$\color{#35bf28}+3.15\%$
test_to_module_speed[False]	1.3788ms	1.2464ms	802.3412 Ops/s	779.9303 Ops/s	$\color{#35bf28}+2.87\%$
test_tc_init	76.4920μs	44.6406μs	22.4012 KOps/s	21.8162 KOps/s	$\color{#35bf28}+2.68\%$
test_tc_init_nested	0.1589ms	86.5809μs	11.5499 KOps/s	11.1188 KOps/s	$\color{#35bf28}+3.88\%$
test_tc_first_layer_tensor	16.8410μs	1.5769μs	634.1392 KOps/s	634.0496 KOps/s	$\color{#35bf28}+0.01\%$
test_tc_first_layer_nontensor	46.7770μs	4.8887μs	204.5529 KOps/s	208.9749 KOps/s	$\color{#d91a1a}-2.12\%$
test_tc_second_layer_tensor	18.0430μs	2.8746μs	347.8732 KOps/s	342.8935 KOps/s	$\color{#35bf28}+1.45\%$
test_tc_second_layer_nontensor	46.9070μs	6.3099μs	158.4811 KOps/s	159.8643 KOps/s	$\color{#d91a1a}-0.87\%$
test_unbind	0.4668s	13.1897ms	75.8165 Ops/s	77.2346 Ops/s	$\color{#d91a1a}-1.84\%$
test_full_like	7.8778ms	6.8227ms	146.5697 Ops/s	123.1951 Ops/s	$\textbf{\color{#35bf28}+18.97\%}$
test_zeros_like	12.1542ms	6.5106ms	153.5945 Ops/s	343.8066 Ops/s	$\textbf{\color{#d91a1a}-55.33\%}$
test_ones_like	17.3182ms	8.1008ms	123.4452 Ops/s	283.8320 Ops/s	$\textbf{\color{#d91a1a}-56.51\%}$
test_clone	16.0662ms	9.6616ms	103.5024 Ops/s	190.0642 Ops/s	$\textbf{\color{#d91a1a}-45.54\%}$
test_squeeze	70.1110μs	12.8497μs	77.8226 KOps/s	77.2793 KOps/s	$\color{#35bf28}+0.70\%$
test_unsqueeze	0.1780ms	94.6330μs	10.5671 KOps/s	10.6218 KOps/s	$\color{#d91a1a}-0.51\%$
test_split	0.5272ms	0.1945ms	5.1414 KOps/s	4.9864 KOps/s	$\color{#35bf28}+3.11\%$
test_permute	0.5094ms	0.2277ms	4.3917 KOps/s	4.3368 KOps/s	$\color{#35bf28}+1.26\%$
test_stack	27.9671ms	25.0587ms	39.9063 Ops/s	39.7633 Ops/s	$\color{#35bf28}+0.36\%$
test_cat	26.6968ms	24.8799ms	40.1931 Ops/s	40.6181 Ops/s	$\color{#d91a1a}-1.05\%$

[github-actions]

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

Total Benchmarks: 228. Improved: $\large\color{#35bf28}50$. Worsened: $\large\color{#d91a1a}4$.

Expand to view detailed results

Name	Max	Mean	Ops	Ops on Repo HEAD	Change
test_plain_set_nested	0.6514ms	14.9261μs	66.9967 KOps/s	66.4069 KOps/s	$\color{#35bf28}+0.89\%$
test_plain_set_stack_nested	39.6020μs	15.0438μs	66.4727 KOps/s	65.5422 KOps/s	$\color{#35bf28}+1.42\%$
test_plain_set_nested_inplace	51.8030μs	15.9367μs	62.7484 KOps/s	62.2163 KOps/s	$\color{#35bf28}+0.86\%$
test_plain_set_stack_nested_inplace	47.5920μs	15.7533μs	63.4787 KOps/s	62.6470 KOps/s	$\color{#35bf28}+1.33\%$
test_items	26.5310μs	2.8922μs	345.7579 KOps/s	345.4870 KOps/s	$\color{#35bf28}+0.08\%$
test_items_nested	0.3550ms	0.3158ms	3.1668 KOps/s	3.2197 KOps/s	$\color{#d91a1a}-1.64\%$
test_items_nested_locked	0.3633ms	0.3157ms	3.1678 KOps/s	3.1755 KOps/s	$\color{#d91a1a}-0.24\%$
test_items_nested_leaf	85.2940μs	63.5398μs	15.7382 KOps/s	15.8589 KOps/s	$\color{#d91a1a}-0.76\%$
test_items_stack_nested	0.3662ms	0.3192ms	3.1326 KOps/s	3.2007 KOps/s	$\color{#d91a1a}-2.13\%$
test_items_stack_nested_leaf	0.1179ms	64.6095μs	15.4776 KOps/s	15.3885 KOps/s	$\color{#35bf28}+0.58\%$
test_items_stack_nested_locked	0.4911ms	0.3172ms	3.1523 KOps/s	3.1736 KOps/s	$\color{#d91a1a}-0.67\%$
test_keys	41.1320μs	3.4333μs	291.2648 KOps/s	293.8637 KOps/s	$\color{#d91a1a}-0.88\%$
test_keys_nested	81.0450μs	55.2874μs	18.0873 KOps/s	18.0979 KOps/s	$\color{#d91a1a}-0.06\%$
test_keys_nested_locked	1.9791ms	60.4679μs	16.5377 KOps/s	16.6672 KOps/s	$\color{#d91a1a}-0.78\%$
test_keys_nested_leaf	75.4850μs	46.8898μs	21.3266 KOps/s	21.4808 KOps/s	$\color{#d91a1a}-0.72\%$
test_keys_stack_nested	89.3750μs	55.5851μs	17.9904 KOps/s	18.1204 KOps/s	$\color{#d91a1a}-0.72\%$
test_keys_stack_nested_leaf	72.2240μs	47.0151μs	21.2698 KOps/s	21.1171 KOps/s	$\color{#35bf28}+0.72\%$
test_keys_stack_nested_locked	98.5550μs	60.4043μs	16.5551 KOps/s	16.5501 KOps/s	$\color{#35bf28}+0.03\%$
test_values	5.2803μs	0.8063μs	1.2403 MOps/s	1.2466 MOps/s	$\color{#d91a1a}-0.51\%$
test_values_nested	55.1430μs	27.4572μs	36.4203 KOps/s	36.2484 KOps/s	$\color{#35bf28}+0.47\%$
test_values_nested_locked	55.3330μs	29.3581μs	34.0622 KOps/s	34.1040 KOps/s	$\color{#d91a1a}-0.12\%$
test_values_nested_leaf	52.1730μs	24.1460μs	41.4148 KOps/s	40.9809 KOps/s	$\color{#35bf28}+1.06\%$
test_values_stack_nested	70.2340μs	27.9314μs	35.8020 KOps/s	34.6152 KOps/s	$\color{#35bf28}+3.43\%$
test_values_stack_nested_leaf	0.1904ms	24.5474μs	40.7375 KOps/s	39.4354 KOps/s	$\color{#35bf28}+3.30\%$
test_values_stack_nested_locked	0.1368ms	29.8349μs	33.5178 KOps/s	32.4375 KOps/s	$\color{#35bf28}+3.33\%$
test_membership	1.7696μs	0.4702μs	2.1269 MOps/s	2.1291 MOps/s	$\color{#d91a1a}-0.11\%$
test_membership_nested	16.9010μs	1.7436μs	573.5240 KOps/s	572.5567 KOps/s	$\color{#35bf28}+0.17\%$
test_membership_nested_leaf	11.5107μs	1.6966μs	589.4073 KOps/s	581.7905 KOps/s	$\color{#35bf28}+1.31\%$
test_membership_stacked_nested	50.3130μs	1.7742μs	563.6446 KOps/s	562.8179 KOps/s	$\color{#35bf28}+0.15\%$
test_membership_stacked_nested_leaf	17.4910μs	1.7768μs	562.8190 KOps/s	565.5517 KOps/s	$\color{#d91a1a}-0.48\%$
test_membership_nested_last	31.4420μs	2.6454μs	378.0137 KOps/s	386.4512 KOps/s	$\color{#d91a1a}-2.18\%$
test_membership_nested_leaf_last	24.5120μs	2.6409μs	378.6530 KOps/s	387.2543 KOps/s	$\color{#d91a1a}-2.22\%$
test_membership_stacked_nested_last	32.4320μs	2.6013μs	384.4222 KOps/s	311.8175 KOps/s	$\textbf{\color{#35bf28}+23.28\%}$
test_membership_stacked_nested_leaf_last	24.1710μs	2.6199μs	381.6881 KOps/s	313.2047 KOps/s	$\textbf{\color{#35bf28}+21.87\%}$
test_nested_getleaf	37.5420μs	6.1533μs	162.5148 KOps/s	163.7981 KOps/s	$\color{#d91a1a}-0.78\%$
test_nested_get	0.2014ms	5.7306μs	174.5011 KOps/s	173.9776 KOps/s	$\color{#35bf28}+0.30\%$
test_stacked_getleaf	30.9610μs	6.0826μs	164.4022 KOps/s	166.3888 KOps/s	$\color{#d91a1a}-1.19\%$
test_stacked_get	40.2030μs	5.7138μs	175.0152 KOps/s	177.8551 KOps/s	$\color{#d91a1a}-1.60\%$
test_nested_getitemleaf	29.5120μs	6.1969μs	161.3722 KOps/s	162.4935 KOps/s	$\color{#d91a1a}-0.69\%$
test_nested_getitem	30.8920μs	5.7839μs	172.8928 KOps/s	172.3563 KOps/s	$\color{#35bf28}+0.31\%$
test_stacked_getitemleaf	52.0320μs	6.1872μs	161.6230 KOps/s	163.4587 KOps/s	$\color{#d91a1a}-1.12\%$
test_stacked_getitem	32.6720μs	5.7103μs	175.1213 KOps/s	173.5111 KOps/s	$\color{#35bf28}+0.93\%$
test_lock_nested	4.6840ms	0.4207ms	2.3772 KOps/s	2.3286 KOps/s	$\color{#35bf28}+2.09\%$
test_lock_stack_nested	0.4896ms	0.3863ms	2.5888 KOps/s	2.5613 KOps/s	$\color{#35bf28}+1.08\%$
test_unlock_nested	0.7771ms	0.3583ms	2.7907 KOps/s	2.7175 KOps/s	$\color{#35bf28}+2.69\%$
test_unlock_stack_nested	0.4442ms	0.3257ms	3.0704 KOps/s	3.0250 KOps/s	$\color{#35bf28}+1.50\%$
test_flatten_speed	0.2614ms	81.2788μs	12.3033 KOps/s	12.5452 KOps/s	$\color{#d91a1a}-1.93\%$
test_unflatten_speed	0.3220ms	0.2817ms	3.5494 KOps/s	3.5589 KOps/s	$\color{#d91a1a}-0.27\%$
test_common_ops	1.4757ms	1.2784ms	782.2133 Ops/s	715.3211 Ops/s	$\textbf{\color{#35bf28}+9.35\%}$
test_creation	22.2810μs	1.4910μs	670.7071 KOps/s	685.7907 KOps/s	$\color{#d91a1a}-2.20\%$
test_creation_empty	54.3130μs	17.2695μs	57.9055 KOps/s	58.0553 KOps/s	$\color{#d91a1a}-0.26\%$
test_creation_nested_1	54.5430μs	19.0441μs	52.5098 KOps/s	52.1250 KOps/s	$\color{#35bf28}+0.74\%$
test_creation_nested_2	51.5030μs	21.9907μs	45.4738 KOps/s	44.8428 KOps/s	$\color{#35bf28}+1.41\%$
test_clone	60.4230μs	29.1330μs	34.3254 KOps/s	33.6433 KOps/s	$\color{#35bf28}+2.03\%$
test_getitem[int]	1.2589ms	16.3756μs	61.0665 KOps/s	59.7698 KOps/s	$\color{#35bf28}+2.17\%$
test_getitem[slice_int]	0.1689ms	27.8924μs	35.8520 KOps/s	34.8501 KOps/s	$\color{#35bf28}+2.88\%$
test_getitem[range]	0.2202ms	0.1096ms	9.1257 KOps/s	9.0603 KOps/s	$\color{#35bf28}+0.72\%$
test_getitem[tuple]	0.1189ms	23.6985μs	42.1968 KOps/s	38.4168 KOps/s	$\textbf{\color{#35bf28}+9.84\%}$
test_getitem[list]	0.2777ms	97.8718μs	10.2174 KOps/s	9.4981 KOps/s	$\textbf{\color{#35bf28}+7.57\%}$
test_setitem_dim[int]	69.1230μs	44.5219μs	22.4609 KOps/s	20.5426 KOps/s	$\textbf{\color{#35bf28}+9.34\%}$
test_setitem_dim[slice_int]	0.2062ms	67.9904μs	14.7080 KOps/s	14.5897 KOps/s	$\color{#35bf28}+0.81\%$
test_setitem_dim[range]	0.1545ms	0.1268ms	7.8836 KOps/s	7.7972 KOps/s	$\color{#35bf28}+1.11\%$
test_setitem_dim[tuple]	0.1887ms	60.9941μs	16.3950 KOps/s	15.3802 KOps/s	$\textbf{\color{#35bf28}+6.60\%}$
test_setitem	0.1886ms	42.4822μs	23.5393 KOps/s	21.2928 KOps/s	$\textbf{\color{#35bf28}+10.55\%}$
test_set	0.2243ms	43.4876μs	22.9951 KOps/s	21.8438 KOps/s	$\textbf{\color{#35bf28}+5.27\%}$
test_set_shared	0.3759ms	51.3771μs	19.4639 KOps/s	19.0598 KOps/s	$\color{#35bf28}+2.12\%$
test_update	0.2019ms	51.1781μs	19.5396 KOps/s	18.0391 KOps/s	$\textbf{\color{#35bf28}+8.32\%}$
test_update_nested	0.2413ms	58.1344μs	17.2015 KOps/s	16.0899 KOps/s	$\textbf{\color{#35bf28}+6.91\%}$
test_update__nested	0.2085ms	58.3793μs	17.1293 KOps/s	15.2957 KOps/s	$\textbf{\color{#35bf28}+11.99\%}$
test_set_nested	0.1889ms	44.3097μs	22.5684 KOps/s	20.7481 KOps/s	$\textbf{\color{#35bf28}+8.77\%}$
test_set_nested_new	0.1969ms	47.7227μs	20.9544 KOps/s	19.3306 KOps/s	$\textbf{\color{#35bf28}+8.40\%}$
test_select	0.5409ms	61.8753μs	16.1615 KOps/s	15.1368 KOps/s	$\textbf{\color{#35bf28}+6.77\%}$
test_select_nested	70.8340μs	42.3386μs	23.6191 KOps/s	23.3867 KOps/s	$\color{#35bf28}+0.99\%$
test_exclude_nested	0.1708ms	58.5018μs	17.0935 KOps/s	16.6223 KOps/s	$\color{#35bf28}+2.83\%$
test_empty[True]	0.2715ms	0.2431ms	4.1131 KOps/s	4.1205 KOps/s	$\color{#d91a1a}-0.18\%$
test_empty[False]	3.5252μs	0.7446μs	1.3429 MOps/s	1.3614 MOps/s	$\color{#d91a1a}-1.35\%$
test_to	73.3540μs	25.2846μs	39.5498 KOps/s	34.2046 KOps/s	$\textbf{\color{#35bf28}+15.63\%}$
test_to_nonblocking	54.2130μs	24.5241μs	40.7761 KOps/s	34.9949 KOps/s	$\textbf{\color{#35bf28}+16.52\%}$
test_unbind_speed	0.3551ms	0.2834ms	3.5281 KOps/s	3.4417 KOps/s	$\color{#35bf28}+2.51\%$
test_unbind_speed_stack0	0.4064ms	0.2826ms	3.5386 KOps/s	3.4559 KOps/s	$\color{#35bf28}+2.39\%$
test_unbind_speed_stack1	93.5340ms	0.7179ms	1.3930 KOps/s	1.3878 KOps/s	$\color{#35bf28}+0.37\%$
test_split	95.4369ms	2.2020ms	454.1411 Ops/s	456.1561 Ops/s	$\color{#d91a1a}-0.44\%$
test_chunk	95.4070ms	2.2101ms	452.4593 Ops/s	453.0936 Ops/s	$\color{#d91a1a}-0.14\%$
test_creation[device0]	0.3415ms	0.1262ms	7.9219 KOps/s	7.7358 KOps/s	$\color{#35bf28}+2.41\%$
test_creation_from_tensor	0.3751ms	0.1280ms	7.8145 KOps/s	7.4055 KOps/s	$\textbf{\color{#35bf28}+5.52\%}$
test_add_one[memmap_tensor0]	0.1314ms	8.9055μs	112.2906 KOps/s	107.1517 KOps/s	$\color{#35bf28}+4.80\%$
test_contiguous[memmap_tensor0]	38.6720μs	2.2045μs	453.6206 KOps/s	434.0575 KOps/s	$\color{#35bf28}+4.51\%$
test_stack[memmap_tensor0]	35.1420μs	6.6089μs	151.3111 KOps/s	136.7417 KOps/s	$\textbf{\color{#35bf28}+10.65\%}$
test_memmaptd_index	1.0846ms	0.4308ms	2.3214 KOps/s	2.2738 KOps/s	$\color{#35bf28}+2.09\%$
test_memmaptd_index_astensor	0.7358ms	0.4877ms	2.0504 KOps/s	2.0006 KOps/s	$\color{#35bf28}+2.49\%$
test_memmaptd_index_op	1.4485ms	1.0602ms	943.2355 Ops/s	903.1674 Ops/s	$\color{#35bf28}+4.44\%$
test_serialize_model	0.1289s	0.1285s	7.7795 Ops/s	7.7773 Ops/s	$\color{#35bf28}+0.03\%$
test_serialize_model_pickle	1.3709s	1.2169s	0.8218 Ops/s	0.8244 Ops/s	$\color{#d91a1a}-0.31\%$
test_serialize_weights	0.2174s	0.1415s	7.0683 Ops/s	7.7707 Ops/s	$\textbf{\color{#d91a1a}-9.04\%}$
test_serialize_weights_returnearly	0.2469s	56.4852ms	17.7037 Ops/s	15.7402 Ops/s	$\textbf{\color{#35bf28}+12.47\%}$
test_serialize_weights_pickle	1.3646s	1.2153s	0.8228 Ops/s	0.8219 Ops/s	$\color{#35bf28}+0.11\%$
test_reshape_pytree	0.1223ms	36.3498μs	27.5105 KOps/s	26.1450 KOps/s	$\textbf{\color{#35bf28}+5.22\%}$
test_reshape_td	0.1697ms	42.4095μs	23.5796 KOps/s	22.3848 KOps/s	$\textbf{\color{#35bf28}+5.34\%}$
test_view_pytree	64.3130μs	36.0009μs	27.7771 KOps/s	26.5827 KOps/s	$\color{#35bf28}+4.49\%$
test_view_td	0.1306ms	47.4525μs	21.0737 KOps/s	20.1497 KOps/s	$\color{#35bf28}+4.59\%$
test_unbind_pytree	0.1396ms	35.1143μs	28.4784 KOps/s	27.6548 KOps/s	$\color{#35bf28}+2.98\%$
test_unbind_td	0.4433ms	43.5653μs	22.9540 KOps/s	22.0131 KOps/s	$\color{#35bf28}+4.27\%$
test_split_pytree	0.2231ms	46.7419μs	21.3941 KOps/s	21.2976 KOps/s	$\color{#35bf28}+0.45\%$
test_split_td	0.6813ms	56.2747μs	17.7700 KOps/s	17.5443 KOps/s	$\color{#35bf28}+1.29\%$
test_add_pytree	0.2056ms	56.5120μs	17.6954 KOps/s	17.0978 KOps/s	$\color{#35bf28}+3.49\%$
test_add_td	0.2396ms	91.6473μs	10.9114 KOps/s	10.1651 KOps/s	$\textbf{\color{#35bf28}+7.34\%}$
test_compile_add_one_nested[tensordict-compile]	0.4101ms	0.2090ms	4.7848 KOps/s	4.5530 KOps/s	$\textbf{\color{#35bf28}+5.09\%}$
test_compile_add_one_nested[tensordict-eager]	0.2956ms	0.1566ms	6.3848 KOps/s	6.3454 KOps/s	$\color{#35bf28}+0.62\%$
test_compile_add_one_nested[pytree-compile]	0.2859ms	0.1458ms	6.8572 KOps/s	6.6952 KOps/s	$\color{#35bf28}+2.42\%$
test_compile_add_one_nested[pytree-eager]	0.3608ms	0.1854ms	5.3923 KOps/s	5.4190 KOps/s	$\color{#d91a1a}-0.49\%$
test_compile_copy_nested[tensordict-compile]	96.2650μs	21.4911μs	46.5309 KOps/s	48.0107 KOps/s	$\color{#d91a1a}-3.08\%$
test_compile_copy_nested[tensordict-eager]	91.5750μs	44.0388μs	22.7073 KOps/s	21.8589 KOps/s	$\color{#35bf28}+3.88\%$
test_compile_copy_nested[pytree-compile]	0.2179ms	62.4114μs	16.0227 KOps/s	15.6158 KOps/s	$\color{#35bf28}+2.61\%$
test_compile_copy_nested[pytree-eager]	81.4540μs	49.4013μs	20.2424 KOps/s	20.2417 KOps/s	$+0.00\%$
test_compile_add_one_flat[tensordict-compile]	0.3718ms	0.3199ms	3.1264 KOps/s	2.9682 KOps/s	$\textbf{\color{#35bf28}+5.33\%}$
test_compile_add_one_flat[tensordict-eager]	0.3460ms	0.2072ms	4.8268 KOps/s	4.7215 KOps/s	$\color{#35bf28}+2.23\%$
test_compile_add_one_flat[tensorclass-compile]	0.2393ms	0.1276ms	7.8378 KOps/s	7.4036 KOps/s	$\textbf{\color{#35bf28}+5.87\%}$
test_compile_add_one_flat[tensorclass-eager]	0.1230ms	60.6710μs	16.4823 KOps/s	15.4431 KOps/s	$\textbf{\color{#35bf28}+6.73\%}$
test_compile_add_one_flat[pytree-compile]	0.4645ms	0.3178ms	3.1464 KOps/s	2.8835 KOps/s	$\textbf{\color{#35bf28}+9.12\%}$
test_compile_add_one_flat[pytree-eager]	0.7824ms	0.6239ms	1.6028 KOps/s	1.5748 KOps/s	$\color{#35bf28}+1.78\%$
test_compile_add_self_flat[tensordict-eager]	0.3529ms	0.2480ms	4.0322 KOps/s	3.9637 KOps/s	$\color{#35bf28}+1.73\%$
test_compile_add_self_flat[tensordict-compile]	0.3755ms	0.3205ms	3.1197 KOps/s	2.9339 KOps/s	$\textbf{\color{#35bf28}+6.33\%}$
test_compile_add_self_flat[tensorclass-eager]	0.1528ms	71.0347μs	14.0776 KOps/s	13.2373 KOps/s	$\textbf{\color{#35bf28}+6.35\%}$
test_compile_add_self_flat[tensorclass-compile]	0.2551ms	0.1287ms	7.7674 KOps/s	7.3379 KOps/s	$\textbf{\color{#35bf28}+5.85\%}$
test_compile_add_self_flat[pytree-eager]	0.6917ms	0.5251ms	1.9042 KOps/s	1.8293 KOps/s	$\color{#35bf28}+4.09\%$
test_compile_add_self_flat[pytree-compile]	0.4046ms	0.3185ms	3.1394 KOps/s	2.9715 KOps/s	$\textbf{\color{#35bf28}+5.65\%}$
test_compile_copy_flat[tensordict-compile]	0.1098ms	18.4235μs	54.2786 KOps/s	53.3083 KOps/s	$\color{#35bf28}+1.82\%$
test_compile_copy_flat[tensordict-eager]	65.5630μs	27.3091μs	36.6178 KOps/s	37.1136 KOps/s	$\color{#d91a1a}-1.34\%$
test_compile_copy_flat[pytree-compile]	0.1127ms	69.8641μs	14.3135 KOps/s	14.3410 KOps/s	$\color{#d91a1a}-0.19\%$
test_compile_copy_flat[pytree-eager]	86.9750μs	51.5708μs	19.3908 KOps/s	19.4543 KOps/s	$\color{#d91a1a}-0.33\%$
test_compile_assign_and_add[tensordict-compile]	2.3247ms	0.8121ms	1.2313 KOps/s	1.0941 KOps/s	$\textbf{\color{#35bf28}+12.55\%}$
test_compile_assign_and_add[tensordict-eager]	3.5099ms	3.1492ms	317.5377 Ops/s	303.9940 Ops/s	$\color{#35bf28}+4.46\%$
test_compile_assign_and_add[pytree-compile]	2.2893ms	0.8028ms	1.2456 KOps/s	1.1045 KOps/s	$\textbf{\color{#35bf28}+12.78\%}$
test_compile_assign_and_add[pytree-eager]	3.3448ms	3.1310ms	319.3817 Ops/s	294.5170 Ops/s	$\textbf{\color{#35bf28}+8.44\%}$
test_compile_indexing[tensor-tensordict-compile]	0.2590ms	0.1088ms	9.1892 KOps/s	8.8437 KOps/s	$\color{#35bf28}+3.91\%$
test_compile_indexing[tensor-tensordict-eager]	0.2027ms	60.1744μs	16.6184 KOps/s	14.7427 KOps/s	$\textbf{\color{#35bf28}+12.72\%}$
test_compile_indexing[tensor-tensorclass-compile]	0.2661ms	0.1037ms	9.6452 KOps/s	9.4257 KOps/s	$\color{#35bf28}+2.33\%$
test_compile_indexing[tensor-tensorclass-eager]	0.2105ms	45.2470μs	22.1009 KOps/s	22.3963 KOps/s	$\color{#d91a1a}-1.32\%$
test_compile_indexing[tensor-pytree-compile]	0.2947ms	0.1068ms	9.3657 KOps/s	9.3817 KOps/s	$\color{#d91a1a}-0.17\%$
test_compile_indexing[tensor-pytree-eager]	0.2174ms	45.2771μs	22.0862 KOps/s	22.3843 KOps/s	$\color{#d91a1a}-1.33\%$
test_compile_indexing[slice-tensordict-compile]	0.2867ms	0.1394ms	7.1724 KOps/s	6.7591 KOps/s	$\textbf{\color{#35bf28}+6.12\%}$
test_compile_indexing[slice-tensordict-eager]	0.2713ms	25.8138μs	38.7389 KOps/s	37.2753 KOps/s	$\color{#35bf28}+3.93\%$
test_compile_indexing[slice-tensorclass-compile]	0.2818ms	0.1326ms	7.5398 KOps/s	7.3647 KOps/s	$\color{#35bf28}+2.38\%$
test_compile_indexing[slice-tensorclass-eager]	70.7940μs	21.1782μs	47.2183 KOps/s	46.5154 KOps/s	$\color{#35bf28}+1.51\%$
test_compile_indexing[slice-pytree-compile]	0.2983ms	0.1376ms	7.2692 KOps/s	7.3426 KOps/s	$\color{#d91a1a}-1.00\%$
test_compile_indexing[slice-pytree-eager]	0.1422ms	21.1067μs	47.3783 KOps/s	47.2710 KOps/s	$\color{#35bf28}+0.23\%$
test_compile_indexing[int-tensordict-compile]	0.2778ms	0.1439ms	6.9509 KOps/s	6.8242 KOps/s	$\color{#35bf28}+1.86\%$
test_compile_indexing[int-tensordict-eager]	0.5163ms	25.7435μs	38.8447 KOps/s	38.7692 KOps/s	$\color{#35bf28}+0.19\%$
test_compile_indexing[int-tensorclass-compile]	0.2755ms	0.1358ms	7.3627 KOps/s	7.0545 KOps/s	$\color{#35bf28}+4.37\%$
test_compile_indexing[int-tensorclass-eager]	60.1640μs	22.4886μs	44.4669 KOps/s	47.7510 KOps/s	$\textbf{\color{#d91a1a}-6.88\%}$
test_compile_indexing[int-pytree-compile]	0.2970ms	0.1356ms	7.3771 KOps/s	7.0429 KOps/s	$\color{#35bf28}+4.74\%$
test_compile_indexing[int-pytree-eager]	0.3866ms	22.3070μs	44.8290 KOps/s	46.8846 KOps/s	$\color{#d91a1a}-4.38\%$
test_mod_add[eager]	0.1962ms	34.3479μs	29.1139 KOps/s	28.4694 KOps/s	$\color{#35bf28}+2.26\%$
test_mod_add[compile]	0.2186ms	73.3171μs	13.6394 KOps/s	13.3429 KOps/s	$\color{#35bf28}+2.22\%$
test_mod_add[compile-overhead]	0.2607ms	0.1359ms	7.3610 KOps/s	6.9809 KOps/s	$\textbf{\color{#35bf28}+5.44\%}$
test_mod_wrap[eager]	0.3864ms	0.2441ms	4.0974 KOps/s	3.7931 KOps/s	$\textbf{\color{#35bf28}+8.02\%}$
test_mod_wrap[compile]	0.4148ms	0.3112ms	3.2132 KOps/s	3.1062 KOps/s	$\color{#35bf28}+3.44\%$
test_mod_wrap[compile-overhead]	7.6371ms	4.0700ms	245.7001 Ops/s	255.1650 Ops/s	$\color{#d91a1a}-3.71\%$
test_mod_wrap_and_backward[eager]	1.5508ms	1.3449ms	743.5661 Ops/s	689.5112 Ops/s	$\textbf{\color{#35bf28}+7.84\%}$
test_mod_wrap_and_backward[compile]	2.4506ms	1.3253ms	754.5177 Ops/s	686.7726 Ops/s	$\textbf{\color{#35bf28}+9.86\%}$
test_mod_wrap_and_backward[compile-overhead]	1.3216ms	0.9064ms	1.1033 KOps/s	971.3738 Ops/s	$\textbf{\color{#35bf28}+13.58\%}$
test_seq_add[eager]	0.2591ms	0.1041ms	9.6096 KOps/s	9.4666 KOps/s	$\color{#35bf28}+1.51\%$
test_seq_add[compile]	0.5786ms	81.0229μs	12.3422 KOps/s	11.6530 KOps/s	$\textbf{\color{#35bf28}+5.91\%}$
test_seq_add[compile-overhead]	0.2559ms	0.1150ms	8.6960 KOps/s	8.3618 KOps/s	$\color{#35bf28}+4.00\%$
test_seq_wrap[eager]	0.5628ms	0.4045ms	2.4721 KOps/s	2.4371 KOps/s	$\color{#35bf28}+1.44\%$
test_seq_wrap[compile]	0.4674ms	0.3165ms	3.1593 KOps/s	2.9739 KOps/s	$\textbf{\color{#35bf28}+6.23\%}$
test_seq_wrap[compile-overhead]	0.3138ms	0.2257ms	4.4307 KOps/s	4.2634 KOps/s	$\color{#35bf28}+3.92\%$
test_func_call_runtime[False-eager]	0.9792ms	0.7851ms	1.2738 KOps/s	1.2522 KOps/s	$\color{#35bf28}+1.73\%$
test_func_call_runtime[False-compile]	0.9952ms	0.8381ms	1.1931 KOps/s	1.2254 KOps/s	$\color{#d91a1a}-2.63\%$
test_func_call_runtime[False-compile-overhead]	0.4943ms	0.3602ms	2.7759 KOps/s	2.6744 KOps/s	$\color{#35bf28}+3.80\%$
test_func_call_runtime[True-eager]	1.0529ms	0.8946ms	1.1178 KOps/s	1.0886 KOps/s	$\color{#35bf28}+2.68\%$
test_func_call_runtime[True-compile]	0.9912ms	0.8276ms	1.2084 KOps/s	1.1793 KOps/s	$\color{#35bf28}+2.47\%$
test_func_call_runtime[True-compile-overhead]	0.5413ms	0.3944ms	2.5353 KOps/s	2.4603 KOps/s	$\color{#35bf28}+3.05\%$
test_func_call_cm_runtime[False-eager]	0.8320ms	0.7268ms	1.3759 KOps/s	1.3249 KOps/s	$\color{#35bf28}+3.85\%$
test_func_call_cm_runtime[False-compile]	0.9537ms	0.8208ms	1.2184 KOps/s	1.2195 KOps/s	$\color{#d91a1a}-0.09\%$
test_func_call_cm_runtime[False-compile-overhead]	0.4854ms	0.3631ms	2.7540 KOps/s	2.6586 KOps/s	$\color{#35bf28}+3.59\%$
test_func_call_cm_runtime[True-eager]	1.1361ms	0.9874ms	1.0128 KOps/s	974.6907 Ops/s	$\color{#35bf28}+3.91\%$
test_func_call_cm_runtime[True-compile]	1.0106ms	0.8526ms	1.1729 KOps/s	1.1243 KOps/s	$\color{#35bf28}+4.32\%$
test_func_call_cm_runtime[True-compile-overhead]	0.5669ms	0.4198ms	2.3820 KOps/s	2.2912 KOps/s	$\color{#35bf28}+3.96\%$
test_vmap_func_call_cm_runtime[eager]	2.5694ms	2.0724ms	482.5296 Ops/s	478.0238 Ops/s	$\color{#35bf28}+0.94\%$
test_vmap_func_call_cm_runtime[compile]	1.0803ms	0.8977ms	1.1140 KOps/s	1.1174 KOps/s	$\color{#d91a1a}-0.31\%$
test_vmap_func_call_cm_runtime[compile-overhead]	0.5252ms	0.4230ms	2.3640 KOps/s	2.2650 KOps/s	$\color{#35bf28}+4.37\%$
test_distributed	3.1119ms	0.2189ms	4.5688 KOps/s	8.4800 KOps/s	$\textbf{\color{#d91a1a}-46.12\%}$
test_tdmodule	45.9230μs	15.1584μs	65.9702 KOps/s	66.6670 KOps/s	$\color{#d91a1a}-1.05\%$
test_tdmodule_dispatch	51.3720μs	30.3527μs	32.9460 KOps/s	32.3328 KOps/s	$\color{#35bf28}+1.90\%$
test_tdseq	35.5420μs	15.9980μs	62.5077 KOps/s	63.8824 KOps/s	$\color{#d91a1a}-2.15\%$
test_tdseq_dispatch	63.5530μs	33.1984μs	30.1220 KOps/s	29.8781 KOps/s	$\color{#35bf28}+0.82\%$
test_instantiation_functorch	2.0231ms	1.8516ms	540.0629 Ops/s	521.4223 Ops/s	$\color{#35bf28}+3.57\%$
test_instantiation_td	1.7997ms	1.1885ms	841.3810 Ops/s	816.5350 Ops/s	$\color{#35bf28}+3.04\%$
test_exec_functorch	0.2901ms	0.2094ms	4.7752 KOps/s	4.6653 KOps/s	$\color{#35bf28}+2.36\%$
test_exec_functional_call	0.3567ms	0.2086ms	4.7939 KOps/s	4.6622 KOps/s	$\color{#35bf28}+2.82\%$
test_exec_td	0.3587ms	0.2137ms	4.6804 KOps/s	4.5080 KOps/s	$\color{#35bf28}+3.82\%$
test_exec_td_decorator	0.5882ms	0.2545ms	3.9288 KOps/s	3.6734 KOps/s	$\textbf{\color{#35bf28}+6.95\%}$
test_vmap_mlp_speed[True-True]	0.8233ms	0.6853ms	1.4591 KOps/s	1.3949 KOps/s	$\color{#35bf28}+4.60\%$
test_vmap_mlp_speed[True-False]	0.8225ms	0.6831ms	1.4639 KOps/s	1.3809 KOps/s	$\textbf{\color{#35bf28}+6.01\%}$
test_vmap_mlp_speed[False-True]	0.9909ms	0.5727ms	1.7462 KOps/s	1.6439 KOps/s	$\textbf{\color{#35bf28}+6.23\%}$
test_vmap_mlp_speed[False-False]	0.7303ms	0.5746ms	1.7405 KOps/s	1.6340 KOps/s	$\textbf{\color{#35bf28}+6.52\%}$
test_vmap_mlp_speed_decorator[True-True]	0.8747ms	0.6971ms	1.4345 KOps/s	1.4164 KOps/s	$\color{#35bf28}+1.28\%$
test_vmap_mlp_speed_decorator[True-False]	1.0107ms	0.7009ms	1.4266 KOps/s	1.4089 KOps/s	$\color{#35bf28}+1.26\%$
test_vmap_mlp_speed_decorator[False-True]	0.7382ms	0.5863ms	1.7056 KOps/s	1.6062 KOps/s	$\textbf{\color{#35bf28}+6.19\%}$
test_vmap_mlp_speed_decorator[False-False]	0.8015ms	0.6134ms	1.6302 KOps/s	1.6065 KOps/s	$\color{#35bf28}+1.48\%$
test_vmap_transformer_speed[True-True]	8.5432ms	8.3447ms	119.8372 Ops/s	117.4525 Ops/s	$\color{#35bf28}+2.03\%$
test_vmap_transformer_speed[True-False]	8.5293ms	8.3130ms	120.2939 Ops/s	117.6782 Ops/s	$\color{#35bf28}+2.22\%$
test_vmap_transformer_speed[False-True]	8.4607ms	8.1389ms	122.8667 Ops/s	120.2123 Ops/s	$\color{#35bf28}+2.21\%$
test_vmap_transformer_speed[False-False]	8.5834ms	8.1946ms	122.0311 Ops/s	120.6638 Ops/s	$\color{#35bf28}+1.13\%$
test_vmap_transformer_speed_decorator[True-True]	20.3675ms	19.5528ms	51.1434 Ops/s	50.3290 Ops/s	$\color{#35bf28}+1.62\%$
test_vmap_transformer_speed_decorator[True-False]	20.2711ms	19.6276ms	50.9488 Ops/s	50.7365 Ops/s	$\color{#35bf28}+0.42\%$
test_vmap_transformer_speed_decorator[False-True]	20.2029ms	19.5081ms	51.2607 Ops/s	51.2882 Ops/s	$\color{#d91a1a}-0.05\%$
test_vmap_transformer_speed_decorator[False-False]	20.1613ms	19.4185ms	51.4973 Ops/s	51.0228 Ops/s	$\color{#35bf28}+0.93\%$
test_to_module_speed[True]	1.5128ms	0.9251ms	1.0809 KOps/s	1.0623 KOps/s	$\color{#35bf28}+1.75\%$
test_to_module_speed[False]	1.3270ms	0.8968ms	1.1151 KOps/s	1.0804 KOps/s	$\color{#35bf28}+3.21\%$
test_tc_init	71.6240μs	34.7541μs	28.7736 KOps/s	28.5148 KOps/s	$\color{#35bf28}+0.91\%$
test_tc_init_nested	0.1079ms	70.0218μs	14.2813 KOps/s	13.6737 KOps/s	$\color{#35bf28}+4.44\%$
test_tc_first_layer_tensor	5.0817μs	0.6548μs	1.5271 MOps/s	1.4771 MOps/s	$\color{#35bf28}+3.39\%$
test_tc_first_layer_nontensor	27.4410μs	2.2216μs	450.1296 KOps/s	451.4787 KOps/s	$\color{#d91a1a}-0.30\%$
test_tc_second_layer_tensor	9.5403μs	1.4135μs	707.4777 KOps/s	726.4368 KOps/s	$\color{#d91a1a}-2.61\%$
test_tc_second_layer_nontensor	82.0840μs	2.9802μs	335.5438 KOps/s	339.1467 KOps/s	$\color{#d91a1a}-1.06\%$
test_unbind	0.1962s	11.9545ms	83.6507 Ops/s	92.4827 Ops/s	$\textbf{\color{#d91a1a}-9.55\%}$
test_full_like	0.7522ms	0.5752ms	1.7386 KOps/s	1.7306 KOps/s	$\color{#35bf28}+0.46\%$
test_zeros_like	0.3535ms	0.1981ms	5.0468 KOps/s	5.0517 KOps/s	$\color{#d91a1a}-0.10\%$
test_ones_like	0.3570ms	0.1981ms	5.0475 KOps/s	5.0547 KOps/s	$\color{#d91a1a}-0.14\%$
test_clone	0.5659ms	0.4140ms	2.4156 KOps/s	2.4102 KOps/s	$\color{#35bf28}+0.22\%$
test_squeeze	0.1542ms	10.8458μs	92.2014 KOps/s	96.0070 KOps/s	$\color{#d91a1a}-3.96\%$
test_unsqueeze	0.2372ms	78.1162μs	12.8014 KOps/s	13.0979 KOps/s	$\color{#d91a1a}-2.26\%$
test_split	0.3846ms	0.1589ms	6.2951 KOps/s	6.0507 KOps/s	$\color{#35bf28}+4.04\%$
test_permute	0.2407ms	0.1809ms	5.5280 KOps/s	5.2144 KOps/s	$\textbf{\color{#35bf28}+6.01\%}$
test_stack	1.3995ms	0.8780ms	1.1389 KOps/s	1.1679 KOps/s	$\color{#d91a1a}-2.49\%$
test_cat	1.2587ms	1.2318ms	811.8374 Ops/s	811.7973 Ops/s	$+0.00\%$