Repository navigation
Sub segment read p2 - #7368
Closed
joseph-isaacs wants to merge 29 commits into
Closed
Sub segment read p2#7368joseph-isaacs wants to merge 29 commits into
joseph-isaacs wants to merge 29 commits into
Conversation
Signed-off-by: Nicholas Gates <nick@nickgates.com>
Signed-off-by: Nicholas Gates <nick@nickgates.com>
Signed-off-by: Nicholas Gates <nick@nickgates.com>
Signed-off-by: Nicholas Gates <nick@nickgates.com>
Signed-off-by: Nicholas Gates <nick@nickgates.com>
Signed-off-by: Nicholas Gates <nick@nickgates.com>
Signed-off-by: Nicholas Gates <nick@nickgates.com>
Signed-off-by: Nicholas Gates <nick@nickgates.com>
Signed-off-by: Joe Isaacs <joe.isaacs@live.co.uk> # Conflicts: # vortex-array/public-api.lock # vortex-array/src/array/vtable/dyn_.rs # vortex-array/src/array/vtable/mod.rs # vortex-array/src/arrays/bool/array.rs # vortex-array/src/arrays/fixed_size_list/compute/slice.rs # vortex-layout/src/layouts/flat/reader.rs # vortex-layout/src/layouts/flat/writer.rs # vortex-layout/src/segments/cache.rs
Contributor
Polar Signals Profiling ResultsLatest Run
Previous Runs (8)
Powered by Polar Signals Cloud |
Contributor
Benchmarks: PolarSignals ProfilingVortex (geomean): 1.018x ➖ datafusion / vortex-file-compressed (1.018x ➖, 0↑ 0↓)
|
Contributor
File Sizes: PolarSignals ProfilingFile Size Changes (1 files changed, +0.0% overall, 1↑ 0↓)
Totals:
|
Contributor
Benchmarks: TPC-H SF=1 on NVMEVerdict: No clear signal (environment too noisy confidence) datafusion / vortex-file-compressed (0.970x ➖, 1↑ 0↓)
datafusion / vortex-compact (0.989x ➖, 1↑ 1↓)
datafusion / parquet (1.032x ➖, 1↑ 4↓)
datafusion / arrow (1.054x ➖, 0↑ 4↓)
duckdb / vortex-file-compressed (0.991x ➖, 0↑ 0↓)
duckdb / vortex-compact (0.996x ➖, 1↑ 1↓)
duckdb / parquet (0.995x ➖, 2↑ 2↓)
duckdb / duckdb (1.056x ➖, 0↑ 5↓)
Full attributed analysis
|
Contributor
File Sizes: TPC-H SF=1 on NVMEFile Size Changes (195 files changed, -98.4% overall, 0↑ 195↓)
Totals:
|
Contributor
Benchmarks: FineWeb NVMeVerdict: No clear signal (low confidence) datafusion / vortex-file-compressed (1.013x ➖, 1↑ 4↓)
datafusion / vortex-compact (1.000x ➖, 1↑ 6↓)
datafusion / parquet (1.170x ❌, 0↑ 8↓)
duckdb / vortex-file-compressed (1.018x ➖, 1↑ 6↓)
duckdb / vortex-compact (1.018x ➖, 1↑ 6↓)
duckdb / parquet (1.182x ❌, 0↑ 9↓)
Full attributed analysis
|
Contributor
File Sizes: FineWeb NVMeNo file size changes detected. |
Contributor
Benchmarks: TPC-DS SF=1 on NVMEVerdict: No clear signal (low confidence) datafusion / vortex-file-compressed (1.010x ➖, 0↑ 2↓)
datafusion / vortex-compact (1.007x ➖, 0↑ 0↓)
datafusion / parquet (1.010x ➖, 0↑ 1↓)
duckdb / vortex-file-compressed (1.002x ➖, 3↑ 3↓)
duckdb / vortex-compact (0.998x ➖, 4↑ 0↓)
duckdb / parquet (1.009x ➖, 0↑ 1↓)
duckdb / duckdb (1.002x ➖, 1↑ 2↓)
Full attributed analysis
|
Contributor
File Sizes: TPC-DS SF=1 on NVMENo file size changes detected. |
Contributor
Benchmarks: TPC-H SF=10 on NVMEVerdict: No clear signal (low confidence) datafusion / vortex-file-compressed (1.017x ➖, 0↑ 1↓)
datafusion / vortex-compact (1.022x ➖, 0↑ 1↓)
datafusion / parquet (1.043x ➖, 0↑ 2↓)
datafusion / arrow (1.038x ➖, 0↑ 0↓)
duckdb / vortex-file-compressed (1.049x ➖, 0↑ 5↓)
duckdb / vortex-compact (0.986x ➖, 0↑ 1↓)
duckdb / parquet (1.009x ➖, 0↑ 0↓)
duckdb / duckdb (0.960x ➖, 0↑ 0↓)
Full attributed analysis
|
Contributor
File Sizes: TPC-H SF=10 on NVMENo file size changes detected. |
Contributor
Benchmarks: FineWeb S3Verdict: No clear signal (low confidence) datafusion / vortex-file-compressed (0.728x ➖, 2↑ 0↓)
datafusion / vortex-compact (0.835x ➖, 2↑ 0↓)
datafusion / parquet (1.072x ➖, 0↑ 0↓)
duckdb / vortex-file-compressed (0.892x ➖, 1↑ 1↓)
duckdb / vortex-compact (0.884x ➖, 1↑ 1↓)
duckdb / parquet (1.018x ➖, 0↑ 0↓)
Full attributed analysis
|
joseph-isaacs
force-pushed
the
ji/sub-segment-read
branch
from
April 13, 2026 14:29
9619e54 to
e20fa4d
Compare
Signed-off-by: Joe Isaacs <joe.isaacs@live.co.uk> # Conflicts: # encodings/fsst/src/canonical.rs
Merging this PR will improve performance by 27.66%
Performance Changes
Comparing 
Footnotes
|
Contributor
|
This PR has been marked as stale because it has been open for 14 days with no activity. Please comment or remove the stale label if you wish to keep it active, otherwise it will be closed in 7 days |
Contributor
|
This PR was closed because it has been inactive for 7 days since being marked as stale. |
joseph-isaacs
pushed a commit
that referenced
this pull request
Oct 8, 2026
The page planner only understands Primitive, BitPacked, Struct and FixedSizeList over ALP-RD, so random access into taxi-style columns (FoR, ALP, Dict, BitPacked with patches, ...) always read whole segments. Reuse the lazy-buffer idea from #7273/#7368 as a fallback for segments the planner cannot handle: - `LazySegmentBuffer` is a `DeviceBuffer` that records a byte range of a segment without reading it; slicing narrows the range and host access errors instead of blocking. - The array is decoded over lazy buffers, with buffers up to a quarter page read up front so slicing code that inspects patches, dictionaries or run ends sees host bytes. Each run of selected rows is sliced, which pushes the slice into the buffers through the encodings' own rules; the narrowed ranges are read in one batch and the tree is rebuilt with `with_slots`/`with_buffers`. - The read falls back to the whole segment when the narrowed bytes cost more than half of it, when a short run still needs nearly every lazy byte (the encoding does not push slices down; remembered per reader), for segments under four pages, and on any error. Taxi random access with partial reads is 4-6% faster than develop; the other datasets are unchanged or faster. Signed-off-by: Claude <noreply@anthropic.com> Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_014Jsub1UwkQ5C4ErXB45K9s
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
2 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.
Signed-off-by: Nicholas Gates nick@nickgates.com<!--
Thank you for submitting a pull request! We appreciate your time and effort.
Please make sure to provide enough information so that we can review your pull
request. The Summary and Testing sections below contain guidance on what to
include.
-->
Summary
Closes: #000
Testing