core/rawdb: add accessors with reusable key buffers #30770
+67
−12
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
Member
|
Can you measure the reduction in memory allocations achieved by applying this change? Please evaluate its impact during both snap sync and full sync. This pull request introduces optimizations to reuse the buffer for keying. These changes are only meaningful if they result in a obvious reduction in memory allocations. |
Member
|
Instead of reusing the same provided buffer for key construction, we can consider defining a global-shared byte slice pool and use it for all db APIs. The implementation can be channel based, e.g. https://github.com/ethereum/go-ethereum/pull/30743/files#diff-81c34779bf9837c7e60ee5a804bb5e951fa34ae8a47a6a6dd1e9bfcea868d728 I will have an attempt and benchmark the performance impact later. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 participants
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Outdated suggestions cannot be applied.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.
This came from an alloc dump on a running sync (mingeth).
Quite a lot of
bytesare allocated temporarily when we write to db (batches), in order to construct the keys. These allocs are very shortlived, since the write-operations are not allowed to hold on to the keys, but instead copies it to some interrnal buffers.This PR changes some of the most heavy mass-write-locations to reuse a key buffer. The locations are for example writing accounts received over snapsync, storage received over snapsync and merging in-memory snapshot-changes to disk.
Not sure how large effect this has. I suspect not a whole lot, but I'll run a benchmark sync and see if it anything is discernable.
Not high prio