chore: PR 6782 w/ PR 6783

[vezenovm]

Description

Problem*

Resolves

Summary*

Pushed to see the bench report of #6782 with the #6783 optimization.

Additional Context

Documentation*

Check one:

• No documentation needed.
• Documentation included in this PR.
• [For Experimental Features] Documentation to be submitted in a separate PR.

PR Checklist*

• I have tested the changes locally.
• I have formatted the changes with Prettier and/or cargo fmt on default settings.

[github-actions]

Changes to Brillig bytecode sizes

Generated at commit: 8e089924e22b61ca4dce0ac0029d284e8eadfcb2, compared to commit: 919149d3413be5232b33611094687fdb5fd86086

🧾 Summary (10% most significant diffs)

Program	Brillig opcodes (+/-)	%
sha256_var_size_regression	-178 ✅	-9.35%
slices	-228 ✅	-9.44%
reference_counts	-63 ✅	-11.75%
nested_array_dynamic	-330 ✅	-11.84%
nested_array_in_slice	-213 ✅	-14.74%
slice_regex	-396 ✅	-15.06%
wildcard_type	-57 ✅	-16.06%
nested_array_dynamic_simple	-33 ✅	-24.09%
fold_complex_outputs	-159 ✅	-24.84%
brillig_nested_arrays	-225 ✅	-55.28%

---

Full diff report 👇

Program	Brillig opcodes (+/-)	%
regression_4709	133,737 (-3)	-0.00%
debug_logs	5,148 (-36)	-0.69%
slice_dynamic_index	2,507 (-18)	-0.71%
fold_numeric_generic_poseidon	763 (-6)	-0.78%
no_predicates_numeric_generic_poseidon	763 (-6)	-0.78%
pedersen_hash	373 (-3)	-0.80%
bench_2_to_17	328 (-3)	-0.91%
array_dynamic	319 (-3)	-0.93%
modulus	1,762 (-18)	-1.01%
check_large_field_bits	292 (-3)	-1.02%
array_dedup_regression	259 (-3)	-1.15%
to_be_bytes	212 (-3)	-1.40%
pedersen_commitment	209 (-3)	-1.42%
fold_2_to_17	584 (-9)	-1.52%
regression	945 (-15)	-1.56%
hashmap	21,801 (-350)	-1.58%
brillig_oracle	370 (-6)	-1.60%
poseidon_bn254_hash_width_3	5,425 (-110)	-1.99%
poseidon_bn254_hash	5,425 (-110)	-1.99%
sha2_byte	2,767 (-57)	-2.02%
7_function	579 (-12)	-2.03%
7	144 (-3)	-2.04%
blake3	144 (-3)	-2.04%
hash_to_field	139 (-3)	-2.11%
aes128_encrypt	552 (-12)	-2.13%
keccak256	2,197 (-48)	-2.14%
loop_invariant_regression	137 (-3)	-2.14%
u128	2,757 (-63)	-2.23%
6_array	392 (-9)	-2.24%
simple_shield	908 (-21)	-2.26%
conditional_regression_661	127 (-3)	-2.31%
conditional_2	116 (-3)	-2.52%
sha256_regression	6,917 (-180)	-2.54%
array_dynamic_blackbox_input	1,032 (-27)	-2.55%
hint_black_box	336 (-9)	-2.61%
conditional_1	1,188 (-33)	-2.70%
uhashmap	14,004 (-411)	-2.85%
strings	917 (-27)	-2.86%
poseidonsponge_x5_254	4,251 (-126)	-2.88%
if_else_chain	101 (-3)	-2.88%
brillig_cow_regression	2,167 (-66)	-2.96%
ram_blowup_regression	969 (-30)	-3.00%
ecdsa_secp256k1	908 (-30)	-3.20%
sha256_var_padding_regression	5,129 (-171)	-3.23%
nested_dyn_array_regression_5782	172 (-6)	-3.37%
regression_5252	4,594 (-162)	-3.41%
conditional_regression_421	83 (-3)	-3.49%
pedersen_check	565 (-21)	-3.58%
brillig_pedersen	565 (-21)	-3.58%
regression_4449	751 (-30)	-3.84%
brillig_cow	372 (-15)	-3.88%
simple_radix	74 (-3)	-3.90%
sha256_var_witness_const_regression	1,308 (-54)	-3.96%
array_dynamic_nested_blackbox_input	901 (-39)	-4.15%
higher_order_functions	677 (-30)	-4.24%
slice_loop	269 (-12)	-4.27%
sha256_brillig_performance_regression	1,662 (-75)	-4.32%
bigint	2,195 (-102)	-4.44%
sha256	2,408 (-120)	-4.75%
conditional_regression_short_circuit	1,264 (-63)	-4.75%
array_to_slice	718 (-36)	-4.77%
6	1,191 (-60)	-4.80%
global_consts	227 (-12)	-5.02%
brillig_uninitialized_arrays	51 (-3)	-5.56%
to_bits	201 (-12)	-5.63%
regression_6674_1	247 (-15)	-5.73%
to_bytes_integration	88 (-6)	-6.38%
regression_5045	87 (-6)	-6.45%
side_effects_constrain_array	129 (-9)	-6.52%
brillig_constant_reference_regression	84 (-6)	-6.67%
merkle_insert	765 (-55)	-6.71%
regression_6674_2	250 (-18)	-6.72%
databus_composite_calldata	390 (-30)	-7.14%
struct_inputs	258 (-21)	-7.53%
simple_2d_array	106 (-9)	-7.83%
brillig_rc_regression_6123	170 (-15)	-8.11%
to_bytes_consistent	135 (-12)	-8.16%
regression_6674_3	555 (-54)	-8.87%
to_le_bytes	122 (-12)	-8.96%
regression_struct_array_conditional	576 (-57)	-9.00%
sha256_var_size_regression	1,726 (-178)	-9.35%
slices	2,186 (-228)	-9.44%
reference_counts	473 (-63)	-11.75%
nested_array_dynamic	2,457 (-330)	-11.84%
nested_array_in_slice	1,232 (-213)	-14.74%
slice_regex	2,233 (-396)	-15.06%
wildcard_type	298 (-57)	-16.06%
nested_array_dynamic_simple	104 (-33)	-24.09%
fold_complex_outputs	481 (-159)	-24.84%
brillig_nested_arrays	182 (-225)	-55.28%

[github-actions]

Changes to number of Brillig opcodes executed

Generated at commit: 8e089924e22b61ca4dce0ac0029d284e8eadfcb2, compared to commit: 919149d3413be5232b33611094687fdb5fd86086

🧾 Summary (10% most significant diffs)

Program	Brillig opcodes (+/-)	%
side_effects_constrain_array	-9 ✅	-9.47%
slice_regex	-387 ✅	-10.02%
nested_array_in_slice	-201 ✅	-11.39%
wildcard_type	-75 ✅	-12.65%
reference_counts	-60 ✅	-12.96%
regression_4449	-38,288 ✅	-16.01%
fold_complex_outputs	-156 ✅	-16.88%
nested_array_dynamic_simple	-33 ✅	-26.83%
array_dedup_regression	-431 ✅	-38.90%
brillig_nested_arrays	-225 ✅	-59.21%

---

Full diff report 👇

Program	Brillig opcodes (+/-)	%
sha2_byte	47,309 (+335)	+0.71%
bench_2_to_17	589,841 (-3)	-0.00%
fold_2_to_17	1,093,925 (-9)	-0.00%
modulus	19,172 (-18)	-0.09%
check_large_field_bits	2,740 (-3)	-0.11%
to_be_bytes	2,452 (-3)	-0.12%
brillig_oracle	2,751 (-6)	-0.22%
u128	24,962 (-63)	-0.25%
aes128_encrypt	4,719 (-12)	-0.25%
loop_invariant_regression	1,164 (-3)	-0.26%
7	1,106 (-3)	-0.27%
blake3	1,106 (-3)	-0.27%
to_bytes_integration	2,065 (-6)	-0.29%
hash_to_field	908 (-3)	-0.33%
keccak256	34,777 (-120)	-0.34%
slice_dynamic_index	4,667 (-18)	-0.38%
pedersen_hash	669 (-3)	-0.45%
7_function	2,537 (-12)	-0.47%
to_le_bytes	1,152 (-6)	-0.52%
regression	2,816 (-15)	-0.53%
array_dynamic	498 (-3)	-0.60%
global_consts	1,773 (-12)	-0.67%
poseidonsponge_x5_254	183,750 (-1,260)	-0.68%
debug_logs	5,160 (-36)	-0.69%
regression_5252	915,026 (-9,354)	-1.01%
pedersen_commitment	293 (-3)	-1.01%
hint_black_box	717 (-9)	-1.24%
6_array	1,631 (-21)	-1.27%
brillig_cow	1,140 (-15)	-1.30%
fold_numeric_generic_poseidon	5,143 (-69)	-1.32%
no_predicates_numeric_generic_poseidon	5,143 (-69)	-1.32%
strings	1,770 (-27)	-1.50%
to_bytes_consistent	754 (-12)	-1.57%
uhashmap	145,549 (-2,455)	-1.66%
regression_6674_1	819 (-15)	-1.80%
slice_loop	942 (-18)	-1.88%
to_bits	614 (-12)	-1.92%
sha256_brillig_performance_regression	23,194 (-456)	-1.93%
brillig_cow_regression	519,086 (-10,376)	-1.96%
ram_blowup_regression	778,650 (-15,953)	-2.01%
array_to_slice	1,670 (-36)	-2.11%
brillig_pedersen	972 (-21)	-2.11%
pedersen_check	972 (-21)	-2.11%
regression_6674_2	822 (-18)	-2.14%
sha256_var_padding_regression	222,216 (-4,943)	-2.18%
higher_order_functions	1,255 (-30)	-2.33%
conditional_regression_661	120 (-3)	-2.44%
conditional_2	119 (-3)	-2.46%
poseidon_bn254_hash	162,594 (-4,644)	-2.78%
poseidon_bn254_hash_width_3	162,594 (-4,644)	-2.78%
sha256_regression	118,707 (-3,397)	-2.78%
ecdsa_secp256k1	6,789 (-200)	-2.86%
regression_struct_array_conditional	1,679 (-51)	-2.95%
if_else_chain	98 (-3)	-2.97%
hashmap	54,143 (-1,739)	-3.11%
regression_6674_3	1,615 (-54)	-3.24%
conditional_regression_421	89 (-3)	-3.26%
nested_dyn_array_regression_5782	171 (-6)	-3.39%
conditional_1	5,717 (-203)	-3.43%
sha256_var_witness_const_regression	7,200 (-266)	-3.56%
struct_inputs	566 (-21)	-3.58%
simple_shield	2,871 (-113)	-3.79%
simple_radix	69 (-3)	-4.17%
sha256	14,996 (-672)	-4.29%
array_dynamic_nested_blackbox_input	4,550 (-209)	-4.39%
databus_composite_calldata	617 (-30)	-4.64%
slices	3,365 (-171)	-4.84%
brillig_rc_regression_6123	290 (-15)	-4.92%
conditional_regression_short_circuit	7,528 (-400)	-5.05%
6	7,450 (-400)	-5.10%
sha256_var_size_regression	16,377 (-957)	-5.52%
array_dynamic_blackbox_input	18,209 (-1,141)	-5.90%
brillig_uninitialized_arrays	44 (-3)	-6.38%
simple_2d_array	125 (-9)	-6.72%
brillig_constant_reference_regression	81 (-6)	-6.90%
merkle_insert	3,769 (-284)	-7.01%
nested_array_dynamic	3,464 (-309)	-8.19%
side_effects_constrain_array	86 (-9)	-9.47%
slice_regex	3,474 (-387)	-10.02%
nested_array_in_slice	1,564 (-201)	-11.39%
wildcard_type	518 (-75)	-12.65%
reference_counts	403 (-60)	-12.96%
regression_4449	200,857 (-38,288)	-16.01%
fold_complex_outputs	768 (-156)	-16.88%
nested_array_dynamic_simple	90 (-33)	-26.83%
array_dedup_regression	677 (-431)	-38.90%
brillig_nested_arrays	155 (-225)	-59.21%

[github-actions]

Peak Memory Sample

Program	Peak Memory
keccak256	79.14M
workspace	121.75M
regression_4709	295.99M
ram_blowup_regression	2.44G
private-kernel-tail	210.43M
private-kernel-reset	862.08M
private-kernel-inner	307.78M
parity-root	175.78M

[github-actions]

Compilation Sample

Program	Compilation Time	%
sha256_regression	0m1.447s	-1%
regression_4709	0m0.775s	1%
ram_blowup_regression	0m16.768s	-2%
private-kernel-tail	0m1.433s	16%
private-kernel-reset	0m9.388s	3%
private-kernel-inner	0m2.300s	-15%
parity-root	0m0.892s	-6%
noir-contracts	2m38.430s	-3%

[github-actions]

Changes to circuit sizes

Generated at commit: 8e089924e22b61ca4dce0ac0029d284e8eadfcb2, compared to commit: 919149d3413be5232b33611094687fdb5fd86086

🧾 Summary (10% most significant diffs)

Program	ACIR opcodes (+/-)	%	Circuit size (+/-)	%
regression_struct_array_conditional	-16 ✅	-18.39%	-24 ✅	-0.74%
nested_array_in_slice	-70 ✅	-6.48%	-151 ✅	-2.65%

---

Full diff report 👇

Program	ACIR opcodes (+/-)	%	Circuit size (+/-)	%
array_dynamic_nested_blackbox_input	253 (-12)	-4.53%	7,326 (-35)	-0.48%
regression_struct_array_conditional	71 (-16)	-18.39%	3,203 (-24)	-0.74%
nested_array_in_slice	1,010 (-70)	-6.48%	5,550 (-151)	-2.65%

[vezenovm]

#6782 has been approved so we can look at the bench on #6783 itself now