Skip to content

perf(l2): compute base blob fee once per block#6006

Open
xqft wants to merge 11 commits into
mainfrom
fake_exp_opt
Open

perf(l2): compute base blob fee once per block#6006
xqft wants to merge 11 commits into
mainfrom
fake_exp_opt

Conversation

@xqft
Copy link
Copy Markdown
Contributor

@xqft xqft commented Jan 22, 2026 •
edited
Loading

Summary

Optimizes EIP-4844 blob gas calculation by computing base_blob_fee_per_gas once per block instead of twice per transaction, eliminating redundant fake_exponential calls per block.

Motivation

The base blob fee per gas is constant for all transactions in a block (determined by parent_excess_blob_gas and base_fee_update_fraction), yet it was being recomputed:

  • in setup_env() (once per transaction)
    -calculate_blob_gas_cost() during transaction execution (via deduct_caller)

This optimization computes the value once at the block level and reuses it.

Changes

Modified:

  • calculate_blob_gas_cost() now accepts pre-computed base_blob_fee_per_gas from Environment
  • execute_block() and execute_block_pipeline() compute the value once before the transaction loop
  • setup_env() accepts base_blob_fee_per_gas as a parameter instead of computing it
  • Updated all call sites: execute_tx, execute_tx_in_block, trace_tx_calls, warm_block, rerun_block

Performance Results

SP1 - Block 24292781 (mainnet, 341 transactions)

Metric Main Optimized Improvement
Total Execution 602,664,256 cycles 526,071,327 cycles -76.6M (-12.7%)
execute_block 380,618,779 cycles 304,025,850 cycles -76.6M (-20.1%)

All cycle savings are concentrated in the transaction execution phase (execute_block), confirming the optimization eliminates redundant calculations during transaction processing.

ZisK - Block 24292781 (mainnet, 341 transactions)

Metric Main Optimized Improvement
Total Steps 718,743,877 676,897,554 -41.8M (-5.82%)
Total Cost 80.3B 76.5B -3.9B (-4.83%)

Cost Distribution Changes:

Component Main Optimized Change
MAIN 48.9B (60.83%) 46.0B (60.20%) -2.8B (-5.93%)
OPCODES 11.3B (14.13%) 10.6B (13.87%) -748M (-6.59%)
PRECOMPILES 13.9B (17.36%) 13.9B (18.24%) 0 (0%)
MEMORY 5.9B (7.31%) 5.6B (7.31%) -287M (-4.89%)

Test Configuration:

  • Hardware: Apple M1 Pro, 16GB RAM
  • Block: 24292781 (mainnet, 341 transactions, 36M gas)
  • SP1 version: 5.2.4
  • ZisK version: 0.15.0
  • Branches: main (cb6e45a) vs fake_exp_opt (0ed64a1)

xqft added 5 commits January 22, 2026 17:23
@xqft
perf(block): add i128 fast path for fake_exponential in zkVM contexts
2a9101d 
Replace U256 arithmetic with i128 for EIP-4844 blob gas calculation when
running in zkVM (zisk/sp1/risc0). This is safe because:
- factor = MIN_BASE_FEE_PER_BLOB_GAS (1)
- numerator = parent_excess_blob_gas (u64, typically < 100M)
- denominator = update_fraction (u64, ~8.8M for mainnet)

Expected savings: 25-30M steps (3.8-4.5%) from reduced checked arithmetic
overhead and smaller integer operations.

Includes test to verify i128 matches U256 for typical values.
@xqft
feat(common): add zisk/sp1 feature flags for zkVM optimizations
b92d219
@xqft
perf(levm): use i128 fake_exponential in get_base_fee_per_blob_gas fo…
f663135 
…r zkVM

Apply same i128 optimization to levm's get_base_fee_per_blob_gas function
which is called during transaction execution. This is the main call site
with ~769 calls per block.
@xqft
optimize base blob fee calculation by computing once per block
6175dd1
@xqft
add zkVM optimization logbook with P2 block-level caching entry
9b82026
@xqft xqft mentioned this pull request Jan 22, 2026
Closed
@xqft xqft changed the title perf(l2): compute base blob fee once per block chore(l2): compute base blob fee once per block Jan 22, 2026
@github-actions
Copy link
Copy Markdown

github-actions Bot commented Jan 22, 2026 •
edited
Loading

Lines of code report

Total lines added: 66
Total lines removed: 4
Total lines changed: 70

Detailed view 
+-------------------------------------------------+-------+------+
| File                                            | Lines | Diff |
+-------------------------------------------------+-------+------+
| ethrex/crates/vm/backends/levm/mod.rs           | 732   | +38  |
+-------------------------------------------------+-------+------+
| ethrex/crates/vm/backends/levm/tracing.rs       | 78    | +17  |
+-------------------------------------------------+-------+------+
| ethrex/crates/vm/backends/mod.rs                | 182   | +11  |
+-------------------------------------------------+-------+------+
| ethrex/crates/vm/levm/src/hooks/default_hook.rs | 349   | -3   |
+-------------------------------------------------+-------+------+
| ethrex/crates/vm/levm/src/utils.rs              | 431   | -1   |
+-------------------------------------------------+-------+------+

@github-actions
Copy link
Copy Markdown

github-actions Bot commented Jan 22, 2026 •
edited
Loading

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 2.933 ± 0.037 2.905 3.032 1.00
main_levm_BubbleSort 3.054 ± 0.104 2.996 3.344 1.04 ± 0.04
pr_revm_BubbleSort 2.940 ± 0.013 2.916 2.953 1.00 ± 0.01
pr_levm_BubbleSort 3.063 ± 0.011 3.050 3.081 1.04 ± 0.01

Benchmark Results: ERC20Approval

Command Mean [ms] Min [ms] Max [ms] Relative
main_revm_ERC20Approval 992.5 ± 13.0 977.0 1014.8 1.00
main_levm_ERC20Approval 1075.4 ± 9.4 1057.0 1088.3 1.08 ± 0.02
pr_revm_ERC20Approval 996.5 ± 10.3 984.9 1015.0 1.00 ± 0.02
pr_levm_ERC20Approval 1077.8 ± 11.1 1064.8 1104.9 1.09 ± 0.02

Benchmark Results: ERC20Mint

Command Mean [ms] Min [ms] Max [ms] Relative
main_revm_ERC20Mint 134.7 ± 1.1 133.8 136.5 1.00
main_levm_ERC20Mint 162.2 ± 5.6 157.8 176.0 1.20 ± 0.04
pr_revm_ERC20Mint 135.8 ± 1.8 134.4 140.6 1.01 ± 0.02
pr_levm_ERC20Mint 160.0 ± 1.4 158.2 163.1 1.19 ± 0.01

Benchmark Results: ERC20Transfer

Command Mean [ms] Min [ms] Max [ms] Relative
main_revm_ERC20Transfer 233.4 ± 3.2 230.7 240.0 1.00
main_levm_ERC20Transfer 268.5 ± 1.1 266.3 270.3 1.15 ± 0.02
pr_revm_ERC20Transfer 236.6 ± 4.4 232.8 247.7 1.01 ± 0.02
pr_levm_ERC20Transfer 273.9 ± 3.0 271.0 279.6 1.17 ± 0.02

Benchmark Results: Factorial

Command Mean [ms] Min [ms] Max [ms] Relative
main_revm_Factorial 220.3 ± 0.9 218.7 222.1 1.00
main_levm_Factorial 255.5 ± 1.3 254.1 257.6 1.16 ± 0.01
pr_revm_Factorial 224.6 ± 2.2 222.3 228.9 1.02 ± 0.01
pr_levm_Factorial 256.2 ± 2.1 253.5 259.4 1.16 ± 0.01

Benchmark Results: FactorialRecursive

Command Mean [s] Min [s] Max [s] Relative
main_revm_FactorialRecursive 1.669 ± 0.047 1.551 1.726 1.00
main_levm_FactorialRecursive 8.476 ± 0.051 8.404 8.523 5.08 ± 0.15
pr_revm_FactorialRecursive 1.696 ± 0.061 1.569 1.777 1.02 ± 0.05
pr_levm_FactorialRecursive 8.274 ± 0.105 8.120 8.447 4.96 ± 0.15

Benchmark Results: Fibonacci

Command Mean [ms] Min [ms] Max [ms] Relative
main_revm_Fibonacci 194.8 ± 2.2 192.9 200.1 1.00
main_levm_Fibonacci 249.2 ± 3.1 244.9 256.7 1.28 ± 0.02
pr_revm_Fibonacci 198.2 ± 0.9 197.3 199.7 1.02 ± 0.01
pr_levm_Fibonacci 249.7 ± 5.8 242.1 263.6 1.28 ± 0.03

Benchmark Results: FibonacciRecursive

Command Mean [ms] Min [ms] Max [ms] Relative
main_revm_FibonacciRecursive 880.5 ± 9.0 865.6 895.2 1.25 ± 0.02
main_levm_FibonacciRecursive 703.5 ± 6.0 696.1 715.8 1.00
pr_revm_FibonacciRecursive 883.7 ± 8.4 871.2 897.2 1.26 ± 0.02
pr_levm_FibonacciRecursive 714.1 ± 32.9 692.8 804.5 1.02 ± 0.05

Benchmark Results: ManyHashes

Command Mean [ms] Min [ms] Max [ms] Relative
main_revm_ManyHashes 8.6 ± 0.2 8.4 9.1 1.00
main_levm_ManyHashes 9.4 ± 0.1 9.3 9.6 1.09 ± 0.02
pr_revm_ManyHashes 8.8 ± 0.1 8.7 9.1 1.02 ± 0.02
pr_levm_ManyHashes 9.3 ± 0.1 9.2 9.5 1.08 ± 0.02

Benchmark Results: MstoreBench

Command Mean [ms] Min [ms] Max [ms] Relative
main_revm_MstoreBench 260.4 ± 3.6 256.7 269.4 1.17 ± 0.02
main_levm_MstoreBench 222.8 ± 1.8 218.9 225.9 1.00
pr_revm_MstoreBench 260.0 ± 1.7 257.9 263.3 1.17 ± 0.01
pr_levm_MstoreBench 224.0 ± 1.1 222.1 225.6 1.01 ± 0.01

Benchmark Results: Push

Command Mean [ms] Min [ms] Max [ms] Relative
main_revm_Push 286.6 ± 1.8 284.9 291.0 1.02 ± 0.01
main_levm_Push 282.0 ± 3.3 273.0 285.0 1.00
pr_revm_Push 286.4 ± 1.3 284.7 288.8 1.02 ± 0.01
pr_levm_Push 282.9 ± 2.1 278.2 286.6 1.00 ± 0.01

Benchmark Results: SstoreBench_no_opt

Command Mean [ms] Min [ms] Max [ms] Relative
main_revm_SstoreBench_no_opt 171.0 ± 6.9 162.1 185.6 1.90 ± 0.09
main_levm_SstoreBench_no_opt 90.1 ± 2.2 87.4 93.6 1.00
pr_revm_SstoreBench_no_opt 170.8 ± 4.0 164.2 180.0 1.90 ± 0.06
pr_levm_SstoreBench_no_opt 90.9 ± 2.0 88.2 93.5 1.01 ± 0.03

@xqft xqft changed the title chore(l2): compute base blob fee once per block perf(l2): compute base blob fee once per block Jan 22, 2026
@github-actions github-actions Bot added L2 Rollup client performance Block execution throughput and performance in general labels Jan 22, 2026
@JereSalo
Copy link
Copy Markdown
Contributor

JereSalo commented Jan 23, 2026 •
edited
Loading

Just wanted to add a quick comment, I tested this with mainnet blocks without zkVMs and the improvement is also noticeable! Maybe I'll upload a picture later of the results.

Edit: It's with a tool that I'm building, I tested random branches and I just saw that yours had a noticeable speedup

@xqft
Copy link
Copy Markdown
Contributor Author

xqft commented Jan 23, 2026 •
edited
Loading

@JereSalo Nice!! yeah would love to include those results in the description too

@xqft xqft changed the title perf(l2): compute base blob fee once per block chore(l2): compute base blob fee once per block Jan 23, 2026
@github-actions github-actions Bot removed the performance Block execution throughput and performance in general label Jan 23, 2026
@xqft xqft changed the title chore(l2): compute base blob fee once per block perf(l2): compute base blob fee once per block Jan 23, 2026
@github-actions github-actions Bot added the performance Block execution throughput and performance in general label Jan 23, 2026
@github-actions
Copy link
Copy Markdown

github-actions Bot commented Jan 23, 2026 •
edited
Loading

Benchmark Block Execution Results Comparison Against Main

Command Mean [s] Min [s] Max [s] Relative
base 65.180 ± 0.243 64.794 65.635 1.00
head 65.646 ± 0.371 65.194 66.234 1.01 ± 0.01

@xqft xqft marked this pull request as ready for review January 23, 2026 17:23
@xqft xqft requested a review from a team as a code owner January 23, 2026 17:24
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@Oppen Oppen Oppen approved these changes

@azteca1998 azteca1998 azteca1998 approved these changes

@ilitteri ilitteri ilitteri approved these changes

Assignees

@xqft xqft

Labels

L2 Rollup client performance Block execution throughput and performance in general

Projects

ethrex_l2
Status: No status
ethrex_performance
Status: Todo

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

5 participants

@xqft @JereSalo @Oppen @azteca1998 @ilitteri