Async Loading

Load data 4-5x faster with parallel async operations

Overview

The tif1 library provides async methods for loading data in parallel, offering 4-5x speedup compared to synchronous loading. This is especially beneficial when loading data with a cold cache.

Why Use Async Loading?

Async loading provides significant performance benefits:

• 4-5x faster data loading with cold cache
• Parallel fetching of multiple data sources
• Non-blocking operations for better resource utilization
• Efficient HTTP/2 multiplexing for concurrent requests

The speedup comes from fetching multiple data files (lap data, telemetry, weather, etc.) in parallel rather than sequentially.

Basic Async Loading

Load session data asynchronously:

import asyncio
import tif1

async def load_session_async():
    """Load session data asynchronously."""
    session = tif1.get_session(2025, "Abu Dhabi Grand Prix", "Practice 1")
    
    # Load laps asynchronously
    laps = await session.laps_async()
    print(f"Loaded {len(laps)} laps")
    
    return laps

# Run the async function
laps = asyncio.run(load_session_async())

Performance Comparison

Compare sync vs async loading:

import asyncio
import time
import tif1

async def load_data_async():
    """Load data asynchronously (4-5x faster)."""
    session = tif1.get_session(2025, "Abu Dhabi Grand Prix", "Practice 1")
    
    start = time.time()
    laps = await session.laps_async()
    elapsed = time.time() - start
    
    print(f"✓ Async: Loaded {len(laps)} laps in {elapsed:.2f}s")
    return laps, elapsed

def load_data_sync():
    """Load data synchronously."""
    session = tif1.get_session(2025, "Abu Dhabi Grand Prix", "Practice 1")
    
    start = time.time()
    laps = session.laps
    elapsed = time.time() - start
    
    print(f"✓ Sync:  Loaded {len(laps)} laps in {elapsed:.2f}s")
    return laps, elapsed

# Clear cache for fair comparison
cache = tif1.get_cache()
cache.clear()

print("ASYNC vs SYNC LOADING COMPARISON")
print("=" * 60)

# Test async loading
print("\nTesting async loading...")
laps_async, time_async = asyncio.run(load_data_async())

# Clear cache again
cache.clear()

# Test sync loading
print("\nTesting sync loading...")
laps_sync, time_sync = load_data_sync()

# Show comparison
print("\n" + "=" * 60)
print("RESULTS")
print("=" * 60)
speedup = time_sync / time_async if time_async > 0 else 0
print(f"Async time: {time_async:.2f}s")
print(f"Sync time:  {time_sync:.2f}s")
print(f"Speedup:    {speedup:.1f}x faster with async")

Typical output:

Async time: 0.82s
Sync time:  3.45s
Speedup:    4.2x faster with async

Loading Multiple Sessions

Load multiple sessions in parallel for maximum performance:

import asyncio
import time
import tif1

async def load_multiple_sessions():
    """Load multiple sessions in parallel."""
    sessions_to_load = [
        (2025, "Monaco Grand Prix", "Qualifying"),
        (2025, "Monaco Grand Prix", "Race"),
        (2025, "British Grand Prix", "Qualifying"),
    ]
    
    async def load_one_session(year, event, session_type):
        session = tif1.get_session(year, event, session_type)
        laps = await session.laps_async()
        return (event, session_type, len(laps))
    
    # Load all sessions in parallel
    start = time.time()
    tasks = [
        load_one_session(year, event, session_type)
        for year, event, session_type in sessions_to_load
    ]
    results = await asyncio.gather(*tasks)
    elapsed = time.time() - start
    
    print(f"\nLoaded {len(results)} sessions in {elapsed:.2f}s:")
    for event, session_type, lap_count in results:
        print(f"  {event} {session_type}: {lap_count} laps")
    
    return results, elapsed

# Run
results, elapsed = asyncio.run(load_multiple_sessions())

Async Telemetry Loading

The fastest lap telemetry methods use async loading internally:

import tif1
import time

session = tif1.get_session(2025, "Monaco Grand Prix", "Race")

print("ASYNC TELEMETRY LOADING")
print("=" * 60)

# Single fastest lap telemetry (uses async internally)
start = time.time()
fastest_tel = session.get_fastest_lap_tel()
elapsed = time.time() - start
print(f"\n1. Single fastest lap: {elapsed:.2f}s")
print(f"   Points: {len(fastest_tel)}")

# Multiple drivers' fastest lap telemetry (parallel!)
start = time.time()
all_fastest_tels = session.get_fastest_laps_tels(by_driver=True)
elapsed = time.time() - start
num_drivers = len(all_fastest_tels["Driver"].unique())
print(f"\n2. All drivers' fastest laps: {elapsed:.2f}s")
print(f"   Drivers: {num_drivers}")
print(f"   Average: {elapsed / num_drivers:.3f}s per driver")
print(f"   Total points: {len(all_fastest_tels):,}")

# Specific drivers' telemetry (parallel)
start = time.time()
top3_tels = session.get_fastest_laps_tels(
    by_driver=True,
    drivers=["VER", "HAM", "LEC"]
)
elapsed = time.time() - start
print(f"\n3. Top 3 drivers: {elapsed:.2f}s")
print(f"   Drivers: {list(top3_tels['Driver'].unique())}")

Async Context Manager

Use async context managers for cleaner code:

import asyncio
import tif1

class AsyncSessionLoader:
    """Async context manager for session loading."""
    
    def __init__(self, year, event, session_type):
        self.year = year
        self.event = event
        self.session_type = session_type
        self.session = None
        self.laps = None
    
    async def __aenter__(self):
        """Load session on enter."""
        self.session = tif1.get_session(
            self.year,
            self.event,
            self.session_type
        )
        self.laps = await self.session.laps_async()
        return self
    
    async def __aexit__(self, exc_type, exc_val, exc_tb):
        """Cleanup on exit."""
        # Cleanup if needed
        pass

async def analyze_with_context():
    """Use async context manager."""
    async with AsyncSessionLoader(
        2025,
        "Monaco Grand Prix",
        "Race"
    ) as loader:
        print(f"Loaded {len(loader.laps)} laps")
        print(f"Drivers: {len(loader.session.drivers_df)}")
        return loader.laps

laps = asyncio.run(analyze_with_context())

Async Error Handling

Handle errors in async context:

import asyncio
import tif1

async def safe_async_load(year, event, session_type):
    """Load session with error handling."""
    try:
        session = tif1.get_session(year, event, session_type)
        laps = await session.laps_async()
        print(f"✓ Loaded {len(laps)} laps")
        return laps
    except tif1.NetworkError as e:
        print(f"✗ Network error: {e}")
        return None
    except tif1.DataNotFoundError as e:
        print(f"✗ Data not found: {e}")
        return None
    except tif1.TIF1Error as e:
        print(f"✗ Error: {e}")
        return None

# Load multiple sessions with error handling
async def load_multiple_safe():
    """Load multiple sessions safely."""
    sessions = [
        (2025, "Monaco Grand Prix", "Race"),
        (2025, "British Grand Prix", "Qualifying"),
        (2025, "Invalid GP", "Race"),  # This will fail
    ]
    
    tasks = [
        safe_async_load(year, event, session_type)
        for year, event, session_type in sessions
    ]
    
    results = await asyncio.gather(*tasks)
    
    # Filter out None results
    valid_results = [r for r in results if r is not None]
    print(f"\nSuccessfully loaded {len(valid_results)}/{len(sessions)} sessions")
    
    return valid_results

results = asyncio.run(load_multiple_safe())

Async with Timeout

Set timeouts for async operations:

import asyncio
import tif1

async def load_with_timeout(year, event, session_type, timeout=30):
    """Load session with timeout."""
    try:
        session = tif1.get_session(year, event, session_type)
        laps = await asyncio.wait_for(
            session.laps_async(),
            timeout=timeout
        )
        print(f"✓ Loaded {len(laps)} laps within {timeout}s")
        return laps
    except asyncio.TimeoutError:
        print(f"✗ Timeout: Loading took longer than {timeout}s")
        return None
    except tif1.TIF1Error as e:
        print(f"✗ Error: {e}")
        return None

laps = asyncio.run(load_with_timeout(
    2025,
    "Monaco Grand Prix",
    "Race",
    timeout=10
))

Concurrent Request Limits

Configure maximum concurrent requests:

# Set max concurrent requests via environment variable
export TIF1_MAX_CONCURRENT_REQUESTS=32
python your_script.py

Or via configuration:

{
  "max_concurrent_requests": 32,
  "telemetry_prefetch_max_concurrent_requests": 64
}

Best Practices

When to Use Async

1. Cold cache: Maximum benefit when data isn’t cached
2. Multiple sessions: Loading multiple sessions in parallel
3. Large datasets: Sessions with many drivers/laps
4. Batch operations: Processing multiple events at once

When Sync Is Fine

1. Warm cache: Data already cached (minimal difference)
2. Simple scripts: Single session, interactive use
3. Learning/debugging: Easier to understand and debug
4. Small datasets: Practice sessions with few laps

Example: Smart Loading

import asyncio
import tif1

def smart_load_session(year, event, session_type):
    """Choose between sync and async based on cache."""
    cache = tif1.get_cache()
    
    # Check if data is likely cached (simplified check)
    # In production, you'd check specific cache keys
    
    session = tif1.get_session(year, event, session_type)
    
    # For now, always use async for cold cache benefit
    # You can add cache checking logic here
    try:
        # Try to use async event loop if available
        loop = asyncio.get_event_loop()
        if loop.is_running():
            # Already in async context
            return session.laps
        else:
            # Use async loading
            return asyncio.run(session.laps_async())
    except RuntimeError:
        # Fallback to sync if async not available
        return session.laps

laps = smart_load_session(2025, "Monaco Grand Prix", "Race")
print(f"Loaded {len(laps)} laps")

Complete Async Example

import asyncio
import time
import tif1

async def comprehensive_async_analysis():
    """Comprehensive async loading example."""
    print("COMPREHENSIVE ASYNC LOADING")
    print("=" * 60)
    
    # 1. Load single session
    print("\n1. Single Session (Async)")
    session = tif1.get_session(2025, "Monaco Grand Prix", "Race")
    start = time.time()
    laps = await session.laps_async()
    elapsed = time.time() - start
    print(f"   Loaded {len(laps)} laps in {elapsed:.2f}s")
    
    # 2. Load multiple sessions in parallel
    print("\n2. Multiple Sessions (Parallel)")
    
    async def load_session(year, event, session_type):
        session = tif1.get_session(year, event, session_type)
        laps = await session.laps_async()
        return (event, session_type, len(laps))
    
    sessions = [
        (2025, "Monaco Grand Prix", "Qualifying"),
        (2025, "British Grand Prix", "Qualifying"),
        (2025, "Silverstone Grand Prix", "Race"),
    ]
    
    start = time.time()
    tasks = [load_session(*s) for s in sessions]
    results = await asyncio.gather(*tasks)
    elapsed = time.time() - start
    
    print(f"   Loaded {len(results)} sessions in {elapsed:.2f}s:")
    for event, session_type, lap_count in results:
        print(f"     • {event} {session_type}: {lap_count} laps")
    
    # 3. Parallel telemetry loading
    print("\n3. Parallel Telemetry Loading")
    session = tif1.get_session(2025, "Monaco Grand Prix", "Race")
    start = time.time()
    all_fastest_tels = session.get_fastest_laps_tels(by_driver=True)
    elapsed = time.time() - start
    num_drivers = len(all_fastest_tels["Driver"].unique())
    print(f"   Loaded {num_drivers} drivers' telemetry in {elapsed:.2f}s")
    print(f"   Average: {elapsed / num_drivers:.3f}s per driver")

# Clear cache for demonstration
cache = tif1.get_cache()
cache.clear()

# Run the async analysis
asyncio.run(comprehensive_async_analysis())

Next Steps

• Learn about caching to maximize async performance
• Explore configuration for async settings
• Understand error handling in async context