Abstract:The whole network data processing of LargeScale GNSS reference station network can avoid the technical problems such as subnetwork division and common station selection of the subnetwork processing strategy. The number of stations whose observation data to be processed should be less than 100. The least square adjustment and SRIF are compared with each other, and then we point out that the least square adjustment is better than the later for large data processing. Further, the high performance computing methods for normal equation are researched; the existing normal equation computing methods are compared with each other, and it is pointed out that the Cholesky matrix decomposition method is better in computing efficiency. At last, the parallel Cholesky matrix decomposition method based on OPENMP is achieved, which can take full advantage of multicore/thread processor to improve the computing efficiency in personal computer platform.