perf(levm): use signed remaining gas, improving gas cost calculation perf #4684

edg-l · 2025-09-29T14:09:03Z

Motivation

Description

Closes #issue_number

github-actions · 2025-09-29T14:12:24Z

Lines of code report

Total lines added: 16
Total lines removed: 0
Total lines changed: 16

Detailed view

+-----------------------------------------------------+-------+------+
| File                                                | Lines | Diff |
+-----------------------------------------------------+-------+------+
| ethrex/crates/vm/levm/src/call_frame.rs             | 354   | +4   |
+-----------------------------------------------------+-------+------+
| ethrex/crates/vm/levm/src/execution_handlers.rs     | 123   | +3   |
+-----------------------------------------------------+-------+------+
| ethrex/crates/vm/levm/src/opcode_handlers/system.rs | 802   | +5   |
+-----------------------------------------------------+-------+------+
| ethrex/crates/vm/levm/src/vm.rs                     | 456   | +4   |
+-----------------------------------------------------+-------+------+

github-actions · 2025-09-29T15:11:31Z

Benchmark Block Execution Results Comparison Against Main

Command	Mean [s]	Min [s]	Max [s]	Relative
`base`	87.833 ± 0.710	86.865	89.181	1.01 ± 0.01
`head`	87.289 ± 0.405	86.745	87.993	1.00

Oppen · 2025-09-30T14:02:52Z

crates/vm/levm/src/opcode_handlers/stack_memory_storage_flow.rs

        // EIP-2200
        let gas_left = self.current_call_frame.gas_remaining;
-        if gas_left <= SSTORE_STIPEND {
+        if gas_left as u64 <= SSTORE_STIPEND {


Maybe just change SSTORE_STIPEND to be i64.

crates/vm/levm/src/gas_cost.rs

JereSalo

This will work for real use cases but gas should theoretically be 64 bits according to the protocol, so we have to decide whether to stop supporting cases in which gas is larger than 63 bits or not.

edg-l · 2025-09-30T14:12:13Z

This will work for real use cases but gas should theoretically be 64 bits according to the protocol, so we have to decide whether to stop supporting cases in which gas is larger than 32 bits or not.

Agree, thats why the pr is in draft yet, we need to. decide for example whether we enforce a maximum gas limit below i64::max

MegaRedHand · 2025-09-30T14:18:26Z

EIP-7825 comes in Fusaka, limiting the gas limit per transaction to 2^24. I think it's safe to use i64 here, even i32, given that a tx with a bigger gas limit never happened before and won't happen after Fusaka.

JereSalo · 2025-09-30T21:08:15Z

EIP-7825 comes in Fusaka, limiting the gas limit per transaction to 2^24. I think it's safe to use i64 here, even i32, given that a tx with a bigger gas limit never happened before and won't happen after Fusaka.

Yeah that's right, just have to make sure to skip related tests and comment why. We should also add a comment above the i64 variable definition justifying briefly why we use i64 and not u64 in case somebody doesn't know

github-actions · 2025-10-01T10:23:19Z

Benchmark Results Comparison

No significant difference was registered for any benchmark run.

Detailed Results

Benchmark Results: BubbleSort

Command	Mean [s]	Min [s]	Max [s]	Relative
`main_revm_BubbleSort`	4.788 ± 0.020	4.775	4.844	1.04 ± 0.01
`main_levm_BubbleSort`	4.610 ± 0.019	4.590	4.642	1.00
`pr_revm_BubbleSort`	4.842 ± 0.025	4.813	4.888	1.05 ± 0.01
`pr_levm_BubbleSort`	4.636 ± 0.028	4.601	4.678	1.01 ± 0.01

Benchmark Results: ERC20Approval

Command	Mean [s]	Min [s]	Max [s]	Relative
`main_revm_ERC20Approval`	1.551 ± 0.008	1.540	1.567	1.00
`main_levm_ERC20Approval`	1.660 ± 0.072	1.626	1.864	1.07 ± 0.05
`pr_revm_ERC20Approval`	1.561 ± 0.012	1.547	1.581	1.01 ± 0.01
`pr_levm_ERC20Approval`	1.638 ± 0.007	1.630	1.652	1.06 ± 0.01

Benchmark Results: ERC20Mint

Command	Mean [ms]	Min [ms]	Max [ms]	Relative
`main_revm_ERC20Mint`	184.9 ± 0.9	183.9	186.8	1.00
`main_levm_ERC20Mint`	196.9 ± 3.7	194.5	205.4	1.07 ± 0.02
`pr_revm_ERC20Mint`	185.8 ± 0.7	185.2	187.4	1.01 ± 0.01
`pr_levm_ERC20Mint`	197.8 ± 3.2	195.0	203.3	1.07 ± 0.02

Benchmark Results: ERC20Transfer

Command	Mean [ms]	Min [ms]	Max [ms]	Relative
`main_revm_ERC20Transfer`	351.3 ± 0.7	350.5	352.3	1.00
`main_levm_ERC20Transfer`	383.3 ± 2.6	380.8	387.2	1.09 ± 0.01
`pr_revm_ERC20Transfer`	352.9 ± 1.1	351.7	354.7	1.00 ± 0.00
`pr_levm_ERC20Transfer`	384.3 ± 2.1	381.6	388.3	1.09 ± 0.01

Benchmark Results: Factorial

Command	Mean [ms]	Min [ms]	Max [ms]	Relative
`main_revm_Factorial`	241.2 ± 0.8	240.1	242.5	1.02 ± 0.01
`main_levm_Factorial`	276.4 ± 0.8	275.2	277.2	1.17 ± 0.01
`pr_revm_Factorial`	236.0 ± 1.1	234.9	238.4	1.00
`pr_levm_Factorial`	278.7 ± 0.9	277.9	280.8	1.18 ± 0.01

Benchmark Results: FactorialRecursive

Command	Mean [s]	Min [s]	Max [s]	Relative
`main_revm_FactorialRecursive`	1.628 ± 0.049	1.505	1.706	1.00 ± 0.04
`main_levm_FactorialRecursive`	8.810 ± 0.078	8.716	8.974	5.44 ± 0.11
`pr_revm_FactorialRecursive`	1.621 ± 0.028	1.556	1.655	1.00
`pr_levm_FactorialRecursive`	8.795 ± 0.080	8.652	8.927	5.43 ± 0.11

Benchmark Results: Fibonacci

Command	Mean [ms]	Min [ms]	Max [ms]	Relative
`main_revm_Fibonacci`	218.5 ± 1.7	214.7	219.9	1.02 ± 0.01
`main_levm_Fibonacci`	262.7 ± 2.7	260.2	267.0	1.23 ± 0.01
`pr_revm_Fibonacci`	213.7 ± 1.3	210.5	215.5	1.00
`pr_levm_Fibonacci`	271.5 ± 22.6	259.7	332.6	1.27 ± 0.11

Benchmark Results: FibonacciRecursive

Command	Mean [ms]	Min [ms]	Max [ms]	Relative
`main_revm_FibonacciRecursive`	872.0 ± 11.6	851.0	886.9	1.01 ± 0.02
`main_levm_FibonacciRecursive`	1086.2 ± 4.9	1080.3	1097.3	1.26 ± 0.02
`pr_revm_FibonacciRecursive`	859.4 ± 13.4	843.8	884.9	1.00
`pr_levm_FibonacciRecursive`	1101.2 ± 4.4	1095.5	1107.7	1.28 ± 0.02

Benchmark Results: ManyHashes

Command	Mean [ms]	Min [ms]	Max [ms]	Relative
`main_revm_ManyHashes`	12.4 ± 0.1	12.4	12.5	1.00 ± 0.01
`main_levm_ManyHashes`	13.6 ± 0.1	13.5	14.0	1.10 ± 0.01
`pr_revm_ManyHashes`	12.4 ± 0.1	12.3	12.6	1.00
`pr_levm_ManyHashes`	13.6 ± 0.1	13.5	13.8	1.10 ± 0.01

Benchmark Results: MstoreBench

Command	Mean [ms]	Min [ms]	Max [ms]	Relative
`main_revm_MstoreBench`	268.6 ± 4.2	264.9	275.0	1.00
`main_levm_MstoreBench`	797.2 ± 5.8	793.0	807.9	2.97 ± 0.05
`pr_revm_MstoreBench`	270.6 ± 11.2	263.3	300.3	1.01 ± 0.04
`pr_levm_MstoreBench`	804.4 ± 29.5	792.8	888.1	2.99 ± 0.12

Benchmark Results: Push

Command	Mean [ms]	Min [ms]	Max [ms]	Relative
`main_revm_Push`	298.4 ± 2.2	293.4	301.7	1.00
`main_levm_Push`	889.6 ± 4.4	885.7	898.8	2.98 ± 0.03
`pr_revm_Push`	298.5 ± 0.5	297.4	299.1	1.00 ± 0.01
`pr_levm_Push`	886.8 ± 1.7	884.7	891.0	2.97 ± 0.02

Benchmark Results: SstoreBench_no_opt

Command	Mean [ms]	Min [ms]	Max [ms]	Relative
`main_revm_SstoreBench_no_opt`	220.2 ± 2.4	218.5	225.9	2.42 ± 0.05
`main_levm_SstoreBench_no_opt`	91.0 ± 1.5	89.2	92.9	1.00
`pr_revm_SstoreBench_no_opt`	219.3 ± 0.5	218.3	220.0	2.41 ± 0.04
`pr_levm_SstoreBench_no_opt`	91.8 ± 3.5	88.9	101.0	1.01 ± 0.04

JereSalo

Good
What if we change the rest of the gas variables to i64 too? Would that make more sense instead of converting types every time? Just wanted to know your opinion, it's not necessary to do it in this PR

edg-l · 2025-10-01T13:09:39Z

Good What if we change the rest of the gas variables to i64 too? Would that make more sense instead of converting types every time? Just wanted to know your opinion, it's not necessary to do it in this PR

we would need to see if its worth it

simple signed remaining gas

8c3958a

github-actions bot assigned edg-l Sep 29, 2025

edg-l changed the title ~~simple signed remaining gas~~ perf(levm): simple signed remaining gas Sep 29, 2025

github-actions bot added levm Lambda EVM implementation performance labels Sep 29, 2025

github-project-automation bot added this to ethrex_performance and ethrex_l1 Sep 29, 2025

github-project-automation bot moved this to Todo in ethrex_performance Sep 29, 2025

Merge branch 'main' into signed_gas

d4dec36

Merge branch 'main' into signed_gas

74a22b3

Oppen reviewed Sep 30, 2025

View reviewed changes

JereSalo reviewed Sep 30, 2025

View reviewed changes

crates/vm/levm/src/gas_cost.rs Outdated Show resolved Hide resolved

JereSalo reviewed Sep 30, 2025

View reviewed changes

Merge branch 'main' into signed_gas

9401f0c

edg-l added 4 commits October 1, 2025 11:40

Merge branch 'main' into signed_gas

b956058

changelog

9152423

lints

1c65a06

change max gas limits

4d1395e

edg-l marked this pull request as ready for review October 1, 2025 10:16

edg-l requested a review from a team as a code owner October 1, 2025 10:16

edg-l moved this from Todo to In review in ethrex_performance Oct 1, 2025

edg-l changed the title ~~perf(levm): simple signed remaining gas~~ perf(levm): use signed remaining gas, improving gas cost calculation perf Oct 1, 2025

edg-l removed this from ethrex_l1 Oct 1, 2025

edg-l requested a review from JereSalo October 1, 2025 10:47

edg-l requested a review from Oppen October 1, 2025 10:47

JereSalo approved these changes Oct 1, 2025

View reviewed changes

Arkenan approved these changes Oct 1, 2025

View reviewed changes

edg-l added this pull request to the merge queue Oct 1, 2025

Merged via the queue into main with commit eedb9dd Oct 1, 2025
63 checks passed

edg-l deleted the signed_gas branch October 1, 2025 13:36

github-project-automation bot moved this from In review to Done in ethrex_performance Oct 1, 2025

Arkenan mentioned this pull request Oct 1, 2025

Evaluate using overflowing_sub instead of i64 for remaining gas #4721

Open

perf(levm): use signed remaining gas, improving gas cost calculation perf #4684

perf(levm): use signed remaining gas, improving gas cost calculation perf #4684

Uh oh!

Conversation

edg-l commented Sep 29, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

github-actions bot commented Sep 29, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Lines of code report

Uh oh!

github-actions bot commented Sep 29, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Benchmark Block Execution Results Comparison Against Main

Uh oh!

Oppen Sep 30, 2025

Choose a reason for hiding this comment

Uh oh!

Uh oh!

JereSalo left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

edg-l commented Sep 30, 2025

Uh oh!

MegaRedHand commented Sep 30, 2025

Uh oh!

JereSalo commented Sep 30, 2025

Uh oh!

github-actions bot commented Oct 1, 2025

Benchmark Results Comparison

Benchmark Results: BubbleSort

Benchmark Results: ERC20Approval

Benchmark Results: ERC20Mint

Benchmark Results: ERC20Transfer

Benchmark Results: Factorial

Benchmark Results: FactorialRecursive

Benchmark Results: Fibonacci

Benchmark Results: FibonacciRecursive

Benchmark Results: ManyHashes

Benchmark Results: MstoreBench

Benchmark Results: Push

Benchmark Results: SstoreBench_no_opt

Uh oh!

JereSalo left a comment

Choose a reason for hiding this comment

Uh oh!

edg-l commented Oct 1, 2025

Uh oh!

Uh oh!

Uh oh!

edg-l commented Sep 29, 2025 •

edited

Loading

github-actions bot commented Sep 29, 2025 •

edited

Loading

github-actions bot commented Sep 29, 2025 •

edited

Loading

JereSalo left a comment •

edited

Loading