Computer_Architecture_HW

This repository contains scripts and outputs related to VLLM service performance testing and usability validation.

Files

• start_server.sh: Shell script to start the VLLM service.
• hf_test.py: Python script for testing performance using method 1.
• vllm_test.py: Python script for testing performance using method 2.
• test.sh: Shell script to verify VLLM service availability.
• inference_stats.txt: Output file containing inference performance statistics.
• vllm_inference_stats.txt: Output file with additional VLLM inference statistics.

Usage

1. Start the VLLM Service:

   bash start_server.sh

2. Check VLLM Usability:

   bash test.sh
   ---

3. Run Performance Tests:

   • Method 1:

     python hf_test.py

   • Method 2:

     python vllm_test.py

4. Outputs:

   • inference_stats.txt and vlm_inference_stats.txt contain performance results.

---

Prerequisites

• Python 3.x
• vLLM 0.6.3.post1

---

Author

• Zh1yuShen