Let A be a (real) large N×N matrix and NNZ the number of its nonzero elements. If NNZ is very small (resp. large), say NNZ=O(N) (resp O(N2)), then A is called sparse (resp. dense). Both storage requirements and computational time in any application operating on large sparse matrices may be dramatically reduced by only storing nonzero elements and avoiding useless operations on zero elements (Bik, 1996). This is achieved by using compressed storage formats for sparse matrices. A sparse matrix may have various structures according to the locations of its nonzero elements. For each sparse matrix structure, one or more dedicated compressed storage formats are known in the literature (i.e. CSR, COO, MCSR, JAD, etc) (Saad, 2019). In this paper, we are only interested in the CSR (Compressed Storage Row) format because it is the most used in the literature (Goharian, El-Ghazawi & Grossman, 2001).