Make Shared transparent to parent kernels#8576
Open
gatesn wants to merge 2 commits into
Open
Conversation
Signed-off-by: Nicholas Gates <nick@nickgates.com>
github-actions
Bot
removed
the
action/benchmark-sql
gatesn
added
changelog/performance
Contributor
Polar Signals Profiling ResultsLatest Run
Previous Runs (1)
Powered by Polar Signals Cloud |
Contributor
|
I think we caused some weird race in github, but benchmarks seem to be running https://github.com/vortex-data/vortex/actions/runs/28109422384 |
Contributor
Benchmarks: PolarSignals Profiling (base)Vortex (geomean): 0.963x ➖ How to read Verdict and Engines
datafusion / vortex-file-compressed (0.963x ➖, 1↑ 0↓)
No file size changes detected. |
Contributor
Benchmarks: TPC-H SF=1 on NVME (base)Verdict: No clear signal (low confidence) How to read Verdict and Engines
datafusion / vortex-file-compressed (0.882x ✅, 16↑ 0↓)
datafusion / parquet (0.906x ➖, 10↑ 0↓)
datafusion / arrow (0.844x ✅, 19↑ 0↓)
duckdb / vortex-file-compressed (0.891x ✅, 16↑ 0↓)
duckdb / parquet (0.935x ➖, 3↑ 0↓)
File Size Changes (17 files changed, -44.4% overall, 5↑ 12↓)
Totals:
|
Contributor
Benchmarks: FineWeb NVMe (base)Verdict: No clear signal (low confidence) How to read Verdict and Engines
datafusion / vortex-file-compressed (1.104x ❌, 0↑ 3↓)
datafusion / parquet (1.084x ➖, 0↑ 2↓)
duckdb / vortex-file-compressed (1.112x ❌, 0↑ 7↓)
duckdb / parquet (1.074x ➖, 0↑ 1↓)
File Size Changes (3 files changed, -46.3% overall, 0↑ 3↓)
Totals:
|
Contributor
Benchmarks: TPC-DS SF=1 on NVME (base)Verdict: No clear signal (low confidence) How to read Verdict and Engines
datafusion / vortex-file-compressed (0.947x ➖, 17↑ 3↓)
datafusion / parquet (0.912x ➖, 31↑ 0↓)
duckdb / vortex-file-compressed (0.926x ➖, 29↑ 0↓)
duckdb / parquet (0.951x ➖, 2↑ 1↓)
File Size Changes (31 files changed, -43.5% overall, 2↑ 29↓)
Totals:
|
Contributor
Benchmarks: FineWeb S3 (base)Verdict: No clear signal (environment too noisy confidence) How to read Verdict and Engines
datafusion / vortex-file-compressed (0.592x ✅, 6↑ 0↓)
datafusion / parquet (0.861x ➖, 0↑ 0↓)
duckdb / vortex-file-compressed (0.786x ➖, 3↑ 0↓)
duckdb / parquet (0.941x ➖, 0↑ 0↓)
|
Contributor
Benchmarks: Statistical and Population Genetics (base)Verdict: No clear signal (low confidence) How to read Verdict and Engines
duckdb / vortex-file-compressed (1.003x ➖, 0↑ 0↓)
duckdb / parquet (0.997x ➖, 0↑ 0↓)
File Size Changes (3 files changed, -32.3% overall, 0↑ 3↓)
Totals:
|
Contributor
Benchmarks: Clickbench Sorted on NVME (base)Verdict: No clear signal (low confidence) How to read Verdict and Engines
datafusion / vortex-file-compressed (1.064x ➖, 1↑ 3↓)
datafusion / parquet (1.001x ➖, 0↑ 0↓)
duckdb / vortex-file-compressed (0.963x ➖, 2↑ 0↓)
duckdb / parquet (0.993x ➖, 0↑ 0↓)
File Size Changes (201 files changed, -42.6% overall, 40↑ 161↓)
Totals:
|
Contributor
Benchmarks: TPC-H SF=10 on NVME (base)Verdict: No clear signal (low confidence) How to read Verdict and Engines
datafusion / vortex-file-compressed (1.020x ➖, 0↑ 2↓)
datafusion / parquet (1.007x ➖, 0↑ 0↓)
datafusion / arrow (1.012x ➖, 0↑ 1↓)
duckdb / vortex-file-compressed (1.003x ➖, 0↑ 0↓)
duckdb / parquet (1.002x ➖, 0↑ 0↓)
File Size Changes (47 files changed, -44.5% overall, 14↑ 33↓)
Totals:
|
Contributor
Benchmarks: Clickbench on NVME (base)Verdict: No clear signal (low confidence) How to read Verdict and Engines
datafusion / vortex-file-compressed (0.986x ➖, 1↑ 4↓)
datafusion / parquet (0.983x ➖, 1↑ 2↓)
duckdb / vortex-file-compressed (0.969x ➖, 4↑ 3↓)
duckdb / parquet (0.967x ➖, 1↑ 0↓)
File Size Changes (201 files changed, -39.1% overall, 54↑ 147↓)
Totals:
|
Contributor
Benchmarks: TPC-H SF=1 on S3 (base)Verdict: No clear signal (environment too noisy confidence) How to read Verdict and Engines
datafusion / vortex-file-compressed (1.086x ➖, 1↑ 5↓)
datafusion / parquet (1.352x ❌, 0↑ 12↓)
duckdb / vortex-file-compressed (0.996x ➖, 0↑ 0↓)
duckdb / parquet (0.976x ➖, 3↑ 2↓)
|
Merging this PR will degrade performance by 15.92%
|
| Mode | Benchmark | BASE |
HEAD |
Efficiency | |
|---|---|---|---|---|---|
| ❌ | Simulation | chunked_bool_canonical_into[(1000, 10)] |
15.9 µs | 26.9 µs | -40.83% |
| ❌ | Simulation | copy_nullable[65536] |
1 ms | 1.4 ms | -24.27% |
| ❌ | Simulation | search_index_below_min_chunked |
1.3 ms | 1.6 ms | -20.33% |
| ❌ | Simulation | search_index_mixed_out_of_range_chunked |
1.3 ms | 1.7 ms | -20.19% |
| ❌ | Simulation | search_index_full_range_random_chunked |
1.4 ms | 1.7 ms | -19.76% |
| ❌ | Simulation | chunked_varbinview_into_canonical[(1000, 10)] |
168.9 µs | 206.6 µs | -18.22% |
| ❌ | Simulation | search_index_above_max_chunked |
1.7 ms | 2 ms | -17.59% |
| ❌ | Simulation | copy_non_nullable[65536] |
908.5 µs | 1,089.2 µs | -16.59% |
| ❌ | Simulation | search_index_in_range_chunked |
1.9 ms | 2.3 ms | -16.23% |
| ⚡ | Simulation | chunked_varbinview_canonical_into[(1000, 10)] |
191.1 µs | 154.7 µs | +23.54% |
| ⚡ | Simulation | compact_sliced[(4096, 90)] |
837.5 ns | 750 ns | +11.67% |
Tip
Investigate this regression by commenting @codspeedbot fix this regression on this PR, or directly use the CodSpeed MCP with your agent.
Comparing ngates/onpair-split-1-shared-parent-kernels (95c4949) with develop (797b650)
Footnotes
-
4 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩
lwwmanning
reviewed
Jun 24, 2026
robert3005
reviewed
Jun 24, 2026
Signed-off-by: "Nicholas Gates" <nick@nickgates.com>
github-actions
Bot
removed
the
action/benchmark-sql
Contributor
Benchmarks: PolarSignals ProfilingVortex (geomean): 0.926x ➖ How to read Verdict and Engines
datafusion / vortex-file-compressed (0.926x ➖, 3↑ 0↓)
No file size changes detected. |
Contributor
Benchmarks: FineWeb NVMeVerdict: No clear signal (low confidence) How to read Verdict and Engines
datafusion / vortex-file-compressed (1.035x ➖, 1↑ 2↓)
datafusion / parquet (1.017x ➖, 0↑ 1↓)
duckdb / vortex-file-compressed (1.023x ➖, 0↑ 1↓)
duckdb / parquet (1.002x ➖, 0↑ 0↓)
File Size Changes (3 files changed, -46.3% overall, 1↑ 2↓)
Totals:
|
Contributor
Benchmarks: TPC-H SF=1 on NVMEVerdict: No clear signal (low confidence) How to read Verdict and Engines
datafusion / vortex-file-compressed (0.989x ➖, 1↑ 3↓)
datafusion / parquet (0.962x ➖, 2↑ 0↓)
datafusion / arrow (0.966x ➖, 1↑ 0↓)
duckdb / vortex-file-compressed (0.957x ➖, 2↑ 1↓)
duckdb / parquet (0.942x ➖, 4↑ 0↓)
File Size Changes (18 files changed, -44.2% overall, 4↑ 14↓)
Totals:
|
Contributor
Benchmarks: TPC-DS SF=1 on NVMEVerdict: No clear signal (low confidence) How to read Verdict and Engines
datafusion / vortex-file-compressed (1.041x ➖, 1↑ 19↓)
datafusion / parquet (0.995x ➖, 2↑ 2↓)
duckdb / vortex-file-compressed (1.004x ➖, 10↑ 19↓)
duckdb / parquet (0.996x ➖, 1↑ 1↓)
File Size Changes (36 files changed, -43.5% overall, 3↑ 33↓)
Totals:
|
Contributor
Benchmarks: FineWeb S3Verdict: No clear signal (low confidence) How to read Verdict and Engines
datafusion / vortex-file-compressed (0.936x ➖, 2↑ 2↓)
datafusion / parquet (1.061x ➖, 0↑ 0↓)
duckdb / vortex-file-compressed (1.023x ➖, 0↑ 0↓)
duckdb / parquet (1.104x ➖, 0↑ 0↓)
|
Contributor
Benchmarks: Statistical and Population GeneticsVerdict: No clear signal (low confidence) How to read Verdict and Engines
duckdb / vortex-file-compressed (1.046x ➖, 0↑ 0↓)
duckdb / parquet (1.041x ➖, 0↑ 0↓)
File Size Changes (3 files changed, -32.3% overall, 1↑ 2↓)
Totals:
|
Contributor
Benchmarks: Clickbench Sorted on NVMEVerdict: No clear signal (environment too noisy confidence) How to read Verdict and Engines
datafusion / vortex-file-compressed (1.002x ➖, 2↑ 2↓)
datafusion / parquet (1.065x ➖, 0↑ 2↓)
duckdb / vortex-file-compressed (1.069x ➖, 0↑ 2↓)
duckdb / parquet (0.923x ➖, 3↑ 0↓)
File Size Changes (301 files changed, -42.6% overall, 100↑ 201↓)
Totals:
|
Contributor
Benchmarks: TPC-H SF=10 on NVMEVerdict: No clear signal (environment too noisy confidence) How to read Verdict and Engines
datafusion / vortex-file-compressed (0.757x ✅, 20↑ 0↓)
datafusion / parquet (0.728x ✅, 21↑ 0↓)
datafusion / arrow (0.692x ✅, 22↑ 0↓)
duckdb / vortex-file-compressed (0.747x ✅, 22↑ 0↓)
duckdb / parquet (0.839x ✅, 20↑ 0↓)
File Size Changes (48 files changed, -44.5% overall, 11↑ 37↓)
Totals:
|
Contributor
Benchmarks: Clickbench on NVMEVerdict: No clear signal (low confidence) How to read Verdict and Engines
datafusion / vortex-file-compressed (0.935x ➖, 18↑ 4↓)
datafusion / parquet (0.943x ➖, 11↑ 0↓)
duckdb / vortex-file-compressed (0.899x ✅, 19↑ 1↓)
duckdb / parquet (1.002x ➖, 1↑ 0↓)
File Size Changes (201 files changed, -39.1% overall, 48↑ 153↓)
Totals:
|
Contributor
Benchmarks: TPC-H SF=1 on S3Verdict: No clear signal (environment too noisy confidence) How to read Verdict and Engines
datafusion / vortex-file-compressed (1.066x ➖, 1↑ 6↓)
datafusion / parquet (0.989x ➖, 0↑ 2↓)
duckdb / vortex-file-compressed (1.167x ➖, 0↑ 3↓)
duckdb / parquet (1.086x ➖, 0↑ 1↓)
|
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
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.
Rational for this change
SharedArrayis a cache wrapper, but parent-kernel execution currently treats it as an ordinary child encoding. That can force canonicalization before the wrapped array gets a chance to handle the parent operation itself. This makesSharedArraytransparent for parent reductions and parent execution, so existing encoding-specific kernels still fire through shared wrappers.No tracked issue.
What changes are included in this PR?
SharedArraynow delegates parent reduction to its current wrapped array. Parent execution also checks the wrapped array before falling back to ordinary execution, avoiding unnecessary cache population and canonicalization.What APIs are changed? Are there any user-facing changes?
No public API changes. Query results are unchanged; this only changes internal execution dispatch.