[Rcpp-devel] Eigen vs Arma -- was: Re: Ceres nonlinear least squares solver

Andreas Alfons andreas.alfons at econ.kuleuven.be
Mon May 7 12:15:49 CEST 2012


Since my name was dropped in an earlier post by Dirk (and to prevent
the conversation from becoming a general discussion on what is nice
about Mac OS X and what isn't), I'll post the benchmarks I sent to
Dirk here, too.

I benchmarked the computation times for my function sparseLTS() from
package robustHD for different problem sizes on my laptop running
Linux:

100 observations, 1000 variables:
With Arma:
  user  system elapsed
 10.312   0.004  10.332
With Eigen:
  user  system elapsed
 6.525   0.000   6.534

100 observations, 10 000 variables:
With Arma:
  user  system elapsed
269.889   2.148 272.606
With Eigen:
  user  system elapsed
214.134   2.320 216.904

If people are interested, I post the code for the benchmarks later on
(I don't have them on the machine that I'm currently working on). As
far as I remember the speed gain with Eigen is much larger on Linux
than on Windows, so I can also run the code on my Windows machine.

- Andreas


On Mon, May 7, 2012 at 7:43 AM, Tim Triche, Jr. <tim.triche at gmail.com> wrote:
> Why Mac OS X?
>
> 1) it's shiny
> 2) it's Unix
> 3) everything's a PDF
>
> Two of my most competent coworkers absolutely love their Macs.  They also ask me the most "interesting" tech support questions, ones where I feel like I can't say "the problem is that you are using Windows" but at the same time, I get vaguely annoyed that Mac OS X tends to be more trouble than say Linux to set up MySQL client libraries, or compile Boost, or whatever.  I can't get *angry* at them, but as a user of *nix for two decades (i.e., a beginner), I want to say "you know, things don't have to be a pain like Solaris or FreeBSD anymore!".
>
> But then I wouldn't be able to dump interesting programming projects onto them, so I just fix whatever it is that's causing the problem.  Look on the bright side, at least they're not using Windows.  I personally prefer Linux over *BSD, but that may be cause I started with Irix.
>
> Oh, and my wife has a MacBook Air, too.  Which is 1000x easier to troubleshoot than her old Windows laptop.  I do not complain.
>
>
>
> On Sun, May 6, 2012 at 10:05 PM, Michael Hannon <jm_hannon at yahoo.com<mailto:jm_hannon at yahoo.com>> wrote:
> Steve Lianoglou <mailinglist.honeypot at gmail.com<mailto:mailinglist.honeypot at gmail.com>> wrote:
>
>>On Sat, May 5, 2012 at 1:21 PM, Dirk Eddelbuettel <edd at debian.org<mailto:edd at debian.org>> wrote:
>>[snip]
>>> At the end of day, and to me as a non-user of OS X, this just seems to make
>>> working on OS X as tedious as working on Windows where you have to patch
>>> your
>>> system together in bits and bobs.  I guess I am spoiled rotten by 15+ years
>>> of dpkg and apt-get and "things just working". Nicer for development in my
>>> book.
>>
>>Can't argue w/ the fact that this makes developing on OSX a bit of a
>>PITA ... it's just that working w/ the rest of OSX is so nice ;-)
>
> This may alleviate some of the PITA:
>
>    http://mxcl.github.com/homebrew/
>
> (I haven't used it.)
>
> I *am* curious to know what is so nice about Mac OS X.  This isn't a troll.  I
> can see that Macs are useful if you have to deal with, say, Adobe products,
> etc.  But I've never been able to fathom their appeal for work in math,
> statistics, science, etc.  (My wife has multiple Macs, and she loves the
> things, even for her statistics research, but I don't seem to "get it".)
>
> -- Mike
> _______________________________________________
> Rcpp-devel mailing list
> Rcpp-devel at lists.r-forge.r-project.org<mailto:Rcpp-devel at lists.r-forge.r-project.org>
> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/rcpp-devel
>
>
>
> --
> A model is a lie that helps you see the truth.
>
> Howard Skipper<http://cancerres.aacrjournals.org/content/31/9/1173.full.pdf>
>


-- 
Andreas Alfons
Faculty of Business and Economics, KU Leuven
www.econ.kuleuven.be/andreas.alfons/public/


More information about the Rcpp-devel mailing list