Tuning Commands¶
Commands for creating and managing tuning configurations for SQL and DataFrame platforms.
tuning init - Create Tuning Configuration¶
Generate sample unified tuning configurations for specific platforms.
Options¶
--platform TEXT: Target platform (required) - duckdb, databricks, snowflake, etc.--output TEXT: Output file path (default: tuning_config.yaml)
Usage Examples¶
# Create sample tuning for Databricks
benchbox tuning init --platform databricks
# Create with custom output path
benchbox tuning init --platform snowflake \
--output ./configs/snowflake-tuning.yaml
DataFrame Tuning via tuning Commands¶
The unified tuning command group handles both SQL and DataFrame platforms.
Use --mode dataframe (or let it auto-detect from the platform name).
Create DataFrame Tuning Configuration¶
# Create sample Polars tuning config (auto-detects DataFrame mode)
benchbox tuning init --platform polars
# Create with smart defaults based on your system
benchbox tuning init --platform polars --smart-defaults
# Explicit DataFrame mode with profile
benchbox tuning init --platform pandas --mode dataframe --profile memory-constrained
# Custom output path
benchbox tuning init --platform dask --output ./configs/dask_tuning.yaml
Options for DataFrame mode:
--platform TEXT: Target platform (polars,pandas,dask,modin,cudf)--mode [sql|dataframe|auto]: Tuning mode (default: auto)--profile [default|optimized|streaming|memory-constrained|gpu]: Configuration profile--output TEXT: Output file path--smart-defaults: Use auto-detected system-optimal settings
Validate Configuration¶
benchbox tuning validate polars_tuning.yaml --platform polars
Show Smart Defaults¶
benchbox tuning defaults --platform polars
Configuration Categories¶
DataFrame tuning supports these configuration sections:
Category |
Settings |
Description |
|---|---|---|
|
|
CPU resource allocation |
|
|
Memory management |
|
|
Execution behavior |
|
|
Type handling |
|
|
I/O optimization |
|
|
GPU settings (cuDF) |
|
|
Write-time physical layout and Parquet encoding |
For complete configuration reference, see DataFrame Platforms - Tuning.