how to speed up code

I am dealing with big amount of data where the for loop takes a long time to process.

tic
% index_tracks_matching_temp is a vector of 60 million (1*60m), also track_length_dns is 150k long 
index_points_matching_temp=index_points_matching;
for i=1:1000 % just first 100 tracks %% or for i=1:track_length_DNS
    mycell_track_index{i,1}=index_tracks_matching_temp(1:track_length_DNS(i));
    mycell_points_index{i,1}=index_points_matching_temp(1:track_length_DNS(i));
    index_tracks_matching_temp(1:track_length_DNS(i))=[];
    index_points_matching_temp(1:track_length_DNS(i))=[];
end
clear index_tracks_matching_temp index_tracks_matching_temp
toc

% processing for i=1000 takes 30 seconds and I need to process up to 500k that's why i need a different way to do it. I always used loops so im not sure about other ways to do it

Answer 1

[Walter Roberson]

If I read the code correctly, then

numused = sum(track_length_DNS);
mycell_tracks_index = mat2cell(index_tracks_matching(1:numused), 1, track_length_DNS);
mycell_points_index = mat2cell(index_points_matching(1:numused), 1, track_length_DNS);

with no loop.

[Tarek]

thanks, Mat2cell function is perfect, I did not know about it before.