Use batch-run when replaying EVM transactions during ingestion

[m-Peter]

Currently, we replay each EVM transaction individually, which requires creating a new EVM emulator, a new BlockView, and to commit the changes in the end.

For blocks with large numbers of transactions, we could achieve a much better performance, by using the existing BatchRunTransactions API.

[coderabbitai]

📝 CodeRabbit Plan Mode

Generate an implementation plan and prompts that you can use with your favorite coding agent.

• Create Plan

Examples

• Example 1
• Example 2

---

🔗 Similar Issues

Related Issues

• Investigate feasibility of storing EVM registers on transaction level #729
• Transactions are not guaranteed to run according to their arrival order #699
• Transaction submission from multiple EOAs in the same batch #926
• Remove tx trace generation & storage from ingestion engine #937

🔗 Related PRs

#239 - Improve tests for batched transactions [merged]
#299 - Transaction pool [merged]
#835 - Optimize logic on BatchTxPool for single tx EOAs [merged]

👤 Suggested Assignees

• m-Peter
• devbugging

---

🧪 Issue enrichment is currently in open beta.

You can configure auto-planning by selecting labels in the issue_enrichment configuration.

To disable automatic issue enrichment, add the following to your .coderabbit.yaml:

issue_enrichment:
  auto_enrich:
    enabled: false

💬 Have feedback or questions? Drop into our discord or schedule a call!

[bluesign]

I don't think there will be much performance gain here, emulator is pretty cheap. blockview can be reused though in existing replayTransactionExecution. ( Which will also get rid of emulator creation) Otherwise precompiles would be a lot of work to handle, with minimal gains.

[m-Peter]

@bluesign I was mostly referring to this use that you uncovered a while back: onflow/flow-go#6958. I will do some local benchmarks, to see if there's any gain.

[bluesign]

yeah here is a bit different situation; it already uses in memory storage ( https://github.com/onflow/flow-go/blob/76749cca97380fff55249a04a2ff05c62d61ebd3/fvm/evm/offchain/sync/replayer.go#L80 ) and construct block proposal afterwards ( https://github.com/onflow/flow-go/blob/76749cca97380fff55249a04a2ff05c62d61ebd3/fvm/evm/offchain/utils/replay.go#L54-L59 )

but I am curious if there will be any gain also, better to benchmark for sure.

[m-Peter]

but I am curious if there will be any gain also, better to benchmark for sure.

I will give it a try, to be sure.