Block-Cholesky for parallel processing (Q1188310)

The paper is devoted to the Cholesky factorization \(A=LL^ T\) with \(A\) a positive definite symmetric matrix and \(L\) a lower triangular matrix. A blocked algorithm based on Level 3 Basic linear Algebra Subroutines (BLAS) is presented which implements parallelism over the blocks. The amount of parallelism is studied and the data dependency graph is discussed. The SCHEDULE package is used to obtain a portable schedule of the tasks. Numerical results of the method are presented and compared with results for a block algorithm parallelized within the Level 3 BLAS kernels.