Updated benchmark runner script

README.md

@@ -108,15 +108,17 @@ Then configure the `--sheet` and `--tab` arguments in benchmark_config.yaml.
 
 ### Running all of the Queries
 
-The included `benchmark_runner.py` script will run all queries sequentially. Configuration for this type of end-to-end run is specified in `benchmark_runner/benchmark_config.yaml`.
+The included `benchmark_runner.sh` script will run all queries sequentially. Configuration for this type of end-to-end run is specified in `benchmark_runner/benchmark_config.yaml`.
+
+First, set `GPU_BDB_HOME` in the bash script to the location of this repository. This is the same environment variable mentioned in the configuration above.
 
 To run all queries, cd to `gpu_bdb/` and:
 
 ```python
-python benchmark_runner.py --config_file benchmark_runner/benchmark_config.yaml
+bash benchmark_runner.sh
 ```
 
-By default, this will run each Dask query once, and, if BlazingSQL queries are enabled in `benchmark_config.yaml`, each BlazingSQL query once. You can control the number of repeats by changing the `N_REPEATS` variable in the script.
+By default, this will run each Dask query once. If BlazingSQL queries are enabled with `INCLUDE_BLAZING` in `benchmark_runner.sh` and in `benchmark_config.yaml`, this will run each BlazingSQL query once. You can control the number of repeats by changing the `N_REPEATS` variable in the script.
 
 
 ## BlazingSQL

gpu_bdb/benchmark_runner.sh

@@ -0,0 +1,42 @@
+#!/bin/bash
+
+USERNAME=$(whoami)
+
+if [ -z "$GPU_BDB_HOME" ]
+then
+    GPU_BDB_HOME=/raid/$USERNAME/prod/gpu-bdb
+else
+    GPU_BDB_HOME=$GPU_BDB_HOME
+fi
+
+INCLUDE_DASK=true
+INCLUDE_BLAZING=false
+N_REPEATS=1
+
+# Dask queries
+if $INCLUDE_DASK; then
+    for qnum in {01..30}
+    do
+        cd $GPU_BDB_HOME/gpu_bdb/queries/q$qnum/
+        for j in $(seq 1 $N_REPEATS)
+        do
+            python gpu_bdb_query_$qnum.py --config_file ../../benchmark_runner/benchmark_config.yaml
+            sleep 3
+        done
+        sleep 3
+    done
+fi
+
+# BlazingSQL Queries
+if $INCLUDE_BLAZING; then
+    for qnum in {01..30}
+    do
+        cd $GPU_BDB_HOME/gpu_bdb/queries/q$qnum/
+        for j in $(seq 1 $N_REPEATS)
+        do
+            python gpu_bdb_query_$qnum\_sql.py --config_file ../../benchmark_runner/benchmark_config.yaml
+            sleep 3
+        done
+        sleep 3
+    done
+fi

gpu_bdb/benchmark_runner/slurm/run_bench.sh

@@ -1,7 +1,6 @@
 set -e pipefail
 
 USERNAME=$(whoami)
-GPU_BDB_HOME=$HOME/gpu-bdb
 LOGDIR=$HOME/dask-local-directory/logs
 STATUS_FILE=${LOGDIR}/status.txt
 
@@ -16,6 +15,8 @@ CONDA_ENV_PATH="/opt/conda/etc/profile.d/conda.sh"
 source $CONDA_ENV_PATH
 conda activate $CONDA_ENV_NAME
 
+export GPU_BDB_HOME=$HOME/gpu-bdb
+
 if [[ "$SLURM_NODEID" -eq 0 ]]; then
     bash $GPU_BDB_HOME/gpu_bdb/cluster_configuration/cluster-startup-slurm.sh SCHEDULER &
     echo "STARTED SCHEDULER"
@@ -29,8 +30,7 @@ if [[ "$SLURM_NODEID" -eq 0 ]]; then
     # echo "Starting load test.."
     # python queries/load_test/gpu_bdb_load_test.py --config_file benchmark_runner/benchmark_config.yaml > $LOGDIR/load_test.log
     echo "Starting E2E run.."
-    python benchmark_runner.py --config_file benchmark_runner/benchmark_config.yaml > $LOGDIR/benchmark_runner.log
-
+    bash benchmark_runner.sh
     echo "FINISHED" > ${STATUS_FILE}
 else
     sleep 15 # Sleep and wait for the scheduler to spin up