GPU-based Parallel Generation Algorithm of Geohash Index Keys

In spatial databases, range queries based on R-tree and K-D tree indexing are more efficient in low-dimensional spaces, and their efficiency in multi-dimensional spaces deteriorates. Geohash divides the space into grids of equal size and passes them through, then maps the points in the grid to a one-dimensional space so that the B+ tree can be used as the index structure of the point set. Taking advantage of GPU parallel processing, a Geohash index key parallel generation algorithm is proposed. Experimental results show that the Geohash index key generation algorithm completed on NVIDIA GeForce GT 730 GPU achieves a speedup ratio of 3 to 5 times compared to traditional CPU operations. Besides, the acceleration effect shows an enhanced trend with the continuous growth of data size. It is verified that GPU has a strong potential advantage in the field of massive spatial data management.