perf(l1): avoid opening tries #5145

iovoid · 2025-10-31T20:11:36Z

Motivation

Currently to get a storage we are:

getting the block header to obtain the state root
getting the account to get the storage root
reading the value

The first step isn't needed because the state root is the same through the block, and the second isn't needed because since we're using a path-based DB we can directly get the value (without even using the trie, if the FKVs are computed).

Description

We make the VM database ask by state_root instead of block hash (skipping step 1), and we skip reading the account storage root when the FlatKeyValues are computed for it.

github-actions · 2025-10-31T20:14:23Z

Lines of code report

Total lines added: 112
Total lines removed: 26
Total lines changed: 138

Detailed view

+---------------------------------------------------------------+-------+------+
| File                                                          | Lines | Diff |
+---------------------------------------------------------------+-------+------+
| ethrex/crates/blockchain/blockchain.rs                        | 1305  | +10  |
+---------------------------------------------------------------+-------+------+
| ethrex/crates/blockchain/payload.rs                           | 676   | +4   |
+---------------------------------------------------------------+-------+------+
| ethrex/crates/blockchain/tracing.rs                           | 130   | +4   |
+---------------------------------------------------------------+-------+------+
| ethrex/crates/blockchain/vm.rs                                | 104   | +3   |
+---------------------------------------------------------------+-------+------+
| ethrex/crates/l2/based/block_fetcher.rs                       | 508   | +9   |
+---------------------------------------------------------------+-------+------+
| ethrex/crates/l2/sequencer/l1_committer.rs                    | 1064  | +5   |
+---------------------------------------------------------------+-------+------+
| ethrex/crates/networking/rpc/eth/account.rs                   | 261   | -1   |
+---------------------------------------------------------------+-------+------+
| ethrex/crates/storage/api.rs                                  | 246   | +3   |
+---------------------------------------------------------------+-------+------+
| ethrex/crates/storage/store.rs                                | 1521  | +8   |
+---------------------------------------------------------------+-------+------+
| ethrex/crates/storage/store_db/rocksdb.rs                     | 1594  | +54  |
+---------------------------------------------------------------+-------+------+
| ethrex/crates/storage/trie_db/rocksdb.rs                      | 194   | -13  |
+---------------------------------------------------------------+-------+------+
| ethrex/crates/storage/trie_db/rocksdb_locked.rs               | 75    | -10  |
+---------------------------------------------------------------+-------+------+
| ethrex/crates/vm/levm/bench/revm_comparison/src/levm_bench.rs | 88    | +4   |
+---------------------------------------------------------------+-------+------+
| ethrex/crates/vm/levm/runner/src/main.rs                      | 277   | +5   |
+---------------------------------------------------------------+-------+------+
| ethrex/tooling/ef_tests/blockchain/test_runner.rs             | 337   | -1   |
+---------------------------------------------------------------+-------+------+
| ethrex/tooling/ef_tests/state/utils.rs                        | 60    | -1   |
+---------------------------------------------------------------+-------+------+
| ethrex/tooling/ef_tests/state_v2/src/modules/utils.rs         | 47    | +3   |
+---------------------------------------------------------------+-------+------+

github-actions · 2025-10-31T21:00:20Z

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`	4.742 ± 0.014	4.718	4.759	1.04 ± 0.00
`main_levm_BubbleSort`	4.560 ± 0.015	4.535	4.580	1.00
`pr_revm_BubbleSort`	4.724 ± 0.011	4.714	4.741	1.04 ± 0.00
`pr_levm_BubbleSort`	4.652 ± 0.288	4.541	5.469	1.02 ± 0.06

Benchmark Results: ERC20Approval

Command	Mean [s]	Min [s]	Max [s]	Relative
`main_revm_ERC20Approval`	1.565 ± 0.021	1.549	1.616	1.00 ± 0.01
`main_levm_ERC20Approval`	1.663 ± 0.011	1.653	1.682	1.07 ± 0.01
`pr_revm_ERC20Approval`	1.559 ± 0.008	1.552	1.578	1.00
`pr_levm_ERC20Approval`	1.663 ± 0.005	1.655	1.670	1.07 ± 0.01

Benchmark Results: ERC20Mint

Command	Mean [ms]	Min [ms]	Max [ms]	Relative
`main_revm_ERC20Mint`	185.9 ± 0.5	185.3	186.6	1.00
`main_levm_ERC20Mint`	205.7 ± 1.2	203.1	207.1	1.11 ± 0.01
`pr_revm_ERC20Mint`	186.4 ± 0.4	185.7	187.0	1.00 ± 0.00
`pr_levm_ERC20Mint`	205.7 ± 1.9	203.2	208.3	1.11 ± 0.01

Benchmark Results: ERC20Transfer

Command	Mean [ms]	Min [ms]	Max [ms]	Relative
`main_revm_ERC20Transfer`	352.4 ± 2.4	350.5	359.0	1.00
`main_levm_ERC20Transfer`	393.8 ± 3.3	389.2	400.4	1.12 ± 0.01
`pr_revm_ERC20Transfer`	353.0 ± 1.2	351.8	355.5	1.00 ± 0.01
`pr_levm_ERC20Transfer`	394.7 ± 1.3	393.0	397.6	1.12 ± 0.01

Benchmark Results: Factorial

Command	Mean [ms]	Min [ms]	Max [ms]	Relative
`main_revm_Factorial`	234.0 ± 2.2	232.3	238.8	1.00
`main_levm_Factorial`	275.6 ± 1.6	273.9	279.4	1.18 ± 0.01
`pr_revm_Factorial`	234.5 ± 1.7	232.8	237.8	1.00 ± 0.01
`pr_levm_Factorial`	276.4 ± 0.9	275.5	277.5	1.18 ± 0.01

Benchmark Results: FactorialRecursive

Command	Mean [s]	Min [s]	Max [s]	Relative
`main_revm_FactorialRecursive`	1.668 ± 0.051	1.529	1.705	1.00
`main_levm_FactorialRecursive`	8.582 ± 0.044	8.513	8.643	5.14 ± 0.16
`pr_revm_FactorialRecursive`	1.679 ± 0.039	1.598	1.731	1.01 ± 0.04
`pr_levm_FactorialRecursive`	8.566 ± 0.041	8.492	8.636	5.13 ± 0.16

Benchmark Results: Fibonacci

Command	Mean [ms]	Min [ms]	Max [ms]	Relative
`main_revm_Fibonacci`	206.9 ± 1.0	205.8	208.7	1.00 ± 0.01
`main_levm_Fibonacci`	258.0 ± 7.5	252.1	274.1	1.25 ± 0.04
`pr_revm_Fibonacci`	206.6 ± 0.5	205.6	207.2	1.00
`pr_levm_Fibonacci`	258.9 ± 3.6	252.0	263.2	1.25 ± 0.02

Benchmark Results: FibonacciRecursive

Command	Mean [ms]	Min [ms]	Max [ms]	Relative
`main_revm_FibonacciRecursive`	880.4 ± 7.2	871.4	891.6	1.14 ± 0.01
`main_levm_FibonacciRecursive`	775.1 ± 6.7	766.9	789.6	1.00
`pr_revm_FibonacciRecursive`	879.1 ± 9.2	871.1	903.4	1.13 ± 0.02
`pr_levm_FibonacciRecursive`	780.4 ± 6.9	769.9	791.7	1.01 ± 0.01

Benchmark Results: ManyHashes

Command	Mean [ms]	Min [ms]	Max [ms]	Relative
`main_revm_ManyHashes`	12.5 ± 0.1	12.4	12.6	1.00
`main_levm_ManyHashes`	13.7 ± 0.1	13.6	13.9	1.09 ± 0.01
`pr_revm_ManyHashes`	12.5 ± 0.1	12.5	12.7	1.00 ± 0.01
`pr_levm_ManyHashes`	13.7 ± 0.1	13.6	13.7	1.09 ± 0.01

Benchmark Results: MstoreBench

Command	Mean [ms]	Min [ms]	Max [ms]	Relative
`main_revm_MstoreBench`	268.8 ± 3.7	264.7	274.0	1.06 ± 0.02
`main_levm_MstoreBench`	254.4 ± 3.4	251.3	261.7	1.00
`pr_revm_MstoreBench`	273.1 ± 10.4	263.9	295.8	1.07 ± 0.04
`pr_levm_MstoreBench`	255.8 ± 5.4	251.1	270.4	1.01 ± 0.03

Benchmark Results: Push

Command	Mean [ms]	Min [ms]	Max [ms]	Relative
`main_revm_Push`	295.8 ± 2.7	292.1	301.3	1.00
`main_levm_Push`	305.0 ± 2.4	303.3	311.1	1.03 ± 0.01
`pr_revm_Push`	319.4 ± 80.5	293.1	548.6	1.08 ± 0.27
`pr_levm_Push`	307.7 ± 3.8	305.4	318.4	1.04 ± 0.02

Benchmark Results: SstoreBench_no_opt

Command	Mean [ms]	Min [ms]	Max [ms]	Relative
`main_revm_SstoreBench_no_opt`	220.2 ± 1.8	218.4	223.0	2.64 ± 0.05
`main_levm_SstoreBench_no_opt`	85.9 ± 3.4	83.2	94.9	1.03 ± 0.05
`pr_revm_SstoreBench_no_opt`	218.8 ± 0.5	218.2	219.7	2.62 ± 0.05
`pr_levm_SstoreBench_no_opt`	83.5 ± 1.6	81.0	85.6	1.00

github-actions · 2025-10-31T21:13:22Z

Benchmark Block Execution Results Comparison Against Main

Command	Mean [s]	Min [s]	Max [s]	Relative
`base`	135.477 ± 0.223	135.056	135.755	1.00 ± 0.00
`head`	135.414 ± 0.223	134.929	135.786	1.00

crates/storage/store_db/in_memory.rs

**Motivation** Currently to get a storage we are: * getting the block header to obtain the state root * getting the account to get the storage root * reading the value The first step isn't needed because the state root is the same through the block, and the second isn't needed because since we're using a path-based DB we can directly get the value (without even using the trie, if the FKVs are computed). **Description** We make the VM database ask by state_root instead of block hash (skipping step 1), and we skip reading the account storage root when the FlatKeyValues are computed for it. --------- Co-authored-by: Ivan Litteri <[email protected]> Co-authored-by: Javier Chatruc <[email protected]> Co-authored-by: Javier Rodríguez Chatruc <[email protected]>

iovoid added 9 commits October 31, 2025 15:31

reduce the overhead added by opening tries

5b9a49a

fix initial value

6282207

use 64 byte fkv-done marker

f7d5446

avoid fetching block header

bcfcc62

also set last_computed_flatkeyvalue when finishing

9919ee0

support old fkv marker

f9ca53a

fix old marker support

54e29d2

fix marker length

cdcb5ae

do not change marker

3208eb9

github-actions bot assigned iovoid Oct 31, 2025

github-actions bot added L1 Ethereum client performance Block execution throughput and performance in general labels Oct 31, 2025

github-project-automation bot added this to ethrex_performance and ethrex_l1 Oct 31, 2025

github-project-automation bot moved this to Todo in ethrex_performance Oct 31, 2025

iovoid added 8 commits October 31, 2025 17:23

unify last_written initialization

a0f8d87

update perf changelog

b0c592e

update bencher

6d24eff

Merge branch 'main' into avoid_opening_trie

f154e24

fix state root

214dfd0

update more users of StoreVmDatabase

e4d55e4

remove unneeded await

37d9fa5

order writes to avoid errors

173b6e0

iovoid added 2 commits October 31, 2025 18:01

Merge branch 'main' into avoid_opening_trie

5ddd6d7

fix docs

11340ab

iovoid marked this pull request as ready for review October 31, 2025 21:08

iovoid requested a review from a team as a code owner October 31, 2025 21:08

ethrex-project-sync bot moved this to In Review in ethrex_l1 Oct 31, 2025

fmt tooling

ac872a6

iovoid mentioned this pull request Oct 31, 2025

perf(l1): optimized VM database for RocksDB #5073

Closed

ilitteri and others added 5 commits October 31, 2025 19:02

Merge branch 'main' into avoid_opening_trie

9ab3a81

fix witness generation

d8b5931

Merge branch 'main' into avoid_opening_trie

e2df98a

fix merge

d3918f8

fix merge in blockchain

631a076

edg-l approved these changes Nov 3, 2025

View reviewed changes

pablodeymo reviewed Nov 3, 2025

View reviewed changes

crates/storage/store_db/in_memory.rs Outdated Show resolved Hide resolved

jrchatruc approved these changes Nov 3, 2025

View reviewed changes

jrchatruc and others added 2 commits November 3, 2025 14:41

Merge branch 'main' into avoid_opening_trie

f3eef26

set flatkeyvalue_computed default implementation

d4234f4

jrchatruc enabled auto-merge November 3, 2025 17:53

pablodeymo approved these changes Nov 3, 2025

View reviewed changes

jrchatruc added this pull request to the merge queue Nov 3, 2025

Merged via the queue into main with commit 1130d27 Nov 3, 2025
54 checks passed

jrchatruc deleted the avoid_opening_trie branch November 3, 2025 18:52

github-project-automation bot moved this from Todo to Done in ethrex_performance Nov 3, 2025

github-project-automation bot moved this from In Review to Done in ethrex_l1 Nov 3, 2025

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

perf(l1): avoid opening tries #5145

perf(l1): avoid opening tries #5145

Uh oh!

iovoid commented Oct 31, 2025

Uh oh!

github-actions bot commented Oct 31, 2025 •

edited

Loading

Uh oh!

github-actions bot commented Oct 31, 2025 •

edited

Loading

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!

github-actions bot commented Oct 31, 2025 •

edited

Loading

Uh oh!

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

6 participants

perf(l1): avoid opening tries #5145

perf(l1): avoid opening tries #5145

Uh oh!

Conversation

iovoid commented Oct 31, 2025

Uh oh!

github-actions bot commented Oct 31, 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 Oct 31, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

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!

github-actions bot commented Oct 31, 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!

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

6 participants

github-actions bot commented Oct 31, 2025 •

edited

Loading

github-actions bot commented Oct 31, 2025 •

edited

Loading

github-actions bot commented Oct 31, 2025 •

edited

Loading