Reputation: 131
I am writing some linear algebra program with openBLAS. The program needs to multiply two matrices; the one is of single-precision and the other one is of double-precision.
I looked up the BLAS reference "http://netlib.org/blas/blasqr.pdf", it seems that BLAS only supports operations on the same type.
Is there anything I miss? Or, to multiply two matrices with different precision, do I need to typecast the single-precision matrix and then multiply them?
Upvotes: 0
Views: 531
Reputation: 26050
Yes, you need to upcast the single-precision matrix (or downcast the double-precision one) and use dgemm
(or sgemm
, resp.)
Upvotes: 1