[R] R badly lags matlab on performance?

[(Ted Harding)]

On 03-Jan-09 18:28:03, Ben Bolker wrote:
> Ajay Shah <ajayshah <at> mayin.org> writes:
>> On Sat, Jan 03, 2009 at 06:59:29PM +0100, Stefan Grosse wrote:
>> > On Sat, 3 Jan 2009 22:25:38 +0530 Ajay Shah <ajayshah <at>
>> > mayin.org> wrote:
>> > 
>> > AS> system.time(for (i in 1:10000000) {a[i] <- a[i] + 1})
>> > 
>> > AS> I wonder what we're doing wrong!
>> > 
>> > it is no secret that R does badly with loops. Thats why it is
>> > recommended to use vectorized operations.
>> 
>> But there's a big difference even on the vectorised version: a <- a +
>> 1. Why should that be? Both systems should merely be handing down to
>> the BLAS. The (stock) R install has a less carefully setup BLAS as
>> compared with the (stock) matlab install?
> 
>   See my other message. I'm suspicious of the real size of
> the difference, I think the difference could well be
> noise.   Also, this particular bit of arithmetic doesn't
> involve BLAS -- see arithmetic.c (dig down until you get to
> integer_binary) ...
>   Ben Bolker

I just ran Ajay's examples 3 times over:
R 2.8.1 on Debian Etch using 1MB of RAM in a VirtualBox VM
running on a 1.73GHz CPU. Results:

          user  system elapsed
Vector:  0.112   0.288   0.393
Loop:   65.276   0.300  65.572

Vector:  0.076   0.312   0.389
Loop:   65.744   0.332  66.076

Vector:  0.068   0.328   0.394
Loop:   65.292   0.308  65.597

Not dissimilar to Ajay's R times (though my loop was about 50% longer).

However, all three runs were very similar -- a little noise,
but not much!

I don't have octave (on the same machine) to compare these with.
And I don't have MatLab at all. So I can't provide a comparison
on that front, I'm afraid.
Ted.

--------------------------------------------------------------------
E-Mail: (Ted Harding) <Ted.Harding at manchester.ac.uk>
Fax-to-email: +44 (0)870 094 0861
Date: 03-Jan-09                                       Time: 19:10:51
------------------------------ XFMail ------------------------------