run-01

#!/bin/bash
set -eu

NODES=1
CORES_PER_NODE=16
PARTITIONS=$(( ${NODES} * ${CORES_PER_NODE} * 2 ))

echo "Nodes: ${NODES}"
echo "Cores/node: ${CORES_PER_NODE}"
mkdir -p logs
sleep 5

for MULTIPLIER in `seq 10 10 130`; do
        N=$(( ${NODES} * ${CORES_PER_NODE} * ${MULTIPLIER} * 1000000 ))
        OUTPUT="logs/$(basename $0)-${N}.log"
	rm -f ${OUTPUT}
        echo "Number of trials: ${N}" | tee -a ${OUTPUT}
	echo "Data Size (MB): $(( ${N} * 8 / 1048576 ))" | tee -a ${OUTPUT}
        /usr/bin/time -f "TIMING: n=${N}, op=totalExecution, duration=%e" spark-submit --master yarn --deploy-mode client \
          --class com.cloudera.datascience.montecarlorisk.MonteCarloRisk \
          --conf "spark.dynamicAllocation.enabled=false" \
          --conf "spark.kryoserializer.buffer.max=512m" \
          --conf "spark.driver.maxResultSize=12576m" \
          --conf "spark.network.timeout=300s" \
          --driver-memory=20g \
          --files data/instruments.csv,data/covariances.csv,data/means.csv \
          target/montecarlo-risk-0.0.1-SNAPSHOT.jar \
          data/instruments.csv ${N} ${PARTITIONS} data/means.csv data/covariances.csv \
          2>&1 | tee -a ${OUTPUT}
done