Skip to content

quickstart.md

Latest commit

 

History

History
154 lines (110 loc) · 3.18 KB

quickstart.md

File metadata and controls

154 lines (110 loc) · 3.18 KB

Quick Start Guide

This guide will help you get up and running with InfiniMetrics in minutes.

Prerequisites

Before you begin, ensure you have:

  • Python 3.8+ installed
  • Git installed (for cloning the repository)
  • GPU (optional, for hardware tests)

Installation

1. Clone the Repository

git clone https://github.com/InfiniTensor/InfiniMetrics.git
cd InfiniMetrics

2. Initialize Submodules

git submodule update --init --recursive

3. Install Dependencies

# Core dependencies
pip install numpy torch pandas

# Optional: For vLLM support
pip install vllm

4. Build Hardware Benchmarks (Optional)

If you plan to run hardware tests:

cd infinimetrics/hardware/cuda-memory-benchmark
bash build.sh

Run Your First Test

Hardware Test

Run a comprehensive hardware benchmark with a single command:

python main.py format_input_comprehensive_hardware.json

This will test:

  • Memory bandwidth (H2D, D2H, D2D, bidirectional)
  • STREAM benchmark
  • GPU cache performance

Inference Test

Run an inference benchmark:

cd infinimetrics/inference
python infer_main.py --config config.json

Understanding the Output

Test Results Location

Results are saved in:

./summary_output/
└── dispatcher_summary_YYYYMMDD_HHMMSS.json    # Overall summary
./output/
└── <test_case_name>/
    ├── seperated_test_result.json              # Seperated test result
    └── metrics.csv                             # Timeseries metrics

Example: Hardware Test Output

{
  "total_tests": 1,
  "successful_tests": 1,
  "failed_tests": 0,
  "results": [{
    "run_id": "test_001",
    "testcase": "hardware.cudaUnified.Comprehensive",
    "result_code": 0,
    "result_file": "./output/hardware.cudaUnified.Comprehensive/metrics.json"
  }]
}

MongoDB Integration (Optional)

InfiniMetrics supports storing test results in MongoDB for persistent storage and dashboard visualization.

Prerequisites

# Install MongoDB dependencies
pip install pymongo watchdog

# Ensure MongoDB is running locally or set connection URL
export MONGO_URI="mongodb://localhost:27017"

CLI Usage

# Start file watcher (auto-import new results, runs forever)
python -m db.watcher --output-dir ./output --summary-dir ./summary_output

# One-time scan only (import existing files and exit)
python -m db.watcher --scan

Python API

from pathlib import Path
from db.watcher import Watcher

# Create watcher and start monitoring
watcher = Watcher(
    output_dir=Path("./output"),
    summary_dir=Path("./summary_output")
)

# One-time scan
result = watcher.scan()
print(f"Imported: {len(result['imported'])}")

# Or run forever (auto-import new files)
watcher.run_forever()  # Blocks until Ctrl+C

Environment Variables

Variable Default Description
MONGO_URI mongodb://localhost:27017 MongoDB connection URL
MONGO_DB_NAME infinimetrics Database name
MONGO_COLLECTION test_runs Test results collection

Next Steps