Use version of chainbase which implements a slab allocator for small size buffers. #1070
Conversation
This file contains 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
|
Note:start |
heifner
reviewed
Dec 16, 2024
| @@ -2191,6 +2195,7 @@ struct controller_impl { | |||
| section.read_row(rows_for_this_tid, db); | |||
| read_row_count.fetch_add(2u, std::memory_order_relaxed); | |||
|
|
|||
| // utils_t::preallocate(db, rows_for_this_tid.value); | |||
There was a problem hiding this comment.
I thought this would help, but upon testing it didn't seem to make a difference, so I commented it out. I thought that maybe we'd want to experiment further in the future which is why I left it commented out. In any case I probably should add a comment that it doesn't appear to make a difference.
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Outdated suggestions cannot be applied.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.
Resolves #1049.
As the issue describes, the current chainbase does a lot of small buffer allocations directly through the bip allocator which hits the segment manager directly. Not only this is inefficient (the rbtree of free buffers need to be rebalanced for every alloc/free), it also has a higher memory overhead.
The slab allocators for small buffers (in the updated chainbase submodule) batches allocations of small buffers into larger blocks, reducing the fragmentattion within the shared memory segment, and providing better efficiency and a smaller memory footprint.
For comparison, here are some results comparing snapshot loading on my linux box (release build)
snapshot from Aug 11, 2024
snapshot from Sept 29, 2024
I also expect a slight performance increase for processing transactions, as the
_undo_stackdeque allocations used to go through the segment allocator as well (for blocks of 16undo_state), and these allocations are now batched.Associated chainbase PR
AntelopeIO/chainbase#54