perf: bypass async I/O pipeline for in-memory buffer reads#6813
Draft
perf: bypass async I/O pipeline for in-memory buffer reads#6813
Conversation
Add BufferSegmentSource that resolves segment requests synchronously by slicing a ByteBuffer directly, and use it in open_buffer() instead of routing through FileSegmentSource. Previously, open_buffer() delegated to open_read() which created a FileSegmentSource backed by the full async I/O pipeline: mpsc channels, IoRequestStream coalescing, buffer_unordered concurrency limiter, and oneshot callbacks. For an in-memory or mmap'd ByteBuffer, the actual data access is a zero-copy slice (~0µs), but each segment request paid the async pipeline overhead. With BufferSegmentSource, each request() returns future::ready() with the sliced buffer. No channels, no coalescing, no callbacks. Signed-off-by: Alexander Droste <alexander.droste@protonmail.com>
0ax1
added
do not merge
github-actions
bot
removed
the
action/benchmark-sql
0ax1
added
action/benchmark-sql
github-actions
bot
removed
the
action/benchmark-sql
Contributor
Polar Signals Profiling ResultsLatest Run
Powered by Polar Signals Cloud |
Contributor
Benchmarks: FineWeb NVMeSummary
Detailed Results Table
|
0ax1
changed the title
Contributor
Benchmarks: TPC-DS SF=1 on NVMESummary
Detailed Results Table
|
Contributor
Benchmarks: TPC-H SF=10 on NVMESummary
Detailed Results Table
|
Contributor
Benchmarks: FineWeb S3Summary
Detailed Results Table
|
Contributor
Benchmarks: PolarSignals ProfilingSummary
Detailed Results Table
|
Contributor
Benchmarks: TPC-H SF=1 on S3Summary
Detailed Results Table
|
Contributor
Benchmarks: TPC-H SF=10 on S3Summary
Detailed Results Table
|
Contributor
Benchmarks: Statistical and Population GeneticsSummary
Detailed Results Table
|
Contributor
Benchmarks: Clickbench on NVMESummary
Detailed Results Table
|
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
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.
Add BufferSegmentSource that resolves segment requests synchronously by slicing a ByteBuffer directly, and use it in open_buffer() instead of routing through FileSegmentSource.
Previously, open_buffer() delegated to open_read() which created a FileSegmentSource backed by the full async I/O pipeline: mpsc channels, IoRequestStream coalescing, buffer_unordered concurrency limiter, and oneshot callbacks. For an in-memory or mmap'd ByteBuffer, the actual data access is a zero-copy slice (~0µs), but each segment request paid the async pipeline overhead.
With BufferSegmentSource, each request() returns future::ready() with the sliced buffer. No channels, no coalescing, no callbacks.