xx={'2017032312_8953';'2017032312_8978'};
yy=[10;11];
for jj = 1 : length( xx )
fprintf( '%s %3.1E \n', xx{jj}, yy(jj) )
end
outputs
2017032312_8953 1.0E+01
2017032312_8978 1.1E+01
Comparison of elapse times of loop codes and the vectorized code by Stephen Cobeldick
"It is better, since my data is huge." "Huge" implies that you want to write to a text file.
Here is a test on R2016a 64bit, Win7 and an old vanilla desktop.
>> et = cssm(1e4)
et =
8.4265 0.6577 0.1814
The difference between the first and the second elapse time value illustrates the influence of automatic flushing on speed. (Both are based on for-loops.) See Improving fwrite performance at Undocumented Matlab by Yair Altman. The third value is the elapse time of the vectorized code by Stephen Cobeldick.
The test is done with this function
function et = cssm(N)
xx = repmat( {'2017032312_8953';'2017032312_8978'}, N,1 );
yy = repmat( [10;11], N, 1 );
tic
fid = fopen( 'test1.txt', 'w' );
for jj = 1 : length( xx )
fprintf( fid, '%s %3.1E \n', xx{jj}, yy(jj) );
end
fclose( fid );
et = toc;
tic
fid = fopen( 'test2.txt', 'W' );
for jj = 1 : length( xx )
fprintf( fid, '%s %3.1E \n', xx{jj}, yy(jj) );
end
fclose( fid );
et(end+1) = toc;
tic
C = xx.';
C(2,:) = num2cell(yy);
fid = fopen( 'test3.txt', 'W' );
fprintf( fid, '%s %3.1E \n', C{:});
fclose( fid );
et(end+1) = toc;
end
