Installing Engines
From the Engine Library
Open Engines Page
Navigate to the Engines section from the sidebar, or press
Cmd+, / Ctrl+, for settings and select Engines.Browse Available Engines
The “Web” tab shows curated engines compatible with your system:
- Automatically filtered for your OS (Windows/Mac/Linux)
- CPU compatibility checked (BMI2 support)
Choose an Engine
Popular options include:
- Stockfish: The strongest open-source engine
- Komodo Dragon: Strong commercial engine (if available)
- Specialty engines for specific styles
Installing Local Engines
If you have a chess engine binary:Select Engine Binary
- Browse to your engine executable
- Windows:
.exefile - Mac/Linux: executable file (no extension usually)
Configure Details
- Name: Give your engine a recognizable name
- Version: Optional version information
- En Croissant will auto-detect engine capabilities
En Croissant supports UCI (Universal Chess Interface) engines. Most modern chess engines use this protocol.
Engine Parameters
Analysis Modes
Choose how long the engine analyzes each position:Depth Mode
Depth Mode
Analyzes to a specific depth (number of moves ahead).Use when:
- You want consistent analysis depth across all positions
- Comparing multiple engines fairly
- Working with tactical puzzles
- Quick analysis: 20
- Standard analysis: 25-30
- Deep analysis: 35-40
- Maximum: 45+ (very slow)
Time Mode
Time Mode
Analyzes for a fixed duration per position.Use when:
- You need consistent analysis speed
- Creating time-limited reports
- Resource-constrained systems
- Quick: 1-3 seconds
- Standard: 5-10 seconds
- Deep: 15-30 seconds
- Thorough: 60+ seconds
Nodes Mode
Nodes Mode
Analyzes a specific number of positions.Use when:
- Precise control over computation
- Benchmarking engine performance
- Research or advanced analysis
- Quick: 1M nodes
- Standard: 10M nodes
- Deep: 100M nodes
Infinite Mode
Infinite Mode
Analyzes continuously until manually stopped.Use when:
- Exploring complex positions
- Waiting for evaluation to stabilize
- Finding hidden tactics
Number of Lines (MultiPV)
Controls how many best moves the engine shows:- 1 line: Show only the engine’s top choice (fastest)
- 2-3 lines: Show top alternatives (recommended)
- 4-5 lines: Show many options (good for studying tactics)
- 6+ lines: Rare use; significantly slower
Each additional line requires roughly 50-70% more computation time. Use fewer lines for faster analysis.
Performance Tuning
Hash Size
Engine memory used for position caching:
How hash size affects performance:
- Larger hash: Faster analysis of repeated positions, better for long games
- Smaller hash: Lower memory usage, better for running multiple engines
- Too large: Can actually slow down due to memory management overhead
Thread Count (Cores)
Number of CPU cores the engine uses:
Thread scaling:
- 1 → 2 threads: ~90% faster
- 2 → 4 threads: ~60% faster
- 4 → 8 threads: ~30% faster
- 8 → 16 threads: ~15% faster
Beyond 8 threads, performance gains diminish. More threads use more memory and may reduce efficiency.
Optimal Configurations
Laptop (4 cores, 8 GB RAM)
Laptop (4 cores, 8 GB RAM)
Single engine:
- Threads: 2-3
- Hash: 512 MB
- MultiPV: 1-2
- 2 engines max
- 1-2 threads each
- 256 MB hash each
Desktop (8 cores, 16 GB RAM)
Desktop (8 cores, 16 GB RAM)
Single engine:
- Threads: 6
- Hash: 2048 MB
- MultiPV: 3
- 2-3 engines
- 2-3 threads each
- 512-1024 MB hash each
Workstation (16+ cores, 32+ GB RAM)
Workstation (16+ cores, 32+ GB RAM)
Single engine:
- Threads: 12
- Hash: 4096 MB
- MultiPV: 5
- 3-4 engines
- 3-4 threads each
- 1024-2048 MB hash each
Advanced Engine Options
Accessing Advanced Settings
Common Advanced Options
Syzygy Tablebase Path
Syzygy Tablebase Path
Configure endgame tablebases for perfect endgame play:
- Download Syzygy tablebases (6-piece or 7-piece)
- Set the path to the tablebase files
- Engine will use perfect play in tablebase positions
Contempt
Contempt
Adjusts engine’s draw evaluation:
- Positive contempt: Engine avoids draws, plays for win
- Zero contempt: Neutral evaluation
- Negative contempt: Engine accepts draws more readily
Ponder
Ponder
Engine thinks during opponent’s time:
- Not typically used in analysis mode
- More relevant for playing against the engine
Skill Level (Stockfish)
Skill Level (Stockfish)
Limits engine strength:
- 20: Full strength
- 10-19: Various playing levels
- 0-9: Beginner levels
Syncing Settings Globally
By default, each tab has independent engine settings. To sync:Multiple Engine Comparison
Setting Up Engine Comparison
Configure Each Engine
Set appropriate resources for each:
- Divide CPU cores evenly
- Divide hash memory proportionally
- Use same depth/time for fair comparison
Interpreting Differences
Small differences (±0.1): Normal evaluation variance, both engines agree Medium differences (±0.3-0.5): Different positional assessments, worth investigating Large differences (±1.0+): Significant disagreement:- Check depth - one engine may need more time
- Look for tactics - one engine may have found a hidden move
- Consider style - some engines are more aggressive
When engines disagree, the strongest engine at the highest depth is usually correct, but not always! Chess is complex.
Engine Specialties
- Stockfish: Balanced, extremely strong in all phases
- Leela Chess Zero: Neural network, excellent positional understanding
- Dragon/Komodo: Strong tactical vision
- Specialized engines: Some excel at specific types of positions
Troubleshooting Engine Issues
Engine Won’t Start
Engine Crashes or Freezes
- Reduce resource usage: Lower threads or hash size
- Kill and restart: Use the stop button on the engine card
- Check system resources: Ensure enough available RAM
- Update engine: Newer versions often fix stability issues
Slow Analysis
Too many lines
Too many lines
Reduce MultiPV from 5 to 2-3 for nearly double the speed.
Insufficient threads
Insufficient threads
Increase thread count if CPU usage is low.
Excessive threads
Excessive threads
Decrease threads if you have 8+ allocated; they may be competing.
Hash size issues
Hash size issues
Very large hash (4096+ MB) can slow down. Try reducing to 2048 MB.
Multiple engines
Multiple engines
Close engines you’re not actively using.
Inconsistent Evaluations
- Low depth: Increase depth to 25+ for stable evaluations
- Tactical positions: Complex positions need more time/depth
- Endgames: Use tablebase support for perfect endgame evaluations
Best Practices
Match Resources to Task
Use quick settings for casual analysis, deep settings for critical positions. Don’t waste resources on simple positions.
Leave System Resources
Reserve 1-2 CPU cores and 2+ GB RAM for the operating system and En Croissant itself.
Test Your Configuration
Analyze a known position at different settings to find your optimal balance of speed vs. depth.
Update Regularly
Engine updates often bring performance improvements and bug fixes. Check periodically for new versions.
Use Global Sync
Enable “Sync Globally” for consistent analysis across all tabs.
Monitor Performance
Check the Logs tab occasionally to ensure engines are running efficiently without errors.