top of page

Sudoku Giant

​

sudoku

INTAGIUM Concept (Beta) handles massive Sudoku grids with speed and precision via a secure C++ backend .

INTAGIUM Concept is a research-driven project developed by the INTAGIUM team to explore scalable solutions to NP-Hard problems. Unlike traditional brute-force, SAT, or backtracking methods, our Concept (a deterministic heuristic) is built from the ground up in C++ with a focus on speed, reliability, and dynamic scalability.

It handles grid dimensions far beyond conventional use cases. From standard 9×9 up to experimental benchmarks like 6400×6400, INTAGIUM maintains consistency, structure, and Sudoku rules compliance with impressive completion rates—often above 98%.

Key Features:

  •  Solves any grid size (tested up to 6400×6400)
  •  User-controlled parameters: grid size, number of initial random values, and number of logic tries
  •  Outputs initial and final grids as downloadable .csv files for full transparency
  •  Fully secured C++ backend, no live code execution, accessed via internal Flask API
  •  Optimized for research, proof-of-concept demos, and large-scale AI experimentation



How It Works :

  1. Random Start – The solver first generates a fully random starting grid; it does not pre-check whether a complete solution exists.
  2. Deterministic Heuristic Passes – The INTAGIUM logic iterates deterministically, filling as many cells as possible within the user-defined try limit.
  3. Transparency of Limits – Any empty cells in the final grid mark either (a) positions that provably have no legal value under the random layout or (b) rare conflicts/limitations in this Beta version—providing clear targets for future optimization.


You can download example results and inspect them yourself:
  •  Initial Grid [SIZE, initalNumbers , Tries ] CSV
  •  Final Grid [SIZE, initalNumbers , Tries ] CSV


INTAGIUM Concept is currently in Beta.You're invited to experiment with it, download real solving results, and help us define the next step of ultra-scale problem solving.



All benchmarks were performed on a regular laptop using an Intel i7-12500H CPU, 16GB RAM 3200MT/s . Sample results (in .csv format) are available for download and verification

Benchmarks : https://datashare.tu-dresden.de/s/wr32FA28iDTtpEJ



 1. Sudoku 6400x6400 with 10,000 and 5 Tries, 99.2% , 14978 seconds






  1. Sudoku 2500x2500 with inital 10,000 random Numbers and 5 Tries , 98.8% , 1740 seconds




  1. Sudoku 1600x1600 , 1000 , 5 Tries, 98.8% , 620 seconds




  1. Sudoku 900x900 with inital 1000 random Numbers and 8 tries, 98.5% , 300 seconds




  1. Sudoku 900x900 with inital 1000 random Number and 5 Tries, 98.8%, 150 seconds.






  1. Sudoku 900x900 1000 50 tries, 99.5% , 1705 seconds




  1. Sudoku 400x400 with 500 inital number and 5 tries, 97.5% , 35 seconds




  1. Sudoku 400x400 with 50 inital Number and 5 Tries ,97.5%, 40 seconds




  1. Sudoku 400x400 with 50 inital Number and 50 Tries, 98%, 240 seconds





  1. Sudoku 400x400 , 50,000 , 5 Tries, 96.64% , 35 seconds






  1. Sudoku 400x400 , 50,000 , 50 Tries, 97.16% , 277 seconds





  1. Sudoku 225x225 , 50,000 , 50 Tries , 95.65% , 90 Seconds






  1. Sudoku 225x225 , 50,000 , 150 tries , 95.91% , 260 seconds,





  1. Sudoku 225x225 , 50,000 , 500 , 97% , 860 seconds






bottom of page