Efficient CSR-Based Sparse Matrix-Vector Multiplication on GPU

Algorithm 7

IPCSR with the adaptive number of threads per block.

Input: , , , , , ;

Output: ;

()    _;

()    _;

()    ; ;

()    ; ;

()    ; ;

    //Assemble into shared memory

()    for to with   +=   do

()      ;

()    done

()    ();

() ;

() if then

()   //Omitted: Perform a scalar-style reduction

() else

     //Omitted: Perform a multiple scalar-style reduction

() end