wip: add script

Author: adam-peaston-SC

llm-inference-benchmarking/benchmark.sh

@@ -0,0 +1,82 @@
+#!/usr/bin/env bash
+set -euo pipefail
+
+# -----------------------------
+# Configuration
+# -----------------------------
+PORT=8000
+HOST=127.0.0.1
+LOG_FILE="vllm.log"
+DATASET_ID=""
+MODEL="/data/${DATASET_ID}"
+TP_SIZE=4
+GPUS="0,1,2,3"
+
+SERVER_PID=""
+
+# -----------------------------
+# Cleanup handler
+# -----------------------------
+cleanup() {
+  if [[ -n "${SERVER_PID}" ]] && kill -0 "${SERVER_PID}" 2>/dev/null; then
+    echo "Cleaning up vLLM server (PID ${SERVER_PID})..."
+    kill "${SERVER_PID}"
+    wait "${SERVER_PID}" 2>/dev/null || true
+  fi
+}
+
+trap cleanup EXIT INT TERM
+
+# -----------------------------
+# Start server in background
+# -----------------------------
+echo "Starting vLLM server..."
+
+CUDA_VISIBLE_DEVICES=${GPUS} \
+nohup python -m vllm.entrypoints.openai.api_server \
+  --model "${MODEL}" \
+  --tensor-parallel-size "${TP_SIZE}" \
+  --host 0.0.0.0 \
+  --port "${PORT}" \
+  > "${LOG_FILE}" 2>&1 &
+
+SERVER_PID=$!
+echo "vLLM server PID: ${SERVER_PID}"
+
+# -----------------------------
+# Wait for server readiness
+# -----------------------------
+echo "Waiting for server to become ready..."
+
+for i in {1..60}; do
+  if curl -sf "http://${HOST}:${PORT}/v1/models" > /dev/null; then
+    echo "Server is up!"
+    break
+  fi
+  sleep 1
+done
+
+if ! curl -sf "http://${HOST}:${PORT}/v1/models" > /dev/null; then
+  echo "ERROR: Server did not start within timeout"
+  echo "Last 50 log lines:"
+  tail -n 50 "${LOG_FILE}"
+  exit 1
+fi
+
+# -----------------------------
+# Curl a test request
+# -----------------------------
+echo "Sending test completion request..."
+
+curl -s "http://${HOST}:${PORT}/v1/chat/completions" \
+  -H "Content-Type: application/json" \
+  -H "Authorization: Bearer test" \
+  -d "{
+    \"model\": \"${MODEL}\",
+    \"messages\": [
+      {\"role\": \"user\", \"content\": \"Hello from the same machine\"}
+    ]
+  }" | jq .
+
+echo "Done."
+echo "Logs: ${LOG_FILE}"