Thank you JJ, that makes sense. I had no idea that R could be so taxing
when bench-marking.
On Thu, Dec 20, 2012 at 1:27 PM, JJ Allaire <jj.alla...@gmail.com> wrote:
> It's possible that the overhead of making R function calls is swamping all
> of the other performance data here. I've re-written the example w/
> sourceCpp and with the calls to the C++ version happening inside C++ rather
> than R. This yields more like 100:1 speedup:
>
> #include <Rcpp.h>
> using namespace Rcpp;
>
> double get_var(const NumericVector& v,double m,int l) {
> double a = 0;
> for (int i = 0; i <l;i++) {
> a += (v[i]-m)*(v[i]-m);
> }
> return(a/l);
> }
>
> double get_sum(const NumericVector& v,int l) {
> double s = 0;
> for (int i = 0; i <l;i++) {
> s += v[i];
> }
> return(s);
> }
>
> // [[Rcpp::export]]
> double rcppVar(const NumericVector& v) {
> int n = v.size();
> double v_mean = get_sum(v,n)/n;
> double v_var = get_var(v,v_mean,n);
> return v_var;
> }
>
> // [[Rcpp::export]]
> void callRcppVar(const NumericVector& v) {
> for (int i=0; i<100; i++)
> rcppVar(v);
> }
>
>
> /*** R
> a=1:1000000
> system.time(callRcppVar(a))
> system.time(for (i in 1:100) var (a))
> */
>
> > a=1:1000000
>
> > system.time(callRcppVar(a))
> user system elapsed
> 0.003 0.001 0.005
>
> > system.time(for (i in 1:100) var (a))
> user system elapsed
> 0.591 0.128 0.719
>
_______________________________________________
Rcpp-devel mailing list
Rcpp-devel@lists.r-forge.r-project.org
https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/rcpp-devel
