Conversation
Polar Signals Profiling ResultsLatest Run
Powered by Polar Signals Cloud |
Signed-off-by: Connor Tsui <connor.tsui20@gmail.com>
0a03eea to
c3ddec1
Compare
Benchmarks: PolarSignals ProfilingVortex (geomean): 1.026x ➖ datafusion / vortex-file-compressed (1.026x ➖, 0↑ 0↓)
|
File Sizes: PolarSignals ProfilingNo file size changes detected. |
Benchmarks: FineWeb NVMeVerdict: No clear signal (low confidence) datafusion / vortex-file-compressed (0.990x ➖, 1↑ 0↓)
datafusion / vortex-compact (1.049x ➖, 0↑ 2↓)
datafusion / parquet (0.990x ➖, 0↑ 0↓)
duckdb / vortex-file-compressed (0.946x ➖, 2↑ 0↓)
duckdb / vortex-compact (1.011x ➖, 0↑ 0↓)
duckdb / parquet (0.940x ➖, 2↑ 0↓)
Full attributed analysis
|
File Sizes: FineWeb NVMeNo file size changes detected. |
Benchmarks: TPC-H SF=1 on NVMEVerdict: No clear signal (low confidence) datafusion / vortex-file-compressed (0.990x ➖, 1↑ 0↓)
datafusion / vortex-compact (0.991x ➖, 0↑ 0↓)
datafusion / parquet (0.988x ➖, 2↑ 0↓)
datafusion / arrow (1.012x ➖, 1↑ 0↓)
duckdb / vortex-file-compressed (1.071x ➖, 0↑ 2↓)
duckdb / vortex-compact (1.024x ➖, 0↑ 2↓)
duckdb / parquet (1.020x ➖, 0↑ 2↓)
duckdb / duckdb (0.997x ➖, 0↑ 0↓)
Full attributed analysis
|
File Sizes: TPC-H SF=1 on NVMENo file size changes detected. |
Benchmarks: TPC-DS SF=1 on NVMEVerdict: No clear signal (low confidence) datafusion / vortex-file-compressed (1.108x ❌, 0↑ 54↓)
datafusion / vortex-compact (1.083x ➖, 0↑ 26↓)
datafusion / parquet (1.095x ➖, 0↑ 41↓)
duckdb / vortex-file-compressed (1.104x ❌, 0↑ 59↓)
duckdb / vortex-compact (1.085x ➖, 0↑ 37↓)
duckdb / parquet (1.068x ➖, 0↑ 20↓)
duckdb / duckdb (1.071x ➖, 0↑ 30↓)
Full attributed analysis
|
File Sizes: TPC-DS SF=1 on NVMENo file size changes detected. |
Benchmarks: FineWeb S3Verdict: No clear signal (environment too noisy confidence) datafusion / vortex-file-compressed (0.740x ➖, 4↑ 1↓)
datafusion / vortex-compact (1.389x ❌, 0↑ 4↓)
datafusion / parquet (1.017x ➖, 0↑ 0↓)
duckdb / vortex-file-compressed (1.006x ➖, 0↑ 0↓)
duckdb / vortex-compact (1.128x ➖, 0↑ 2↓)
duckdb / parquet (1.028x ➖, 0↑ 0↓)
Full attributed analysis
|
Benchmarks: Random AccessVortex (geomean): 0.891x ✅ unknown / unknown (0.979x ➖, 7↑ 3↓)
|
Benchmarks: Statistical and Population GeneticsVerdict: No clear signal (low confidence) duckdb / vortex-file-compressed (1.032x ➖, 0↑ 1↓)
duckdb / vortex-compact (1.030x ➖, 0↑ 0↓)
duckdb / parquet (0.999x ➖, 0↑ 0↓)
Full attributed analysis
|
File Sizes: Statistical and Population GeneticsNo file size changes detected. |
Benchmarks: TPC-H SF=10 on NVMEVerdict: No clear signal (low confidence) datafusion / vortex-file-compressed (1.067x ➖, 0↑ 7↓)
datafusion / vortex-compact (1.064x ➖, 0↑ 0↓)
datafusion / parquet (1.071x ➖, 0↑ 5↓)
datafusion / arrow (1.079x ➖, 0↑ 4↓)
duckdb / vortex-file-compressed (1.050x ➖, 0↑ 0↓)
duckdb / vortex-compact (1.059x ➖, 0↑ 1↓)
duckdb / parquet (1.023x ➖, 0↑ 0↓)
duckdb / duckdb (1.042x ➖, 0↑ 0↓)
Full attributed analysis
|
File Sizes: TPC-H SF=10 on NVMENo file size changes detected. |
Benchmarks: TPC-H SF=1 on S3Verdict: No clear signal (environment too noisy confidence) datafusion / vortex-file-compressed (0.942x ➖, 1↑ 0↓)
datafusion / vortex-compact (1.090x ➖, 0↑ 4↓)
datafusion / parquet (1.053x ➖, 1↑ 1↓)
duckdb / vortex-file-compressed (1.046x ➖, 0↑ 0↓)
duckdb / vortex-compact (1.033x ➖, 0↑ 0↓)
duckdb / parquet (1.035x ➖, 0↑ 0↓)
Full attributed analysis
|
| // NB: We need to execute the constant array to be canonical because there is a | ||
| // reduction rule that turns `Extension(Constant(..))` into `Constant(Extension(..))`, | ||
| // and if we don't do this we create an infinite cycle. | ||
| // See `ExtensionConstantRule` for more details. |
There was a problem hiding this comment.
should we just remove the reduction rule?
There was a problem hiding this comment.
It's kind of hard to say since if we don't remove that reduction rule, all of our optimizations that look for a ConstantArray child now don't work.
There was a problem hiding this comment.
We should keep the rule 100%. We must have a standard way to check is constant.
Benchmarks: Clickbench on NVMEVerdict: No clear signal (low confidence) datafusion / vortex-file-compressed (1.030x ➖, 0↑ 3↓)
datafusion / parquet (1.052x ➖, 0↑ 5↓)
duckdb / vortex-file-compressed (0.997x ➖, 8↑ 2↓)
duckdb / parquet (1.016x ➖, 0↑ 0↓)
duckdb / duckdb (0.974x ➖, 4↑ 0↓)
Full attributed analysis
|
File Sizes: Clickbench on NVMEFile Size Changes (1 files changed, -0.0% overall, 0↑ 1↓)
Totals:
|
Benchmarks: CompressionVortex (geomean): 1.002x ➖ unknown / unknown (1.004x ➖, 0↑ 1↓)
|
Benchmarks: TPC-H SF=10 on S3Verdict: No clear signal (environment too noisy confidence) datafusion / vortex-file-compressed (1.032x ➖, 0↑ 2↓)
datafusion / vortex-compact (1.114x ➖, 0↑ 0↓)
datafusion / parquet (1.017x ➖, 1↑ 1↓)
duckdb / vortex-file-compressed (1.087x ➖, 0↑ 0↓)
duckdb / vortex-compact (1.084x ➖, 0↑ 0↓)
duckdb / parquet (1.071x ➖, 0↑ 1↓)
Full attributed analysis
|
|
i dont really believe the TPC-H SF1 regression is real given SF10 is fine |
joseph-isaacs
left a comment
There was a problem hiding this comment.
We should add a cycle detection mode to executor for debug
Summary
Fixes the regression caused by #7507
The reduction rule would go from
Extension(Constant(..))intoConstant(Extension(..)), but theexecutewould just revert that, creating a cycle!Testing
N/A