Is there sparse matrix-dense vector multiplication routine with matrix stored in int and vector in double..?

kirankumarm — Wed, 28 Jul 2010 08:50:22 GMT

Hi All,
I want to use sparse matrix-dense vector multiplication. Matrix is stored in CSR(compressed sparse row) format.

My matrix is of datatype int and
vector is of datatype double.

I searched for routines and found mkl_dcsrmv, mkl_dcsrgemv but both need to have same data type(double).

Is there any other way so that I can use matrix in int and vector in double and perform sparse matrix-dense vector multiplication using routines..?

Regards,
M.Kiran Kumar.

Is there sparse matrix-dense vector multiplication routine with

Gennady_F_Intel — Wed, 28 Jul 2010 09:29:32 GMT

Kiran, you right. No such m-v routines in MKL ( only for single and double precision).

for your case, you can try to explicity convert int -> double

say:

int val;

(double)val;

static_cast(number);

but, it wiil take an additional time ...

--Gennady

topic Is there sparse matrix-dense vector multiplication routine with in Intel® oneAPI Math Kernel Library

Is there sparse matrix-dense vector multiplication routine with matrix stored in int and vector in double..?

Is there sparse matrix-dense vector multiplication routine with