MEX to improve a recursive computation

Hi everybody,

I have a simple function of the form:

function t = RungeKutta_secondorder(c1,c2,w0,dcr,d0,E)
dw=0;
d=d0;
dt=0.001;
t=0;
while d<dcr
    kp1=0.5*(c1-(E*sin(d)/c2))*dt;
    kpp1=w0*dw*dt;
    kp2=0.5*(c1-(E*sin(d+kpp1)/c2))*dt;
    kpp2=w0*(dw+kp1)*dt;
    t=t+dt;
    dw=dw+0.5*(kp1+kp2);
    d=d+0.5*(kpp1+kpp2);    
end
end

The typical values for the arguments are: c1=0.4,c2=6,w0=300,dcr=1.1,d0=0.3,E=1.

Another program calls this function many many times. Therefore, it is very important to reduce its computation time to the least possible. How can I achieve this goal by MEX and/or filter techniques?

with MEX I mean including a MEX function instead of the above piece of code (I do not know how to write that MEX).

with filter I mean the filter command in signal processing toolbox which is sometimes used to speed up recursive computations (I do not know how to use it instead of the above function).

Thanks a lot,