The BLAS Level 2 routine cblas_?ger computes A := alpha*x*y'+ A. Is there a simpler routine that just calculates A := alpha*x*y'?
Setting A=0 offers the same results, but does it provide good performance too? i.e. am I wasting computation in doing the additions?
Hello Tang, Wei,
MKL only provides A := alpha*x*y'+ A for cblas_?ger. Yes, you can set A=0 to meet your expectation of A := alpha*x*y'. And it has good performance without any performance degradation.