Conversation
Signed-off-by: Adam Gutglick <adam@spiraldb.com>
Signed-off-by: Adam Gutglick <adam@spiraldb.com>
Signed-off-by: Adam Gutglick <adam@spiraldb.com>
Polar Signals Profiling ResultsLatest Run
Previous Runs (5)
Powered by Polar Signals Cloud |
Benchmarks: PolarSignals ProfilingVortex (geomean): 1.744x ❌ datafusion / vortex-file-compressed (1.744x ❌, 1↑ 6↓)
|
Benchmarks: TPC-H SF=1 on NVMEVerdict: No clear signal (environment too noisy confidence) datafusion / vortex-file-compressed (1.086x ➖, 0↑ 10↓)
datafusion / vortex-compact (1.106x ❌, 0↑ 7↓)
datafusion / parquet (1.031x ➖, 2↑ 4↓)
datafusion / arrow (1.066x ➖, 0↑ 4↓)
duckdb / vortex-file-compressed (1.151x ❌, 0↑ 10↓)
duckdb / vortex-compact (1.118x ❌, 0↑ 8↓)
duckdb / parquet (1.089x ➖, 2↑ 8↓)
duckdb / duckdb (1.014x ➖, 0↑ 2↓)
Full attributed analysis
|
Benchmarks: FineWeb NVMeVerdict: No clear signal (medium confidence) datafusion / vortex-file-compressed (0.730x ✅, 7↑ 1↓)
datafusion / vortex-compact (0.639x ✅, 7↑ 2↓)
datafusion / parquet (0.911x ➖, 2↑ 0↓)
duckdb / vortex-file-compressed (1.032x ➖, 3↑ 4↓)
duckdb / vortex-compact (0.958x ➖, 7↑ 2↓)
duckdb / parquet (0.947x ➖, 0↑ 0↓)
Full attributed analysis
|
Benchmarks: TPC-DS SF=1 on NVMEVerdict: No clear signal (low confidence) datafusion / vortex-file-compressed (0.932x ➖, 17↑ 0↓)
datafusion / vortex-compact (0.958x ➖, 9↑ 0↓)
datafusion / parquet (0.947x ➖, 5↑ 0↓)
duckdb / vortex-file-compressed (0.948x ➖, 23↑ 8↓)
duckdb / vortex-compact (0.975x ➖, 8↑ 7↓)
duckdb / parquet (0.956x ➖, 6↑ 0↓)
duckdb / duckdb (0.937x ➖, 16↑ 0↓)
Full attributed analysis
|
Benchmarks: TPC-H SF=10 on NVMEVerdict: No clear signal (low confidence) datafusion / vortex-file-compressed (1.030x ➖, 0↑ 0↓)
datafusion / vortex-compact (1.020x ➖, 1↑ 3↓)
datafusion / parquet (0.958x ➖, 7↑ 1↓)
datafusion / arrow (1.104x ❌, 0↑ 12↓)
duckdb / vortex-file-compressed (1.099x ➖, 0↑ 10↓)
duckdb / vortex-compact (1.095x ➖, 0↑ 8↓)
duckdb / parquet (1.022x ➖, 0↑ 0↓)
duckdb / duckdb (1.028x ➖, 0↑ 0↓)
Full attributed analysis
|
Benchmarks: Clickbench on NVMEVerdict: No clear signal (low confidence) datafusion / vortex-file-compressed (0.944x ➖, 8↑ 1↓)
datafusion / parquet (0.937x ➖, 5↑ 1↓)
duckdb / vortex-file-compressed (0.884x ✅, 21↑ 2↓)
duckdb / parquet (0.949x ➖, 1↑ 0↓)
duckdb / duckdb (0.970x ➖, 5↑ 0↓)
Full attributed analysis
|
Signed-off-by: Adam Gutglick <adam@spiraldb.com>
Benchmarks: FineWeb S3Verdict: No clear signal (low confidence) datafusion / vortex-file-compressed (0.695x ✅, 2↑ 0↓)
datafusion / vortex-compact (0.669x ✅, 3↑ 0↓)
datafusion / parquet (0.957x ➖, 0↑ 0↓)
duckdb / vortex-file-compressed (0.970x ➖, 1↑ 1↓)
duckdb / vortex-compact (0.956x ➖, 1↑ 1↓)
duckdb / parquet (0.961x ➖, 0↑ 0↓)
Full attributed analysis
|
Benchmarks: TPC-H SF=10 on S3Verdict: No clear signal (environment too noisy confidence) datafusion / vortex-file-compressed (1.171x ➖, 0↑ 5↓)
datafusion / vortex-compact (1.069x ➖, 0↑ 2↓)
datafusion / parquet (0.904x ➖, 0↑ 0↓)
duckdb / vortex-file-compressed (1.021x ➖, 0↑ 0↓)
duckdb / vortex-compact (1.044x ➖, 1↑ 1↓)
duckdb / parquet (0.895x ➖, 0↑ 0↓)
Full attributed analysis
|
|
Do you think the vortex-compact improvements are from more parallelism in decompression? |
Benchmarks: Statistical and Population GeneticsVerdict: No clear signal (low confidence) duckdb / vortex-file-compressed (0.987x ➖, 1↑ 0↓)
duckdb / vortex-compact (0.980x ➖, 1↑ 1↓)
duckdb / parquet (0.990x ➖, 0↑ 0↓)
Full attributed analysis
|
Benchmarks: TPC-H SF=1 on S3Verdict: No clear signal (environment too noisy confidence) datafusion / vortex-file-compressed (0.890x ➖, 3↑ 2↓)
datafusion / vortex-compact (0.880x ➖, 2↑ 2↓)
datafusion / parquet (0.925x ➖, 0↑ 0↓)
duckdb / vortex-file-compressed (0.965x ➖, 3↑ 2↓)
duckdb / vortex-compact (1.042x ➖, 0↑ 1↓)
duckdb / parquet (0.914x ➖, 0↑ 0↓)
Full attributed analysis
|
Signed-off-by: Adam Gutglick <adam@spiraldb.com>
This is a mostly codex-driven experiment to play around with some general ideas I had and never got around to diving into, at least locally I've reached a point where perf looks interesting enough that I want to actually run benchmarks.
While I wat for benchmarks to run, some initial numbers from f3c23a6, running clickbench against develop (using GNU time):