Date: Apr 21, 2013 4:57 AM
Subject: time consuming problem and MKL and other solutions to speed up
I have to simulate a 3d domain within a time-loop with n=10000 step with matlab.
but my problem is that Matlab is time consuming in this case.
I used from best possible commands in matlab.for example,I replaced nested for-loops as following:
X(1:nx,2:ny,2:nz) = coeff1(1:nx,2:ny,2:nz).* (Y(1:nx,2:ny,2:nz) - Y(1:nx,1:ny-1,2:nz)) ...
But my simulation is very time consuming,yet.
for example in a 3d domain with 100*100*100 point,my program need 500 minutes for running!
in my program I use from "mldivide" command and sparse arrays for 3d,that are fastest methods.
I want to use from "intel Math Kernel Library (MKL),that is good for speed up.
but,I don't know how use from it in matlab.
is there anyone who direct me?(best and fastest method for long simulations with matlab)