Chess Engine in Jai

This Chess Engine in Jai is a hobby project designed to find bugs and benchmark the Jai Compiler. You can find video demos of this Chess Engine here

How to Build:

This project includes both a Chess Graphical User Interface and a UCI compatible Chess Engine.

Type jai build.jai to build both the UI and AI on release mode

Type jai build.jai - ui to build the GUI.

Type jai build.jai - ai to build the AI.

Type jai build.jai - ui ai to build both the GUI and the AI.

Type jai build.jai - release to build an optimized build.

Type jai build.jai - release ai nnue_probe to build the engine using Daniel Shawul's NNUE-Probe library. This C library is entirely optional, and is used to provide the best SIMD code for unsupported computers (e.g. AVX512, ARM NEON, MMX, etc.)

The build script will try to detect the CPU you are using, and pick either AVX2, SSE, or cpu automatically based on your processor. Send the flags avx2, sse, or cpu to manually toggle what you want. cpu means running with no SIMD support. Because Neural Networks are based on matrix multiplication, SIMD is needed to obtain the best performance.

Code Organization

This code is divided up into the following files:

• build.jai is a build script for both the User Interface and GUI.
• eval.jai contains chess evaluation parameters
• movegen.jai contains the Chess structs and Piece definitions
• search.jai is the main Chess search function. This file contains code for Multi-threaded Parallel Search
• uci.jai contains the main function for the AI. This file parses the Universal Chess Interface protocol for the engine
• ui.jai is the main function for the UI. This file handles the User Interface with Simp and GetRect
• linux.jai is code specific to the Linux OS.
• windows.jai is code specific to the Windows OS.
• macos.jai is code specific to Mac OS.
• the resources contains the following information
  • Fonts
  • Sound Effects
  • Chess Piece Picture
  • Neural Network Model for AI
• The Efficiently Updatable Neural Network Code is organized as follows:
  • nnue_avx2.jai contains code for AVX2 processors.
  • nnue_sse.jai contains code for SSE processors.
  • nnue_probe.jai contains code for interacting with the NNUE-Probe C library by Daniel Shawul. This can be used to get the best optimization for SIMD architectures unsupported by the existing Jai code (e.g. AVX512 SIMD, ARM NEON SIMD, X86-64 MMX SIMD, etc.)
  • nnue_cpu.jai is a default code with no SIMD. Because it is general with no SIMD whatsoever, this allows someone to run it on any platform.

User Interface Features

• Drag and drop behavior to move pieces
• Click to move pieces
• Legal move generation and detection
• Flip the board using the X Key
• New game using the Ctrl+N Key
• Undo move using the Ctrl+ZKey
• Redo move using the Ctrl+YKey
• Implements UCI (Universal Chess Interface) Protocol to communicate with any UCI compatible chess engine (e.g. Stockfish, RubiChess)
• Can do Human vs Computer, Computer vs Human, Human vs Human, and Computer vs Computer.
• Parse and Loads FEN strings
• Engine vs. Engine Implemented (e.g. Stockfish vs Komodo)
• Multi-engine support
• UCI Engine Options
• Console command-line
• Windows, Mac, and Linux support
• Chess Clock with Increment
• Blindfold Mode
• Highlight Squares with Right Click
• Draw Arrows on the Board with Right Click
• Display Engine Lines
• Dark Background/Light Background Theme

Chess Engine in Jai

• Estimated elo: 3100
• UCI protocol

Chess Engine Options

• Clear Hash
• Transposition Table Memory Size (in MB).
• Number of Threads
• Multiple Principle Variation Lines
• Difficulty Levels 1 to 8

Parallel Search

• Lazy Shared Memory Processing Parallel Search
• Uses the Thread_Group Jai Module to implement threading

Board Representation

• 8x8 Board
• Legal/Psuedo Legal Move Generation
• Staged Move Generation
• Bitboards with Little Endian Rank-File Mapping
• Fancy Magic Bitboards with Parallel Bit Extract
• Kogge-Stone Algorithm
• Move generator generates 324 million positions per second
• Moves encoded as 16-bit integers

Search

• Negamax Search with Alpha-Beta Pruning
• Principle Variation Search
• Iterative Deepening
• Aspiration Window Search
• Internal Iterative Deepening

Pruning and Reductions

• Null Move Pruning w/ Verification
• Low Depth SEE Pruning
• Reverse Futility Pruning/Static Move Pruning
• Delta Pruning
• SEE Quiescene Search Pruning
• Razoring
• Mate Distance Pruning
• History Leaf Reduction/Pruning
• Prob Cut
• Late Move Reduction/Pruning

Extensions

• Singular Search Extensions
• Check Extensions
• Recapture Extensions

Transposition Table

• 16-byte Transposition Table Hash Entries
• 3-fold repetition & Fifty-move Rule
• Zobrist Hashing w/ Incremental Update
• Transposition Table Probing in Non PV Nodes
• Prefetch Transposition Table Entries
• Depth-Preferred Replacement Scheme
• Aging

Move Ordering

• MVV-LVA (Most Valuable Victim, Least Valuable Attacker) Implementation
• Transposition Table Hash Move
• Killer Moves Move Ordering
• History Moves Move Ordering
• Tactical History Move Ordering
• Countermove and Follow Up History Move Ordering
• Relative History Move Ordering

Evaluation

• Efficiently Updatable Neural Networks with Incremental Update
• Simple Mop-Up Endgame Evaluation
• NNUE Stockfish 12 HalfKP Architecture
• Supports the following CPU Architectures:
  • AVX2
  • SSE
  • CPU w/o SIMD

Credits

• Thank you to Jai Beta Users Patrik Smělý and Don Swet (github.com/cookednick) for testing the Chess program on Mac OS.
• Thank you to Maksim Korzh, author of the BBC chess engine. His work provided the main inspiration for this project. Here is a link to the BBC Chess Engine.
• Thank you to Daniel Shawul for translating Stockfish NNUE into C. His work was the basis for my NNUE implementation. nnue-probe
• Thank you to Jonathan Blow, for allowing me access to the Jai Compiler Beta. Jai-Community-Wiki
• Berserk Chess Engine was a great chess engine I learned a lot from. The code is well-organized and well-written. I highly recommend looking through Jay Honnold's work if you want to learn how to take your chess engine to the next level.
• The Stockfish PyTorch NNUE Guide is an incredibly detailed look at how NNUE works. I recommend reading it over and over again to get a full grasp about what is going on.