Pass execution context through list view rebuild#8274
Open
connortsui20 wants to merge 2 commits into
Open
Conversation
joseph-isaacs
added
changelog/break
connortsui20
force-pushed
the
ct/rebuild-execute
branch
2 times, most recently
from
9514a28 to
d85c62e
Compare
Merging this PR will improve performance by 23.88%
|
| Mode | Benchmark | BASE |
HEAD |
Efficiency | |
|---|---|---|---|---|---|
| ❌ | Simulation | decompress_rd[f64, (100000, 0.01)] |
669.1 µs | 981.2 µs | -31.8% |
| ❌ | Simulation | decompress_rd[f64, (100000, 0.1)] |
669.1 µs | 981.1 µs | -31.8% |
| ❌ | Simulation | decompress_rd[f32, (100000, 0.1)] |
413.4 µs | 543.9 µs | -24% |
| ❌ | Simulation | decompress_rd[f32, (100000, 0.01)] |
413.4 µs | 543.9 µs | -23.99% |
| ❌ | Simulation | compare[1] |
59.3 µs | 67.1 µs | -11.75% |
| ❌ | Simulation | compare[6] |
71.3 µs | 80.5 µs | -11.45% |
| ❌ | Simulation | patched_take_10k_adversarial |
260 µs | 289 µs | -10.03% |
| ⚡ | Simulation | dict_canonicalize_zipfian[16, 1000] |
50.5 µs | 23.6 µs | ×2.1 |
| ⚡ | Simulation | compare[56] |
327.4 µs | 224.7 µs | +45.73% |
| ⚡ | Simulation | compare[62] |
363.8 µs | 249.9 µs | +45.56% |
| ⚡ | Simulation | compare[63] |
371.2 µs | 255.4 µs | +45.31% |
| ⚡ | Simulation | compare[60] |
353.8 µs | 243.6 µs | +45.25% |
| ⚡ | Simulation | compare[61] |
362.6 µs | 250.7 µs | +44.67% |
| ⚡ | Simulation | compare[58] |
347.2 µs | 240.8 µs | +44.17% |
| ⚡ | Simulation | compare[59] |
354.3 µs | 246.1 µs | +43.95% |
| ⚡ | Simulation | compare[57] |
345.8 µs | 241.3 µs | +43.32% |
| ⚡ | Simulation | compare[54] |
330.4 µs | 231.4 µs | +42.77% |
| ⚡ | Simulation | compare[55] |
337.1 µs | 236.4 µs | +42.61% |
| ⚡ | Simulation | compare[52] |
320.4 µs | 225.1 µs | +42.33% |
| ⚡ | Simulation | compare[48] |
295.4 µs | 207.5 µs | +42.33% |
| ... | ... | ... | ... | ... | ... |
ℹ️ Only the first 20 benchmarks are displayed. Go to the app to view all benchmarks.
Tip
Investigate this regression by commenting @codspeedbot fix this regression on this PR, or directly use the CodSpeed MCP with your agent.
Comparing ct/rebuild-execute (01701be) with develop (5e3aedb)
github-actions
Bot
removed
the
action/benchmark
Signed-off-by: Connor Tsui <connor.tsui20@gmail.com>
connortsui20
force-pushed
the
ct/rebuild-execute
branch
from
d85c62e to
03c6d2f
Compare
Contributor
Polar Signals Profiling ResultsLatest Run
Powered by Polar Signals Cloud |
Contributor
Benchmarks: PolarSignals ProfilingVortex (geomean): 1.102x ❌ How to read Verdict and Engines
datafusion / vortex-file-compressed (1.102x ❌, 0↑ 6↓)
No file size changes detected. |
Contributor
Benchmarks: FineWeb NVMeVerdict: No clear signal (low confidence) How to read Verdict and Engines
datafusion / vortex-file-compressed (0.993x ➖, 0↑ 0↓)
datafusion / vortex-compact (1.006x ➖, 0↑ 0↓)
datafusion / parquet (0.997x ➖, 0↑ 0↓)
duckdb / vortex-file-compressed (0.973x ➖, 1↑ 0↓)
duckdb / vortex-compact (1.013x ➖, 0↑ 0↓)
duckdb / parquet (1.018x ➖, 0↑ 0↓)
File Size Changes (1 files changed, -0.0% overall, 0↑ 1↓)
Totals:
|
Contributor
Benchmarks: TPC-H SF=1 on NVMEVerdict: No clear signal (low confidence) How to read Verdict and Engines
datafusion / vortex-file-compressed (1.178x ❌, 0↑ 22↓)
datafusion / vortex-compact (1.154x ❌, 0↑ 21↓)
datafusion / parquet (1.088x ➖, 0↑ 10↓)
datafusion / arrow (1.172x ❌, 0↑ 18↓)
duckdb / vortex-file-compressed (1.117x ❌, 0↑ 18↓)
duckdb / vortex-compact (1.106x ❌, 0↑ 14↓)
duckdb / parquet (1.071x ➖, 0↑ 5↓)
duckdb / duckdb (1.077x ➖, 0↑ 5↓)
File Size Changes (9 files changed, +0.0% overall, 5↑ 4↓)
Totals:
|
Contributor
Benchmarks: TPC-DS SF=1 on NVMEVerdict: No clear signal (low confidence) How to read Verdict and Engines
datafusion / vortex-file-compressed (0.995x ➖, 0↑ 1↓)
datafusion / vortex-compact (0.997x ➖, 1↑ 1↓)
datafusion / parquet (1.002x ➖, 1↑ 1↓)
duckdb / vortex-file-compressed (1.001x ➖, 1↑ 2↓)
duckdb / vortex-compact (1.000x ➖, 1↑ 0↓)
duckdb / parquet (1.004x ➖, 1↑ 1↓)
duckdb / duckdb (0.994x ➖, 0↑ 0↓)
File Size Changes (5 files changed, -0.0% overall, 1↑ 4↓)
Totals:
|
Contributor
Benchmarks: FineWeb S3Verdict: No clear signal (environment too noisy confidence) How to read Verdict and Engines
datafusion / vortex-file-compressed (1.183x ➖, 0↑ 1↓)
datafusion / vortex-compact (1.077x ➖, 0↑ 1↓)
datafusion / parquet (1.288x ➖, 0↑ 4↓)
duckdb / vortex-file-compressed (1.071x ➖, 0↑ 0↓)
duckdb / vortex-compact (1.056x ➖, 0↑ 0↓)
duckdb / parquet (1.090x ➖, 0↑ 0↓)
|
Contributor
Benchmarks: Statistical and Population GeneticsVerdict: No clear signal (low confidence) How to read Verdict and Engines
duckdb / vortex-file-compressed (0.997x ➖, 0↑ 1↓)
duckdb / vortex-compact (1.008x ➖, 0↑ 0↓)
duckdb / parquet (1.000x ➖, 0↑ 0↓)
File Size Changes (1 files changed, -0.0% overall, 0↑ 1↓)
Totals:
|
Contributor
Benchmarks: Random AccessVortex (geomean): 0.931x ➖ How to read Verdict and Engines
unknown / unknown (1.002x ➖, 12↑ 4↓)
|
Contributor
Benchmarks: TPC-H SF=10 on NVMEVerdict: No clear signal (low confidence) How to read Verdict and Engines
datafusion / vortex-file-compressed (1.000x ➖, 0↑ 0↓)
datafusion / vortex-compact (0.992x ➖, 0↑ 0↓)
datafusion / parquet (0.995x ➖, 1↑ 0↓)
datafusion / arrow (0.999x ➖, 0↑ 0↓)
duckdb / vortex-file-compressed (0.991x ➖, 0↑ 0↓)
duckdb / vortex-compact (0.999x ➖, 0↑ 0↓)
duckdb / parquet (0.998x ➖, 1↑ 0↓)
duckdb / duckdb (0.998x ➖, 0↑ 0↓)
File Size Changes (26 files changed, +0.0% overall, 14↑ 12↓)
Totals:
|
Contributor
Benchmarks: Clickbench on NVMEVerdict: No clear signal (environment too noisy confidence) How to read Verdict and Engines
datafusion / vortex-file-compressed (1.029x ➖, 2↑ 4↓)
datafusion / parquet (1.036x ➖, 1↑ 1↓)
duckdb / vortex-file-compressed (1.063x ➖, 1↑ 8↓)
duckdb / parquet (1.002x ➖, 1↑ 0↓)
duckdb / duckdb (1.014x ➖, 1↑ 0↓)
File Size Changes (104 files changed, +0.0% overall, 53↑ 51↓)
Totals:
|
Contributor
Author
|
I dont trust codspeed here |
Contributor
Benchmarks: Appian on NVMEVerdict: No clear signal (low confidence) How to read Verdict and Engines
datafusion / vortex-file-compressed (1.092x ➖, 0↑ 2↓)
datafusion / parquet (1.079x ➖, 0↑ 2↓)
duckdb / vortex-file-compressed (1.063x ➖, 0↑ 1↓)
duckdb / parquet (1.051x ➖, 0↑ 0↓)
duckdb / duckdb (1.048x ➖, 0↑ 0↓)
File Size Changes (4 files changed, -0.0% overall, 0↑ 4↓)
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 (1.151x ➖, 0↑ 4↓)
datafusion / vortex-compact (1.102x ➖, 0↑ 3↓)
datafusion / parquet (1.083x ➖, 0↑ 3↓)
duckdb / vortex-file-compressed (1.133x ➖, 0↑ 3↓)
duckdb / vortex-compact (1.105x ➖, 0↑ 1↓)
duckdb / parquet (1.070x ➖, 0↑ 0↓)
|
Contributor
Benchmarks: CompressionVortex (geomean): 1.000x ➖ How to read Verdict and Engines
unknown / unknown (1.011x ➖, 0↑ 1↓)
|
Contributor
Benchmarks: TPC-H SF=10 on S3Verdict: No clear signal (environment too noisy confidence) How to read Verdict and Engines
datafusion / vortex-file-compressed (1.084x ➖, 0↑ 4↓)
datafusion / vortex-compact (1.158x ➖, 0↑ 2↓)
datafusion / parquet (1.035x ➖, 0↑ 0↓)
duckdb / vortex-file-compressed (1.105x ➖, 0↑ 0↓)
duckdb / vortex-compact (1.165x ➖, 0↑ 4↓)
duckdb / parquet (1.118x ➖, 0↑ 1↓)
|
Hoist `create_execution_ctx()` out of the divan-timed closures into the untimed `with_inputs` setup (using `bench_values` / `bench_local_values` where a by-value context is needed) so the benchmarks measure the operation under test rather than per-sample context creation. Replace every non-CUDA benchmark use of the deprecated global `LEGACY_SESSION` with a local `static SESSION` built from `VortexSession::empty().with::<ArraySession>()`, which is exactly what `LEGACY_SESSION` is defined as, so the change is behavior-preserving while reducing reliance on the legacy global. Document why `ListViewBuilder::extend_from_array_unchecked` still mints a `LEGACY_SESSION` context: the `ArrayBuilder` trait does not thread an `ExecutionCtx` through its extend methods. Signed-off-by: Connor Tsui <connor.tsui20@gmail.com>
| /// will not have any leading or trailing garbage data. | ||
| /// The output `ListViewArray` will be zero-copyable back to a `ListArray`, and additionally it will | ||
| /// not have any leading or trailing garbage data. | ||
| pub fn list_view_from_list(list: ListArray, ctx: &mut ExecutionCtx) -> VortexResult<ListViewArray> { |
Contributor
There was a problem hiding this comment.
Also should this be a method on the ListArray?
Contributor
Author
There was a problem hiding this comment.
I wanted it standalone since it is unclear if this should live on list view or list
| .offsets() | ||
| .clone() | ||
| .execute::<PrimitiveArray>(&mut ctx) | ||
| .vortex_expect("failed to execute list view offsets into a PrimitiveArray"); |
Contributor
Author
There was a problem hiding this comment.
probably, but not much we can do about it because of the builders API
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.
Summary
TODO is there a tracking issue for deprecating the old to_canonical path? This PR makes a solid amount of progress in removing deprecated stuff.
Threads
ExecutionCtxthroughListView::rebuildand the paths it reaches (reset_offsets,compact,list_from_list_view,recursive_list_from_list_view) so they reuse the caller's session instead of each minting a freshLEGACY_SESSIONcontext internally.reset_offsetsinlist_view_from_listnow propagates with?instead ofvortex_expectsince execution can fail once it runs through it (#8274 (comment))API Changes
A few places that now get an
ExecutionCtxpassed through them.Testing
Tests and benchmarks also now use a local empty session. This is just a lot of mechanical / not semantic changes.