For the axpy, the calling sequences are given by place HR here
place HR here The data type of the operands and parameters like vector length and stride are now passed as part of linear algebra objects alpha, x and y. Instead of returning the result as the function value, it is returned in alpha.
When a node calls PLA_Local_dot, the operation
is equivalent to
extracting the local information (data type, length, and pointer to the
buffers) and locally performing the inner product.
The local dimensions of alpha must be ,
but it can be of any object type.
The local dimensions of x, and y must be such
that the operation is well-defined. All objects must have the same data type.
Notice: no communication is performed as part of this call.
Thus, all information must be locally available.
All nodes must simultaneously call PLA_Dot, which performs
the BLAS call on the distributed objects.
Parameter alpha must
be a multiscalar.
Parameters x and y
must be of object type vector or (duplicated) projected
vector.
The global dimensions of the different objects must be such
that the operation is well-defined.