https://hgpu.org/?p=6577
Toward Accelerating the Matrix Inversion Computation of Symmetric Positive-Definite Matrices on Heterogeneous GPU-Based Systems