run_perf.sh

Go to the documentation of this file.

#!/bin/bash
# ===========================================
# run_perf.sh
# ===========================================
 

 
 
ROOT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")/.." && pwd)"
export PYTHONPATH="$ROOT_DIR:$PYTHONPATH"
 
set -e
 
# -------- Config --------
DEFAULT_TRIALS=100000000
ALL_METHODS=("Sequential" "Heap" "Pool" "SIMD")
SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
BATCHID=$(uuidgen | cut -d'-' -f1)
BUILD_PATH="./build/montecarlo"
GLOBAL_TIMESTAMP=$(date "+%Y-%m-%d_%H-%M-%S")
 
# -------- CLI Args --------
ARG1="$1"
ARG2="$2"
$ARG3=$3
 
INSERT_DB=true
 
if [[ "$ARG3" == "insert_db=false" ]]; then
    INSERT_DB=false
fi
 
if [[ -z "$ARG1" && -z "$ARG2" ]]; then
    TRIALS=$DEFAULT_TRIALS
    METHODS=("${ALL_METHODS[@]}")
elif [[ "$ARG1" =~ ^[0-9]+$ && -z "$ARG2" ]]; then
    TRIALS="$ARG1"
    METHODS=("${ALL_METHODS[@]}")
elif [[ "$ARG1" =~ ^[a-zA-Z]+$ && -z "$ARG2" ]]; then
    TRIALS=$DEFAULT_TRIALS
    METHODS=("$ARG1")
else
    TRIALS="$ARG1"
    METHODS=("$ARG2")
fi
 
# -------- Info --------
echo "[INFO] Trials   : $TRIALS"
echo "[INFO] Methods  : ${METHODS[*]}"
echo "[INFO] Batch ID : $BATCHID"
echo "[INFO] Timestamp: $GLOBAL_TIMESTAMP"
 
LOG_DIR="db/logs/batch_${BATCHID}_${GLOBAL_TIMESTAMP}"
mkdir -p "$LOG_DIR"
 
# -------- Perf Event Creation --------
PERF_EVENTS=""
PERF_EVENTS+="cycles,instructions,"
PERF_EVENTS+="cache-references,cache-misses,"
PERF_EVENTS+="branch-instructions,branch-misses,"
PERF_EVENTS+="L1-dcache-loads,L1-dcache-load-misses,"
PERF_EVENTS+="dTLB-loads,dTLB-load-misses"
 
echo "[INFO] Using perf events:"
echo "$PERF_EVENTS" | tr ',' '\n' | sed 's/^/  - /'
 
# -------- Run Each Method --------
for METHOD in "${METHODS[@]}"; do
    METHOD_TIMESTAMP=$(date +"%Y-%m-%d %H:%M:%S")
    echo "[▶] Running: $METHOD"
 
    LOG_PATH="$LOG_DIR/perf_${METHOD}_${METHOD_TIMESTAMP}.csv"
    PERF_PARQUET="$LOG_DIR/perf_results_${METHOD}_${METHOD_TIMESTAMP}_${BATCHID}.parquet"
 
    mkdir -p "$(dirname "$LOG_PATH")"
 
    START_NS=$(date +%s%N)
    perf stat -x, -o "$LOG_PATH" -e $PERF_EVENTS "$BUILD_PATH" "$TRIALS" "$METHOD" > /dev/null
    END=$(date +%s%N)
 
    WALL_NS=$((END - START_NS))
    WALL_S=$(awk "BEGIN {printf \"%.6f\", $WALL_NS / 1000000000}")
 
    # Pull metrics into shell vars
    eval "$(python3 pipeline/parse_perf_metrics.py "$LOG_PATH" "$TRIALS")"
 
    python3 pipeline/gen_perf_parquet_logs.py \
        --out_path "$PERF_PARQUET" \
        --wall_time_s "$WALL_S" \
        --wall_time_ns "$WALL_NS" \
        --timestamp "$METHOD_TIMESTAMP" \
        --batchid "$BATCHID" \
        --method "$METHOD" \
        --trials "$TRIALS" \
        --cycles "$CYCLES" \
        --instr "$INSTR" \
        --ipc "$IPC" \
        --cache_loads "$CACHE_LOADS" \
        --cache_miss "$CACHE_MISS" \
        --l1_loads "$L1_LOADS" \
        --l1_misses "$L1_MISSES" \
        --l2_loads "$L2_LOADS" \
        --l2_misses "$L2_MISSES" \
        --l3_loads "$L3_LOADS" \
        --l3_misses "$L3_MISSES" \
        --tlb_loads "$TLB_LOADS" \
        --tlb_misses "$TLB_MISSES" \
        --branch_instr "$BRANCH_INSTR" \
        --branch_misses "$BRANCH_MISSES" \
        --miss_per_trial "$MISS_PER_TRIAL" \
        --cycles_per_trial "$CYCLES_PER_TRIAL"
done
 
python3 pipeline/combine_batch_parquets.py \
  "$LOG_DIR" \
  "$LOG_DIR/perf_results_all_${BATCHID}.parquet"
 
if [ "$INSERT_DB" = true ]; then
    python3 pipeline/insert_to_clickhouse.py \
    --batchid "$BATCHID"
else
  echo "[INFO] Skipping ClickHouse insertion (insert_db=false)"
fi
 
echo "[INFO] Simulation Finished:"
echo "     └─ Exported CSV & Parquet logs to  : $LOG_DIR"
echo "     └─ Combined batch Parquet logs  : $LOG_DIR/perf_results_all_${BATCHID}.parquet"