What is the fastest way to perform x'Ay ?

In my code I have a bottleneck when I try to perform the following calculation: Given a square matrix A and vectors of equal length x,y (assume all dimensions are compatible), compute

    result  = x'*A*y

Is there a way that speeds this up that is simple to implement?

Answer 1

[James Tursa]

I don't know that there is anything faster than what you have written. A matrix multiply followed by an inner product using BLAS routines in the background. How large are your variables? Is this calculation in a loop where everything changes at each iteration, or do some of the variables remain constant during the loop?