Requests to triton server with ensemble model

encode.sh

#!/bin/bash

# Define variables
TRITON_URL="localhost:8000"
MODEL_NAME="executor_model"
MODEL_VERSION="1"

# Create JSON payload
PAYLOAD=$(cat <<EOF
{
  "inputs": [
    {
        "name": "col1",
        "shape": [1],
        "datatype": "INT64",
        "data": [132131]
      },
      {
        "name": "col2",
        "shape": [1],
        "datatype": "INT64",
        "data": [4]
      }
  ]
}
EOF
)

# Send curl request
make_request() {
    curl -X POST "http://${TRITON_URL}/v2/models/${MODEL_NAME}/versions/${MODEL_VERSION}/infer" \
         -H "Content-Type: application/json" \
         -d "$PAYLOAD"
}

# Function to measure execution time
measure_time() {
    start_time=$(date +%s.%N)
    "$@"
    end_time=$(date +%s.%N)
    duration=$(echo "$end_time - $start_time" | bc)
    echo "Execution time: $duration seconds"
}


# Run multiple times:
total_time=0
runs=50
for i in $(seq 1 $runs); do
    #echo "Run $i:"
    start_time=$(date +%s%N)
    make_request #> /dev/null  # Suppress output 
    end_time=$(date +%s%N)
    duration_ns=$((end_time - start_time))
    duration=$((duration_ns / 1000000))
    total_time=$((total_time + duration))
    #echo "Execution time: $duration ms"
    echo ""
done
average_time=$((total_time / runs))
echo "Average execution time over $runs runs: $average_time ms"