The document discusses advancements in efficient breadth-first search (BFS) algorithms for large-scale supercomputers, focusing on techniques such as bitmap-based sparse matrix representation, vertex reordering, and load balancing. The proposed methods resulted in a significant performance increase, achieving 38,621 giga traversed edges per second on the K computer, which contributes to its top ranking in the Graph500 benchmark. The authors address challenges in partitioning and accessing hyper sparse matrices while introducing innovative solutions that enhance computational efficiency.