Suitability of parallel computing?

I want to make this code run as fast as possible:

 f1=[50 20 3 -50 -27]*(1000000); %frequency
    p1=[10 200 50 9 0]; %phase
    a1=[1 0.2 0.7 0.5 0.1]; % amplitude 

 s=1000000;  
tot=s/max(abs(f1))*2.5;
 t=linspace(0 ,tot ,s);

    x=zeros(length(f1),s);
    x=f1'*t;
    y=repmat(p1',1,s);
    I=a1*sin((2*pi*x)+y);
    Q=a1*cos((2*pi*x)+y);

Is this the sort of problem where using parallel computing could help?

Answer 1

[Edric Ellis]

You could use the gpuArray support from PCT to perform this calculation. On my machine, this runs about 15x faster. (I have a fairly old Tesla C2070 GPU, and a quad-core CPU). Here's what I timed:

d = gpuDevice();
tic
x=gpuArray(f1)' * gpuArray.linspace(0, tot, s);
y=gpuArray(p1');
I2=a1 * arrayfun(@(xx,yy) sin((2*pi*xx)+yy), x, y);
Q2=a1 * arrayfun(@(xx,yy) cos((2*pi*xx)+yy), x, y);
wait(d);
toc

Note that it's unlikely that using parfor or spmd will speed this computation up much since the CPU computation is already intrinsically multi-threaded by MATLAB.