Hi, I am an IDL novice so please bear with me if this question is trivial. I have a piece of code that on the basis of two images generates a new image. I've written a function that for each pixels (i,j) in the input images iteratively solves for the new image value. The way I step through the two images is in for loops (I now this is probably bad...) like below for i=1,dim1 for j=1,dim2 call to function which returns new value of pixel(i,j) endfor endfor My calculation is rather slow (~0.1s pr. pixel) which becomes long with 512x512 images and uses only one thread in the processor (probably due to the nested for-loops?). My calculation is independent from pixel to pixel and should hence gain from parallelization which potentially could increase the speed a lot. I have looked at the split_for method which is floating around the net but have not been able to compile it on IDL 8.4. Anybody else been having luck with this? And furthermore, how would one avoid the nested for loops? Thanks in advance for any replies/comments