A reason why you should avoid MATLAB-builtin waitbar() and instead use fprintf() ref: http://qiita.com/kouichi-c-nakamura/items/ae9dc371eade0fd0a945

file0.mat

% waitbarVAfprintf

trials = 10;

elapsed1 = zeros(10,1);
for j = 1:10
    clear n
    tic
    n = 1000;
    fprintf('%d to go\n',n)
    for i = 1:n
        fprintf('*')
    end
    fprintf('\n')
    elapsed1(j) = toc;
end

elapsed2 = zeros(10,1);
for j = 1:10
    clear n
    tic
    n = 1000;
    for i = 1:n
        if round(i/n*100) > round((i-1)/n*100)
            fprintf('*')
        end
    end
    fprintf('\n')
    elapsed2(j) = toc;
end

elapsed3 = zeros(10,1);
for j = 1:10
    clear n wbmsg wb
    tic
    n = 1000;
    for i = 1:n
        wbmsg = sprintf('In progress: %.1f%%', i/n*100);
        if ~exist('wb', 'var')
            wb = waitbar((i-1)/n, wbmsg);
        else
            waitbar((i-1)/n, wb, wbmsg);
        end
    end
    delete(wb) % cannot use 'close'!!!
    elapsed3(j) = toc;
end

elapsed4 = zeros(10,1);
for j = 1:10
    clear n wbmsg wb
    tic
    n = 1000;
    for i = 1:n
        if round(i/n*100) > round((i-1)/n*100)
            wbmsg = sprintf('In progress: %.1f%%', i/n*100);
            if ~exist('wb', 'var')
                wb = waitbar((i-1)/n, wbmsg);
            else
                waitbar((i-1)/n, wb, wbmsg);
            end
        end
    end
    delete(wb) % cannot use 'close'!!!
    elapsed4(j) = toc;
end

%%
fprintf('1. fprintf: %f %s %f msec per cycle\n',...
    mean(elapsed1),char(177),std(elapsed1));
fprintf('2. fprintf for each %%: %f %s %f msec per cycle\n',...
    mean(elapsed2),char(177),std(elapsed2));
fprintf('3. waitbar: %f %s %f msec per cycle\n',...
    mean(elapsed3),char(177),std(elapsed3));
fprintf('4. waitbar for each %%: %f %s %f msec per cycle\n',...
    mean(elapsed4),char(177),std(elapsed4));

fprintf('fprintf is %.1f-fold faster than waitbar updating at every cycle\n',...
    mean(elapsed3./elapsed1));
fprintf('fprintf for each %% is %.1f-fold faster than waitbar for each %%\n',...
    mean(elapsed4./elapsed2));

fprintf('fprintf for each %% is %.1f-fold faster than frpintf updating at every cycle\n',...
    mean(elapsed1./elapsed2));
fprintf('waitbar for each %% is %.1f-fold faster than waitbar updating at every cycle\n',...
    mean(elapsed3./elapsed4));

fprintf('fprintf for each %% is %.1f-fold faster than waitbar updating at every cycle\n',...
    mean(elapsed3./elapsed2));

file1.txt

fprintf is 360.9-fold faster than waitbar updating at every cycle
fprintf for each % is 510.3-fold faster than waitbar for each %
fprintf for each % is 10.7-fold faster than frpintf updating at every cycle
waitbar for each % is 7.5-fold faster than waitbar updating at every cycle
fprintf for each % is 3829.7-fold faster than waitbar updating at every cycle

file2.mat

n = 1000;
for i = 1:n
    if round(i/n*100) > round((i-1)/n*100)
        wbmsg = sprintf('In progress: %.1f%%', i/n*100);
        if ~exist('wb', 'var')
            wb = waitbar((i-1)/n, wbmsg);
        else
            waitbar((i-1)/n, wb, wbmsg);
        end
    end
end