perf(levm): reuse stack pool #5179
Open
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Lines of code reportTotal lines added: Detailed view |
Benchmark Results ComparisonNo significant difference was registered for any benchmark run. Detailed ResultsBenchmark 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
|
iovoid
force-pushed
the
reuse_stack_pool
branch
from
b9855fe to
b3b636e
Compare
Benchmark Block Execution Results Comparison Against Main
|
pablodeymo
reviewed
Nov 6, 2025
Comment on lines
+258
to
+260
| std::mem::swap(&mut vm.stack_pool, stack_pool); | ||
| let result = vm.execute().map_err(VMError::into); | ||
| std::mem::swap(&mut vm.stack_pool, stack_pool); |
There was a problem hiding this comment.
I think it's better to use std::mem::take at the beggining and then take. We don't need to actually swap.
There was a problem hiding this comment.
Pull Request Overview
This PR introduces a stack pool optimization for transaction execution within blocks to improve performance by reusing stack allocations across multiple transactions.
- Adds a new
execute_tx_in_blockfunction that accepts a mutable stack pool parameter for reuse across transaction executions - Updates
execute_block_pipelineto create and maintain a shared stack pool throughout block execution - Swaps the stack pool with the VM before and after transaction execution to enable stack reuse
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
Comment on lines
+244
to
+254
| pub fn execute_tx_in_block( | ||
| // The transaction to execute. | ||
| tx: &Transaction, | ||
| // The transactions recovered address | ||
| tx_sender: Address, | ||
| // The block header for the current block. | ||
| block_header: &BlockHeader, | ||
| db: &mut GeneralizedDatabase, | ||
| vm_type: VMType, | ||
| stack_pool: &mut Vec<Stack>, | ||
| ) -> Result<ExecutionReport, EvmError> { |
There was a problem hiding this comment.
This new public function is missing a documentation comment. Consider adding a doc comment explaining the purpose of this function and how it differs from execute_tx, particularly the role of the stack_pool parameter in optimizing stack allocations across multiple transaction executions within a block.
Comment on lines
+258
to
+260
| std::mem::swap(&mut vm.stack_pool, stack_pool); | ||
| let result = vm.execute().map_err(VMError::into); | ||
| std::mem::swap(&mut vm.stack_pool, stack_pool); |
There was a problem hiding this comment.
If vm.execute() panics, the second std::mem::swap won't execute, potentially leaving the stack_pool in an inconsistent state. Consider using a guard pattern or RAII wrapper to ensure the swap-back happens even in case of panics, or document that panic safety is not guaranteed here.
Suggested change
| std::mem::swap(&mut vm.stack_pool, stack_pool); | |
| let result = vm.execute().map_err(VMError::into); | |
| std::mem::swap(&mut vm.stack_pool, stack_pool); | |
| struct StackPoolGuard<'a> { | |
| a: &'a mut Vec<Stack>, | |
| b: &'a mut Vec<Stack>, | |
| } | |
| impl<'a> Drop for StackPoolGuard<'a> { | |
| fn drop(&mut self) { | |
| std::mem::swap(self.a, self.b); | |
| } | |
| } | |
| std::mem::swap(&mut vm.stack_pool, stack_pool); | |
| let _guard = StackPoolGuard { a: &mut vm.stack_pool, b: stack_pool }; | |
| let result = vm.execute().map_err(VMError::into); |
| ) -> Result<BlockExecutionResult, EvmError> { | ||
| Self::prepare_block(block, db, vm_type)?; | ||
|
|
||
| let mut shared_stack_pool = Vec::with_capacity(1024); |
There was a problem hiding this comment.
The magic number 1024 for the stack pool capacity is not explained. Consider either adding a comment explaining why this specific capacity was chosen, or extracting it as a named constant (e.g., STACK_POOL_CAPACITY) to improve code readability and maintainability.
| pub fn execute_tx_in_block( | ||
| // The transaction to execute. | ||
| tx: &Transaction, | ||
| // The transactions recovered address |
There was a problem hiding this comment.
Grammar issue in comment: "The transactions recovered address" should be "The transaction's recovered address" (possessive form).
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Motivation
Currently, allocation of a new stack object is taking ~8% of execution.
Description
We make a single allocation of every layer.