GPU Implementation: Accelerating 3D-Bin Packing Problem
N Murali Krishna, Ashwin M. Krishnan, Manoj Nambiar · 2023
This study addresses the challenging 3D bin packing problem (3D-BPP) using state-of-the-art algorithms and GPU acceleration. The focus is on enhancing memory-bound packing operations and compute-intensive Deep Q-network using CUDA kernels, resulting in more than a 50-fold speed increase compared to a sequential CPU version. This acceleration enables training of the Deep Q-Network without heuristics, improving packing efficiency by 2% to 4.2%. The study includes a comprehensive performance analysis across standard 3D-BPP test cases, showcasing the advantages of combining application and hardware architecture analysis for achieving high performance.