Hello everyone,I'm a high school student interested in Game Theory. I've developed a tool to find 2 and 3 ways equity using NVIDIA CUDA. The engine generates flat binary files for instant O(1) access during possible later CFR training. I used the 7-card HandRanks LUT and optimized the kernel for shared memory tiling. I’m sharing this to get feedback from the community and to help anyone working on GTO solvers. GitHub Link: https://github.com/3piee/cuda-poker-3-2-... (https://github.com/3piee/cuda-poker-3-2-ways-equity-engine)
Newbie1 posts
Posted May 9, 2026, 4:35 PM#1