Programming

Preallocating Memory

Repeatedly expanding the size of an array over time, (for example, adding more elements to it each time through a programming loop), can adversely affect the performance of your program. This is because

• MATLAB has to spend time allocating more memory each time you increase the size of the array.
• This newly allocated memory is likely to be noncontiguous, thus slowing down any operations that MATLAB needs to perform on the array.

The preferred method for sizing an array that is expected to grow over time is to estimate the maximum possible size for the array, and preallocate this amount of memory for it at the time the array is created. In this way, your program performs one memory allocation that reserves one contiguous block.

The following command preallocates enough space for a 25,000 by 10,000 matrix, and initializes each element to zero:

• A = zeros(25000, 10000);
  

Building a Preallocated Array

Once memory has been preallocated for the maximum estimated size of the array, you can store your data in the array as you need it, each time appending to the existing data. This example preallocates a large array, and then reads blocks of data from a file into the array until it gets to the end of the file:

• blocksize = 5000;
  maxrows = 2500000; cols = 20;
  rp = 1;     % row pointer
  
  % Preallocate A to its maximum possible size
  A = zeros(maxrows, cols);
  
  % Open the data file, saving the file pointer.
  fid = fopen('statfile.dat', 'r');
   
  while true
     % Read from file into a cell array.  Stop at EOF.
     block = textscan(fid, '%n', blocksize*cols);
     if isempty(block{1})   break,   end;
   
     % Convert cell array to matrix, reshape, place into A.
     A(rp:rp+blocksize-1, 1:cols) = ...
        reshape(cell2mat(block), blocksize, cols);
   
     % Process the data in A.
     evaluate_stats(A);               % User-defined function
   
     % Update row pointer
     rp = rp + blocksize;
  end
  

Note    If you eventually need more room in a matrix than you had preallocated, you can preallocate additional storage in the same manner, and concatenate this additional storage onto the original array.

Reshaping a Matrix

Shifting and Sorting Matrices

© 1994-2005 The MathWorks, Inc.