Remove inlining constraint for large functions when targetting the new code transform. #12731

ekpyron · 2022-03-02T14:53:09Z

Depends on #12937.

libyul/optimiser/FullInliner.cpp

chriseth · 2022-03-02T14:55:18Z

Changelog.md

@@ -5,6 +5,7 @@ Language Features:

 Compiler Features:
 * JSON-AST: Added selector field for errors and events.
+ * Yul Optimizer: Improve inlining heauristics for via IR code generation and pure Yul compilation.


Suggested change

* Yul Optimizer: Improve inlining heauristics for via IR code generation and pure Yul compilation.

* Yul Optimizer: Improve inlining heuristic for via IR code generation and pure Yul compilation.

chriseth · 2022-03-02T14:56:03Z

test/cmdlineTests/name_simplifier/output

-                    mstore(add(add(headStart, length), 96), 0)
-                }
-                tail := add(add(headStart, and(add(length, 31), not(31))), 96)
-            }
            function panic_error_0x32()


Why does it not inline those?

Apparently it has a code size of at least 6 and thereby generally doesn't qualify...

chriseth · 2022-03-02T14:57:34Z

test/libyul/yulOptimizerTests/fullInliner/no_inline_into_big_function.yul

@@ -57,6 +57,24 @@
 //         let a_56 := b_54


Is this test still meaningful?

Nope, but we may want to run it in a setup that'll go through the old code transform...

ekpyron · 2022-03-02T15:04:50Z

On our tests we get this. But more curious about the external test stats. Even more so since they distinguish code size from gas cost better.

Click for a table of gas differences

File name	IR-optimized (%)	Legacy-Optimized (%)	Legacy (%)
constructor_with_params_inheritance.sol	-0.186633	0	0
constructor_with_params.sol	-0.094985	0	0
constructor_with_params_diamond_inheritance.sol	-3.09367	0	0
smoke/constructor.sol	-3.65267	0	0
events/event_indexed_string.sol	-2.36025	0	0
events/event_dynamic_nested_array_storage_v2.sol	-1.11354	0	0
events/event_dynamic_array_storage_v2.sol	-0.965611	0	0
events/event_emit_from_other_contract.sol	-0.724017	0	0
events/event_dynamic_array_storage.sol	-0.965611	0	0
functionCall/failed_create.sol	-5.70945	0	0
functionCall/creation_function_call_with_salt.sol	-2.69084	0	0
functionCall/gas_and_value_basic.sol	-3.17486	0	0
functionCall/external_call_to_nonexisting_debugstrings.sol	-2.43625	0	0
functionCall/external_call_to_nonexisting.sol	-8.02858	0	0
functionCall/mapping_array_internal_argument.sol	-0.173324	0	0
functionCall/creation_function_call_with_args.sol	-2.69303	0	0
functionCall/send_zero_ether.sol	-2.72404	0	0
functionCall/gas_and_value_brace_syntax.sol	-3.17486	0	0
externalContracts/ramanujan_pi.sol	-3.43954	0	0
externalContracts/prbmath_unsigned.sol	-1.26607	0	0
externalContracts/deposit_contract.sol	-2.67888	0	0
externalContracts/prbmath_signed.sol	-0.0573687	0	0
externalContracts/FixedFeeRegistrar.sol	-3.58586	0	0
externalContracts/strings.sol	-8.95921	0	0
constructor/constructor_static_array_argument.sol	-0.208113	0	0
constructor/arrays_in_constructors.sol	-0.974948	0	0
constructor/no_callvalue_check.sol	-0.00243829	0	0
constructor/constructor_arguments_external.sol	0.178213	0	0
constructor/bytes_in_constructors_packer.sol	-2.87035	0	0
constructor/bytes_in_constructors_unpacker.sol	-4.89873	0	0
calldata/copy_from_calldata_removes_bytes_data.sol	-0.00837731	0	0
abiencodedecode/abi_decode_simple_storage.sol	-0.0956459	0	0
inheritance/value_for_constructor.sol	-1.69169	0	0
inheritance/inherited_function_calldata_memory_interface.sol	-2.06227	0	0
storage/packed_storage_structs_bytes.sol	-1.50832	0	0
abiEncoderV2/abi_encode_calldata_slice.sol	0.120142	0	0
abiEncoderV2/storage_array_encoding.sol	-0.0364603	0	0
abiEncoderV2/abi_encode_v2_in_function_inherited_in_v1_contract.sol	1.38454	0	0
abiEncoderV2/calldata_array.sol	-25.4391	0	0
abiEncoderV2/abi_encode_v2.sol	-0.295678	0	0
userDefinedValueType/erc20.sol	-8.14631	0	0
userDefinedValueType/calldata.sol	-0.149494	0	0
libraries/using_library_mappings_return.sol	-0.139453	0	0
libraries/internal_types_in_library.sol	-3.00202	0	0
libraries/using_library_mappings_public.sol	0.152809	0	0
abiEncoderV1/abi_encode_calldata_slice.sol	0.120142	0	0
abiEncoderV1/abi_decode_v2_storage.sol	-0.0727945	0	0
abiEncoderV1/struct/struct_storage_ptr.sol	-0.652464	0	0
viaYul/array_storage_length_access.sol	-0.000246334	0	0
viaYul/array_storage_index_boundary_test.sol	-0.913714	0	0
viaYul/array_memory_index_access.sol	-0.198685	0	0
viaYul/array_storage_index_access.sol	-1.87115	0	0
viaYul/array_storage_push_empty_length_address.sol	-0.000308725	0	0
viaYul/array_storage_push_empty.sol	-10.2648	0	0
viaYul/copy_struct_invalid_ir_bug.sol	-0.0742732	0	0
viaYul/array_storage_index_zeroed_test.sol	-2.12432	0	0
salted_create/salted_create_with_value.sol	-0.023133	0	0
state/blockhash_basic.sol	-3.5543	0	0
array/constant_var_as_array_length.sol	-0.282734	0	0
array/function_array_cross_calls.sol	-1.38798	0	0
array/reusing_memory.sol	-0.179653	0	0
array/byte_array_transitional_2.sol	-0.473622	0	0
array/fixed_arrays_in_constructors.sol	-0.177818	0	0
array/bytes_length_member.sol	-0.0335027	0	0
array/dynamic_multi_array_cleanup.sol	-5.26303	0	0
array/dynamic_array_cleanup.sol	-0.0186579	0	0
array/fixed_arrays_as_return_type.sol	-5.42211	0	0
array/dynamic_arrays_in_storage.sol	-6.67631	0	0
array/arrays_complex_from_and_to_storage.sol	-0.0653075	0	0
array/fixed_array_cleanup.sol	-0.00214802	0	0
array/create_memory_array.sol	-4.07982	0	0
array/delete/bytes_delete_element.sol	-7.94842	0	0
array/push/nested_bytes_push.sol	-0.233309	0	0
array/push/array_push_struct_from_calldata.sol	-0.293371	0	0
array/push/array_push_nested_from_calldata.sol	-1.03384	0	0
array/push/byte_array_push_transition.sol	-1.67488	0	0
array/push/push_no_args_2d.sol	-0.0262091	0	0
array/push/array_push_struct.sol	-0.26663	0	0
array/push/push_no_args_bytes.sol	-1.19988	0	0
array/push/array_push.sol	-0.417726	0	0
array/copying/array_copy_storage_storage_different_base_nested.sol	-0.397417	0	0
array/copying/array_copy_storage_to_memory_nested.sol	-0.65892	0	0
array/copying/array_copy_storage_storage_static_static.sol	-0.0012707	0	0
array/copying/array_copy_different_packing.sol	-1.50914	0	0
array/copying/array_copy_storage_storage_static_dynamic.sol	-0.00811939	0	0
array/copying/array_copy_storage_storage_dynamic_dynamic.sol	-0.261058	0	0
array/copying/array_copy_target_simple_2.sol	-0.28325	0	0
array/copying/array_nested_calldata_to_storage.sol	-0.419844	0	0
array/copying/memory_dyn_2d_bytes_to_storage.sol	-6.08932	0	0
array/copying/array_storage_multi_items_per_slot.sol	-0.275892	0	0
array/copying/array_nested_memory_to_storage.sol	-0.561084	0	0
array/copying/array_copy_including_array.sol	-0.0351168	0	0
array/copying/array_copy_target_simple.sol	-0.387079	0	0
array/copying/copy_function_internal_storage_array.sol	-0.311187	0	0
array/copying/array_of_function_external_storage_to_storage_dynamic.sol	-0.0401265	0	0
array/copying/copy_byte_array_in_struct_to_storage.sol	-0.037608	0	0
array/copying/array_of_function_external_storage_to_storage_dynamic_different_mutability.sol	-0.0402522	0	0
array/copying/array_copy_target_leftover.sol	-3.32283	0	0
array/copying/calldata_array_dynamic_to_storage.sol	-0.11065	0	0
array/copying/array_copy_storage_storage_struct.sol	-0.0762061	0	0
array/copying/bytes_inside_mappings.sol	-0.0218622	0	0
array/copying/array_of_struct_memory_to_storage.sol	-0.608482	0	0
array/copying/copy_byte_array_to_storage.sol	-1.01944	0	0
array/copying/array_copy_calldata_storage.sol	0.0238224	0	0
array/copying/copy_removes_bytes_data.sol	-0.00169119	0	0
array/copying/array_copy_nested_array.sol	-0.199651	0	0
array/copying/array_of_struct_calldata_to_storage.sol	-0.453539	0	0
array/copying/copying_bytes_multiassign.sol	-0.0901274	0	0
array/copying/storage_memory_nested.sol	-0.917776	0	0
array/copying/storage_memory_packed_dyn.sol	-3.07921	0	0
array/copying/array_of_structs_containing_arrays_memory_to_storage.sol	-0.585874	0	0
array/copying/array_copy_storage_storage_dyn_dyn.sol	-0.0251333	0	0
array/copying/storage_memory_nested_bytes.sol	0.0478209	0	0
array/copying/storage_memory_nested_struct.sol	-0.305636	0	0
array/copying/array_copy_storage_storage_different_base.sol	-0.400078	0	0
array/copying/array_copy_clear_storage.sol	-0.480392	0	0
array/copying/array_of_structs_containing_arrays_calldata_to_storage.sol	-2.2852	0	0
array/copying/storage_memory_nested_from_pointer.sol	-0.917776	0	0
array/copying/bytes_storage_to_storage.sol	-12.566	0	0
array/copying/array_copy_cleanup_uint40.sol	-4.14068	0	0
array/pop/array_pop_uint24_transition.sol	-0.6207	0	0
array/pop/byte_array_pop_masking_long.sol	-0.27375	0	0
array/pop/array_pop_uint16_transition.sol	-0.955743	0	0
array/pop/byte_array_pop_copy_long.sol	0.415514	0	0
array/pop/byte_array_pop_long_storage_empty_garbage_ref.sol	-0.0539821	0	0
array/pop/byte_array_pop_long_storage_empty.sol	-7.6756	0	0
array/pop/array_pop_array_transition.sol	-0.192385	0	0
immutable/use_scratch.sol	-3.26079	0	0
immutable/multi_creation.sol	-0.0219949	0	0
structs/struct_copy_via_local.sol	-0.418383	0	0
structs/struct_delete_storage_nested_small.sol	-0.252123	0	0
structs/struct_delete_storage_with_array.sol	-0.381504	0	0
structs/structs.sol	0.0141335	0	0
structs/memory_structs_nested_load.sol	-0.972791	0	0
structs/struct_copy.sol	-0.00699766	0	0
structs/struct_delete_storage_with_arrays_small.sol	-0.248445	0	0
structs/copy_struct_array_from_storage.sol	-1.35973	0	0
structs/struct_containing_bytes_copy_and_delete.sol	-0.0934733	0	0
structs/struct_memory_to_storage_function_ptr.sol	-0.400123	0	0
structs/conversion/recursive_storage_memory.sol	-0.710806	0	0
structs/calldata/calldata_struct_with_nested_array_to_storage.sol	-0.529728	0	0
various/erc20.sol	-7.51651	0	0
various/senders_balance.sol	-3.0777	0	0
various/destructuring_assignment.sol	-1.06028	0	0
various/value_complex.sol	-5.48924	0	0
various/address_code.sol	-9.33142	0	0
various/value_insane.sol	-6.0074	0	0
various/swap_in_storage_overwrite.sol	-0.0173179	0	0
various/skip_dynamic_types_for_structs.sol	-0.0838823	0	0
isoltestTesting/balance_other_contract.sol	-6.03229	0	0

ekpyron · 2022-03-02T16:16:38Z

test/externalTests/elementfi.sh

@@ -89,6 +89,8 @@ function elementfi_test
    # TODO: Remove when https://github.com/element-fi/elf-contracts/issues/243 is fixed.
    sed -i 's|^\s*require(_expiration - block\.timestamp < _unitSeconds);\s*$||g' contracts/ConvergentCurvePool.sol

+    find . -name "*.sol" -exec sed -i -e 's/^\(\s*\)\(assembly\)/\1\/\/\/ @solidity memory-safe-assembly\n\1\2/' '{}' \;


This one actually isn't memory-safe... and apparently I was fooled by our test cases in that just inlining unconditionally will just work and not break anything :-(.

ekpyron · 2022-03-02T17:08:23Z

For the working external tests I get:

--- summarized-benchmarks.json  2022-03-02 18:02:11.482076212 +0100
+++ summarized-benchmarks-more-inlining.json   2022-03-02 18:00:44.111559280 +0100
     "bleeps": {
         "ir-optimize-evm+yul": {
-            "bytecode_size": 132839,
+            "bytecode_size": 129306,
             "deployment_gas": 0,
-            "method_gas": 39289198,
+            "method_gas": 39158584,
             "version": "bb90cd0"
         },
...
    "gnosis": {
         "ir-optimize-evm+yul": {
-            "bytecode_size": 56037,
+            "bytecode_size": 53436,
             "deployment_gas": null,
             "method_gas": null,
...
     "gnosis2": {
         "ir-optimize-evm+yul": {
-            "bytecode_size": 46641,
+            "bytecode_size": 44358,
             "deployment_gas": null,
             "method_gas": null,
...
     "prb-math": {
         "ir-optimize-evm+yul": {
-            "bytecode_size": 38354,
+            "bytecode_size": 37545,
             "deployment_gas": null,
             "method_gas": null,
...
     "yield_liquidator": {
         "ir-optimize-evm+yul": {
-            "bytecode_size": 89120,
-            "deployment_gas": 15449067,
-            "method_gas": 2947797,
+            "bytecode_size": 85129,
+            "deployment_gas": 14831189,
+            "method_gas": 2938537,
             "version": "74f969a"
...
     "zeppelin": {
         "ir-optimize-evm+yul": {
-            "bytecode_size": 546357,
+            "bytecode_size": 520326,
             "deployment_gas": null,
             "method_gas": null,

Unfortunately we don't get runtime gas stats for some yet... but interestingly code size also seems to generally decrease...

But it still sucks a bit that it does still bring back more stack-too-deep cases.

chriseth · 2022-03-02T17:33:23Z

Impressive!

ekpyron · 2022-03-16T17:55:15Z

libyul/optimiser/FullInliner.cpp

 	// Do not inline into already big functions.
-	if (m_functionSizes.at(_callSite) > 45)
+	if (m_functionSizes.at(_callSite) > (usesNewCodeTransform ? 450 : 45))


So this version seems to fix all stack-too-deep errors in the external tests.
But I also have a hunch that #12132 might actually help a lot with stack pressure as well, so maybe on top of that we can even drop any boundary...

cameel · 2022-03-28T16:48:07Z

All the external failures here at this point seem to be due to NomicFoundation/hardhat#2453. I think it would be fine to just disable these tests until the issue is fixed.

leonardoalt · 2022-04-04T10:08:29Z

@ekpyron maybe fix the conflicts here so you can rebase and test the external contracts again?

ekpyron · 2022-04-13T10:41:06Z

libyul/optimiser/FullInliner.cpp

@@ -196,7 +205,7 @@ bool FullInliner::shallInline(FunctionCall const& _funCall, YulString _callSite)
 			break;
 		}

-	return (size < 6 || (constantArg && size < 12));
+	return (size < (usesNewCodeTransform ? 8 : 6) || (constantArg && size < (usesNewCodeTransform ? 16 : 12)));


@cameel Actually: tweaking this is also something to do here.

cameel · 2022-04-13T11:23:17Z

Just for reference, today's discussion with @ekpyron about what is still to be done here. The most important part:

@ekpyron

But generally, disabling tests that fail due to heuristics is really the main thing to do for now I guess...
Or actually: tweaking https://github.com/ethereum/solidity/pull/12731/files#r849338706 also.
But we don't necessarily need to do that in one step.
But that's something where #12132 might help. The thing is that some of our panic helpers don't fall under the current return (size < 6 || (constantArg && size < 12)); threshold. And due to that, the current code transform won't notice that those paths revert and will keep merging stack layouts, etc... while with #12132 we may get away with just not touching https://github.com/ethereum/solidity/pull/12731/files#r849338706 at all in a first step and only tweaking it in a second step...

@cameel

You were doing this tweaking based on gas improvements and whether ext tests fail or did you also have other criteria?

@ekpyron

Yes - resp. we were wondering somewhere else why optimized IR code still had panic_*() function calls, since we'd have expected them to be inlined - so I wanted to see which values would be needed to inline those - not sure if the values currently in the PR do it or not.
The ideal version of the PR would just remove https://github.com/ethereum/solidity/pull/12731/files#r828298176 entirely without any new threshold. (And without changing https://github.com/ethereum/solidity/pull/12731/files#r849338706 at first, even though that would be the next step)
I think the only external test that was failing due to that was elementfi...
I actually also wanted to have a look at that - that looked like it emitted a memoryguard, so it allowed stack-to-memory, but still ran into a stack-too-deep error, which at this point I'd consider a bug case.

Failing compilation with a stack-too-deep error.
(due to the heightened stack pressure due to more inlining - where the hope was that this wouldn't happen anymore these days ;-))

libyul/optimiser/FullInliner.cpp

cameel · 2022-05-20T19:00:21Z

#12937 updated. Now all external tests should be passing here.

cameel · 2022-05-23T12:23:15Z

Updated benchmarks from 1070566.

`ir-optimize-evm+yul`

Relative

project	bytecode_size	deployment_gas	method_gas
bleeps	`-3.25% ✅`	`0%`	`-0%`
brink	`-7.5% ✅`
colony	`0%`
elementfi	`-2.45% ✅`
ens	`0%`	`-6.37% ✅`	`-0.64% ✅`
euler	`-2.52% ✅`	`-1.68% ✅`	`-2.07% ✅`
gnosis	`-3.83% ✅`	`-4.33% ✅`	`-0.02% ✅`
perpetual-pools	`-4.56% ✅`	`-2.66% ✅`	`-0.99% ✅`
pool-together	`-7.06% ✅`	`-5.58% ✅`	`-0.37% ✅`
prb-math	`-3.53% ✅`	`-3.48% ✅`	`0%`
trident	`-9.78% ✅`	`-7.95% ✅`	`-6.25% ✅`
uniswap	`-4.28% ✅`	`-4.5% ✅`	`-1.24% ✅`
yield_liquidator	`-7.12% ✅`	`-6.1% ✅`	`-0.5% ✅`
zeppelin	`-4.75% ✅`	`-3.21% ✅`	`-0.18% ✅`

Absolute

project	bytecode_size	deployment_gas	method_gas
bleeps	-4261	0	-113
brink	-1018
colony	0
elementfi	-14314
ens	0	-1984437	-763921
euler	-4604	-601676	-62018944
gnosis	-2593	-169903	-11399
perpetual-pools	-9474	-1107332	-8007246
pool-together	-15248	-2061198	-234934
prb-math	-1341	-285835	0
trident	-31030	-2178611	-27085001
uniswap	-7385	-905462	-33994790
yield_liquidator	-6306	-931624	-14831
zeppelin	-27540	-3400500	-770162

cameel · 2022-05-23T12:32:50Z

test/libsolidity/semanticTests/userDefinedValueType/erc20.sol

@@ -113,7 +113,7 @@ contract ERC20 {
 // ----
 // constructor()
 // ~ emit Transfer(address,address,uint256): #0x00, #0x1212121212121212121212121212120000000012, 0x14
-// gas irOptimized: 418388
+// gas irOptimized: 364503


Wow, looks like UDVTs are getting a nice boost.

Nice, as it should be!

chriseth · 2022-05-23T13:18:14Z

It's great that the bytecode size also improves for all external tests!

chriseth reviewed Mar 2, 2022

View reviewed changes

libyul/optimiser/FullInliner.cpp Show resolved Hide resolved

chriseth reviewed Mar 2, 2022

View reviewed changes

ekpyron force-pushed the removeInliningConstraint branch from 8b11252 to 8fef645 Compare March 2, 2022 16:02

ekpyron changed the base branch from develop to assemblyAnnotationDialectString March 2, 2022 16:02

ekpyron force-pushed the assemblyAnnotationDialectString branch from caf191b to 46d8611 Compare March 2, 2022 16:07

ekpyron force-pushed the removeInliningConstraint branch from 8fef645 to 7f1f87a Compare March 2, 2022 16:10

ekpyron commented Mar 2, 2022

View reviewed changes

ekpyron force-pushed the removeInliningConstraint branch from 7f1f87a to 2775355 Compare March 2, 2022 16:35

Base automatically changed from assemblyAnnotationDialectString to develop March 7, 2022 12:03

cameel mentioned this pull request Mar 11, 2022

allow specifying a function should be inlined in the signature #12782

Closed

ekpyron force-pushed the removeInliningConstraint branch 2 times, most recently from 409ea3c to 2d52471 Compare March 16, 2022 17:10

ekpyron commented Mar 16, 2022

View reviewed changes

cameel added the optimizer label Mar 18, 2022

ekpyron commented Apr 13, 2022

View reviewed changes

cameel force-pushed the removeInliningConstraint branch 3 times, most recently from 6ea76c7 to 41faa38 Compare April 13, 2022 17:50

cameel mentioned this pull request Apr 13, 2022

Prevent parallel soltest runs from overwriting each other's XML test output #12936

Merged

cameel force-pushed the removeInliningConstraint branch from 41faa38 to 4c350b8 Compare April 13, 2022 20:02

cameel added the has dependencies The PR depends on other PRs that must be merged first label Apr 13, 2022

cameel mentioned this pull request Apr 13, 2022

Disable external test cases in which Hardhat's heuristics break after relaxing inlining constraints #12937

Merged

ekpyron changed the base branch from develop to disable-ext-test-cases-broken-after-relaxing-inlining May 11, 2022 18:06

ekpyron force-pushed the removeInliningConstraint branch from a6d231e to 65695e3 Compare May 11, 2022 18:33

ekpyron commented May 11, 2022

View reviewed changes

libyul/optimiser/FullInliner.cpp Show resolved Hide resolved

ekpyron force-pushed the removeInliningConstraint branch 2 times, most recently from ff71e3d to c592dcb Compare May 11, 2022 19:50

ekpyron force-pushed the disable-ext-test-cases-broken-after-relaxing-inlining branch from ed44ef0 to 826fc47 Compare May 20, 2022 10:42

ekpyron force-pushed the removeInliningConstraint branch from c592dcb to a59c0ce Compare May 20, 2022 10:48

cameel force-pushed the disable-ext-test-cases-broken-after-relaxing-inlining branch from 826fc47 to cc66991 Compare May 20, 2022 18:54

cameel force-pushed the removeInliningConstraint branch from a59c0ce to 7b7409a Compare May 20, 2022 18:54

cameel force-pushed the disable-ext-test-cases-broken-after-relaxing-inlining branch from cc66991 to 52e796c Compare May 20, 2022 18:58

cameel force-pushed the removeInliningConstraint branch from 7b7409a to 2817678 Compare May 20, 2022 18:58

chriseth force-pushed the disable-ext-test-cases-broken-after-relaxing-inlining branch from 52e796c to 46eff4b Compare May 23, 2022 10:30

chriseth force-pushed the removeInliningConstraint branch 2 times, most recently from 57e12cb to 0fb13eb Compare May 23, 2022 11:35

chriseth force-pushed the disable-ext-test-cases-broken-after-relaxing-inlining branch from 46eff4b to 49d5c0a Compare May 23, 2022 11:37

cameel and others added 3 commits May 23, 2022 13:37

Separate some inlining tests into pre- and post-homestead

f08d349

Relax inliner heuristics.

7168c27

Update test expectations.

77038aa

chriseth force-pushed the removeInliningConstraint branch from 0fb13eb to 77038aa Compare May 23, 2022 11:37

cameel reviewed May 23, 2022

View reviewed changes

Base automatically changed from disable-ext-test-cases-broken-after-relaxing-inlining to develop May 23, 2022 13:17

chriseth dismissed a stale review via 77038aa May 23, 2022 13:17

chriseth marked this pull request as ready for review May 23, 2022 13:17

chriseth approved these changes May 23, 2022

View reviewed changes

chriseth merged commit 45030f0 into develop May 23, 2022

chriseth deleted the removeInliningConstraint branch May 23, 2022 13:18

ekpyron mentioned this pull request Jun 7, 2022

Use side-effects of user defined functions in evm code transform. #12132

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Remove inlining constraint for large functions when targetting the new code transform. #12731

Remove inlining constraint for large functions when targetting the new code transform. #12731

ekpyron commented Mar 2, 2022 •

edited by cameel

Loading

chriseth Mar 2, 2022

chriseth Mar 2, 2022

ekpyron Mar 16, 2022

chriseth Mar 2, 2022

ekpyron Mar 2, 2022

ekpyron commented Mar 2, 2022

ekpyron Mar 2, 2022

ekpyron commented Mar 2, 2022

chriseth commented Mar 2, 2022

ekpyron Mar 16, 2022

cameel commented Mar 28, 2022

leonardoalt commented Apr 4, 2022

ekpyron Apr 13, 2022

cameel commented Apr 13, 2022

cameel commented May 20, 2022

cameel commented May 23, 2022

cameel May 23, 2022

chriseth May 23, 2022

chriseth commented May 23, 2022

	* Yul Optimizer: Improve inlining heauristics for via IR code generation and pure Yul compilation.
	* Yul Optimizer: Improve inlining heuristic for via IR code generation and pure Yul compilation.

Remove inlining constraint for large functions when targetting the new code transform. #12731

Remove inlining constraint for large functions when targetting the new code transform. #12731

Conversation

ekpyron commented Mar 2, 2022 • edited by cameel Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

ekpyron commented Mar 2, 2022

Choose a reason for hiding this comment

ekpyron commented Mar 2, 2022

chriseth commented Mar 2, 2022

Choose a reason for hiding this comment

cameel commented Mar 28, 2022

leonardoalt commented Apr 4, 2022

Choose a reason for hiding this comment

cameel commented Apr 13, 2022

cameel commented May 20, 2022

cameel commented May 23, 2022

ir-optimize-evm+yul

Relative

Absolute

Choose a reason for hiding this comment

Choose a reason for hiding this comment

chriseth commented May 23, 2022

ekpyron commented Mar 2, 2022 •

edited by cameel

Loading

`ir-optimize-evm+yul`