Example script to run tpc-h queries with query-specific parameters.

run_tpch_benchmarks.sh

#!/bin/bash

# Run TPC-H benchmarks for queries 1-22 sequentially
# Usage: bash ./run_tpch_benchmarks.sh > summary.out 2>&1 &

export CUDA_VISIBLE_DEVICES=3
ITERATIONS=3

# Configuration (Viking)
PDSH_FILE=/raid/rzamora/benchmarking_rapidsmpf/cudf/python/cudf_polars/cudf_polars/experimental/benchmarks/pdsh.py
DATASET_PATH=/raid/rapidsmpf/data/tpch/scale-1000

# Configuration (PDX)
# PDSH_FILE=/home/rapids/src/cudf/python/cudf_polars/cudf_polars/experimental/benchmarks/pdsh.py
# DATASET_PATH=/data/tpch-rs/scale-1000

# Default values
DEFAULT_BLOCKSIZE=5_000_000_000
DEFAULT_SPILL_DEVICE=0.9
DEFAULT_MAX_IO_THREADS=8

# Per-query settings (only specify non-default values)
declare -A QUERY_BLOCKSIZE
QUERY_BLOCKSIZE[1]=5_000_000_000
QUERY_BLOCKSIZE[4]=1_000_000_000
QUERY_BLOCKSIZE[13]=1_000_000_000
QUERY_BLOCKSIZE[18]=1_000_000_000
QUERY_BLOCKSIZE[21]=1_000_000_000

declare -A QUERY_SPILL_DEVICE
QUERY_SPILL_DEVICE[1]=0.9
QUERY_SPILL_DEVICE[4]=0.8
QUERY_SPILL_DEVICE[13]=0.8
QUERY_SPILL_DEVICE[18]=0.85
QUERY_SPILL_DEVICE[21]=0.4

declare -A QUERY_MAX_IO_THREADS
QUERY_MAX_IO_THREADS[1]=4
QUERY_MAX_IO_THREADS[4]=4
QUERY_MAX_IO_THREADS[13]=4
QUERY_MAX_IO_THREADS[18]=4
QUERY_MAX_IO_THREADS[21]=4


for QUERY_NUM in {1..22}; do
    # Get query-specific settings or use defaults
    BLOCKSIZE=${QUERY_BLOCKSIZE[$QUERY_NUM]:-$DEFAULT_BLOCKSIZE}
    SPILL_DEVICE=${QUERY_SPILL_DEVICE[$QUERY_NUM]:-$DEFAULT_SPILL_DEVICE}
    MAX_IO_THREADS=${QUERY_MAX_IO_THREADS[$QUERY_NUM]:-$DEFAULT_MAX_IO_THREADS}

    echo "Starting Query ${QUERY_NUM} (blocksize=${BLOCKSIZE}, spill-device=${SPILL_DEVICE})..."
    python $PDSH_FILE $QUERY_NUM \
        --iterations ${ITERATIONS} \
        --path ${DATASET_PATH} \
        --suffix '' \
        --spill-device ${SPILL_DEVICE} \
        --blocksize ${BLOCKSIZE} \
        --max-io-threads ${MAX_IO_THREADS} \
        --rmm-async \
        --explain \
        --runtime rapidsmpf > "output_q${QUERY_NUM}.out" 2>&1

    exit_code=$?
    if [ $exit_code -eq 0 ]; then
        echo "Query ${QUERY_NUM} completed successfully"
    else
        echo "Query ${QUERY_NUM} failed with exit code ${exit_code}"
    fi
    echo "---"
done

echo "All benchmarks completed!"