Re: For loop benchmarks between Octave versions

[Daniel J Sebald]

On 06/24/2014 02:04 AM, Julien Bect wrote:

> Here are the new results on 3.8.2-RC1:
>
> without patch with patch
> doubleForLoop/1 181.8 [ 1.8] 210.2 [ 7.9] evol: +15.6%

Strange one there, but it doesn't matter because such loop isn't to be 
used in practice.

> vectorOfSquares 326.8 [ 4.0] 591.9 [ 2.1] evol: +81.1%

I'm surprised given all that really is being done with this test is 
populating a memory array:

function vectorOfSquares (nrep, n)

for r = 1:nrep
    x = zeros (1, n);
    for i = 1:n
        x(i) = i * i;
    end
end

end

I guess there is a multiplication there as well, but that's pretty 
simple too.

In any case, one thing to keep in mind is that the vectorOfSquares tests 
a for-loop implementation.  But a more-likely faster approach would be

function vectorOfSquaresFaster (nrep, n)

for r = 1:nrep
    i = 1:n;
    x = i.*i;
end

end

Just depends on what it is one wants to measure (looping or vector 
multiplication).

Dan