> What gives? What I found that does speed up the code dramatically
> is the -march switch. I guess that can't be repo-ed because it is
> CPU dependent, right? Here's the important settings that I used to
> compile R from source:
>
> CC="gcc"
> CFLAGS="-g -O2 -march=amdfam10"
> CXX="g++"
> CXXF
Thank you Doug and Dirk.
I was afraid I had to move to Eigen... The problem now is that in the
transition I am working with matrices both of type NumericMatrix and
Map.
The following snippet is a loop that is giving me troubles...
wz1 <- lapply(1:K, function(k) matrix(0.0, N, M)) # this is
On Tue, Oct 23, 2012 at 12:56 PM, Giovanni Petris wrote:
> Hello,
>
> I have a 3d array and I want to compute the eigendecomposition of each slice
> [,, k]. I am trying to make Rcpp see the relevant slice of the 3d array as a
> matrix. My experiments, along the lines illustrated below, have so f
Hi Giovanni,
On 23 October 2012 at 17:56, Giovanni Petris wrote:
| Hello,
|
| I have a 3d array and I want to compute the eigendecomposition of each slice
[,, k]. I am trying to make Rcpp see the relevant slice of the 3d array as a
matrix. My experiments, along the lines illustrated below, hav
Hello,
I have a 3d array and I want to compute the eigendecomposition of each slice
[,, k]. I am trying to make Rcpp see the relevant slice of the 3d array as a
matrix. My experiments, along the lines illustrated below, have so far been
unsuccessful. Does anybody have a suggestion? Thank you in
I continued this, and now we can also expose free functions returning
pointers. For example:
Foo* make_foo(){ return new Foo(3, 4) ; }
See svn version of RcppBDT for a real example.
Romain
Le 23/10/12 15:55, Romain Francois a écrit :
Hello,
I've added the possibility for exposed classes to
On 23 October 2012 at 09:46, Rodney Sparapani wrote:
| On 09/25/2012 04:45 PM, Dirk Eddelbuettel wrote:
| >
| > I use and recommend ~/.R/Makevars for that --- as setting certain values at
| > the package level is very verboten ("non portable code" and all that) as far
| > as CRAN is concerned.
| >
On 09/25/2012 04:45 PM, Dirk Eddelbuettel wrote:
I use and recommend ~/.R/Makevars for that --- as setting certain values at
the package level is very verboten ("non portable code" and all that) as far
as CRAN is concerned.
Now, that's the "how do you" part. As for "does it matter" that is a
d
Hello,
I've added the possibility for exposed classes to expose a method that
returns a pointer of the class. That is not easy to explain in a
sentence, here is an example code:
class Foo{
public:
Foo( double x_, double y_) : x(x_), y(y_){}
Foo* clone(){
Hi Romain,
> > Because `$` doesn't evaluate its field-name argument, is
> >
> > field.name <- "value"
> > do.call(`$`, list(x, field.name))
> >
> > the recommended way of accessing methods whose names are computed at
> > run-time? It does work; just wondering if there's a better way.
>
10 matches
Mail list logo
