High-Frequency Poker Engine (HFT-C++)

A low-latency Texas Hold'em equity calculator engineered for high-throughput Monte Carlo simulation and real-time decision making.

🚀 Key Technical Features

• O(1) Hand Evaluation: Implemented "Cactus Kev" Prime-Product algorithm with perfect hashing (evaluator.cpp) for constant-time lookups.
• Bitmask Optimization: State representation using 32-bit integers to minimize cache misses and memory overhead.
• Parallelized Monte Carlo: Multi-threaded simulation engine using std::async and std::future (monte_carlo.cpp) to batch simulations across available CPU cores.
• Foreign Function Interface (FFI): Zero-copy extern "C" bridge (wrapper.cpp) allowing Python to drive the C++ engine without serialization overhead.

🛠️ Architecture

The system follows a hybrid architecture optimized for performance:

1. Core Engine (C++): Handles memory management, bitwise logic, and the simulation loop.
2. Control Layer (Python): Manages strategy logic, game flow, and data visualization via ctypes.

⚡ Benchmarks

• Throughput: ~30 Million hand evaluations/sec (Single Thread)
• Latency: < 20ms per 50k simulation batch on standard hardware.

📦 Build & Run

Prerequisites

• G++ (C++20 support recommended)
• Python 3.8+

Compilation

# 1. Generate the Lookup Table (Pre-computation)
g++ -std=c++20 -O3 generator.cpp -o generator
./generator

# 2. Build the Shared Library (Engine)
# For Linux/Mac:
g++ -std=c++20 -O3 -shared -fPIC wrapper.cpp poker.cpp evaluator.cpp monte_carlo.cpp -o libpoker.dylib
# Note: Output name 'libpoker.dylib' matches bot.py configuration. Use .so for Linux.