Matlab cputime with kmeans return 0

I am measuring the cputime taken by kmeans algorithm for each iteration using cputime feature. However, some iterations return cputime = 0. Here's my implementation:

   load fisheriris;
   [~,C] = kmeans(meas, 3, 'options',statset('MaxIter', 1),'Display', 'off');
   results=[];
   for i = 1:15
          t=cputime;
          [~,C] = kmeans(meas, 3, 'options',statset('MaxIter', 1),'Start',C, 'Display', 'off');
          elapsedCPUTime=cputime-t;
          results=[results;elapsedCPUTime];
   end 

This is the results I got for 15 iterations: 0, 0, 0.046875, 0, 0, 0, 0, 0, 0.03125, 0, 0, 0, 0, 0 ,0.03125. My first thought is that the computational time was too quick, hence 0 second. Is it true? If so, how can we achieve more precise cputime?

Thanks a lot.

From the documentation:

To measure performance, it is recommended that you use the timeit or tic and toc functions. For more information, see Using tic and toc Versus the cputime Function.

Following that link:

Time a significant enough portion of code. Ideally, the code you are timing should take more than 1/10 second to run.

Try running kmeans in a loop.

load fisheriris;
[~,C] = kmeans(meas, 3, 'options',statset('MaxIter', 1),'Display', 'off');
results=[];
for ii = 1:15
      t=cputime;
      for k = 1:100
          [~,Ctemp] = kmeans(meas, 3, 'options',statset('MaxIter', 1),'Start',C, 'Display', 'off');
      end
      elapsedCPUTime=(cputime-t)/100;
      C = Ctemp;
      results=[results;elapsedCPUTime];
end

Using i as the loop variable might yield unexpected results, which is why I changed it to ii.