MATLAB: How to use the MATLAB SAVE function inside a PARFOR loop in Parallel Computing Toolbox 5.1 (R2011a)

Parallel Computing Toolbox

I want to use the SAVE function inside PARFOR loop

Best Answer

It is possible to use the exact SAVE function inside a PARFOR loop. All you need to do is to create a wrapper function on top of SAVE and call it inside PARFOR. An example code snippet is shown below:

parfor ii=1:10
    x = magic(ii);
    fname = sprintf( 'file_%d', ii );
    iSaveX( fname, x );
end
% wrapper around SAVE function
function iSaveX( fname, x )
    save( fname, 'x' );
end

Related Solutions

MATLAB: How to use SAVE with a PARFOR loop using Parallel Computing Toolbox

Transparency is violated by the SAVE command because in general MATLAB cannot determine which variables from the workspace will be saved to a file.

The solution is to move the call to SAVE to a separate function and to call that function from inside the PARFOR loop. Pass any variables that are to be saved as arguments to the function. That way MATLAB can determine which ones will be saved and transparency is not violated.

For example:

Save the following as "parsave.m":

function parsave(fname, x,y)
  save(fname, 'x', 'y')
end

Then run it with:

parfor ii = 1:4
  x = rand(10,10);
  y = ones(1,3);
  parsave(sprintf('output%d.mat', ii), x, y);
end

Similarly, for loading several files with enumerated file names, we can use the following code.

function [output] = par_load(i)
  eval(['load file_' num2str(i) '.mat']);
  output = time;
end

Which can be run with:

parfor i=1:4
    data = par_load(i);
    a(i) = 10 + data(1); 
end

MATLAB: Undefined function ‘matlabpool’ for input arguments of type ‘char’.

If you encounter that issue on R2012a, then the difficulty would be that you do not have the Parallel Computing Toolbox installed (or perhaps that it is not licensed.)

If you encounter the issue on R2013b or later, then you might potentially have PCT installed and licensed, but you might be encountering the difficulty that as of R2013b, matlabpool() was renamed to parpool(), which takes slightly different options.

I wrote a transition routine to make it easier to deal with code that uses matlabpool in R2013b and later. I wrote it for use with one particular third party software package that only needed a subset of the possible matlabpool inputs, so it is not a complete replacement for matlabpool: it only handles a small number of common inputs. You could try adding this transition code as matlabpool.m and seeing if it works for you. If you start getting messages about gcp or parpool not existing then your problem is with not having PCT installed or licensed.

function varargout = matlabpool(varargin)
  %matlabpool shim for R2013b and later, limited functionality
    %written July 2018, by Canada Eat The Cookie Foundation
    if nargin == 1 && ischar(varargin{1}) && strcmp(varargin{1}, 'size')
        pool = gcp('nocreate');
        if isempty(pool)
            varargout{1} = 0;
        else
            varargout{1} = pool.NumWorkers;
        end
    elseif nargin == 1 && isnumeric(varargin{1})
         pool = gcp('nocreate');
         if ~isempty(pool)
             if pool.NumWorkers ~= varargin{1}
                 delete(pool);
             end
             pool = parpool(varargin{1});
         else
             pool = parpool(varargin{1});
         end
         varargout{1} = pool;
    elseif nargin == 1 && ischar(varargin{1}) && strcmp(varargin{1}, 'open')
         pool = gcp();
         varargout{1} = pool;
    else
        [varargout{:}] = parpool(varargin{:});
    end

Best Answer

Related Solutions

MATLAB: How to use SAVE with a PARFOR loop using Parallel Computing Toolbox

MATLAB: Undefined function ‘matlabpool’ for input arguments of type ‘char’.

Related Question