Skip to content

Instantly share code, notes, and snippets.

@romainfrancois
Forked from hadley/apply-benchmark.cpp
Last active December 25, 2015 12:09
Show Gist options
  • Star 0 You must be signed in to star a gist
  • Fork 0 You must be signed in to fork a gist
  • Save romainfrancois/6974012 to your computer and use it in GitHub Desktop.
Save romainfrancois/6974012 to your computer and use it in GitHub Desktop.
rowApply
#include <Rcpp.h>
using namespace Rcpp;
// [[Rcpp::export]]
NumericVector rowApply0(NumericMatrix& x, const Function& FUN)
{
int n = x.nrow();
NumericVector result = no_init(n);
for (int r = 0; r < n; r++) {
result[r] = as<double>(FUN(x(r, _) ) );
}
return result;
}
// [[Rcpp::export]]
NumericVector rowApply1(NumericMatrix& x, const Function& FUN)
{
int n = x.nrow();
NumericVector result = no_init(n);
for (int r = 0; r < n; r++) {
Language call(FUN, x(r, _)) ;
result[r] = as<double>(call.fast_eval() );
}
return result;
}
// [[Rcpp::export]]
NumericVector rowApply2(NumericMatrix& x, const Function& FUN)
{
int n = x.nrow();
NumericVector result = no_init(n);
Language call(FUN, R_NilValue);
Language::Proxy proxy(call, 1);
for (int r = 0; r < n; r++) {
proxy = x(r, _) ;
result[r] = as<double>(call.fast_eval() );
}
return result;
}
// [[Rcpp::export]]
NumericVector rowApply3(NumericMatrix& x, const Function& FUN)
{
int n = x.nrow();
NumericVector result = no_init(n);
NumericVector row( x.ncol() ) ;
Language call(FUN, row);
for (int r = 0; r < n; r++) {
row = x(r, _) ;
result[r] = as<double>(call.fast_eval() );
}
return result;
}
inline void grab_row( const NumericMatrix& x, int row, NumericVector& target, int ncol, int nrow){
for( int i=0, k = row; i<ncol; i++, k+= nrow)
target[i] = x[k];
}
// [[Rcpp::export]]
NumericVector rowApply4(NumericMatrix& x, const Function& FUN){
int n = x.nrow();
NumericVector result = no_init(n);
int ncol = x.ncol() ;
NumericVector row( ncol ) ;
Language call(FUN, row);
for (int r = 0; r < n; r++) {
grab_row( x, r, row, ncol, n ) ;
result[r] = as<double>(call.fast_eval() );
}
return result;
}
/*** R
library(microbenchmark)
options(digits = 3)
mean_ <- function(.) .Internal(mean(.))
M <- matrix(rnorm(15L), nrow=3L);
identical(rowMeans(M), apply(M, 1L, mean));
identical(rowMeans(M), rowApply0(M, mean));
identical(rowMeans(M), rowApply1(M, mean));
identical(rowMeans(M), rowApply2(M, mean));
identical(rowMeans(M), rowApply3(M, mean));
identical(rowMeans(M), rowApply3(M, mean_));
identical(rowMeans(M), rowApply4(M, mean_));
microbenchmark(rowMeans(M),
apply(M, 1L, mean),
rowApply0(M, mean),
rowApply1(M, mean),
rowApply2(M, mean),
rowApply3(M, mean),
rowApply3(M, mean_),
rowApply4(M, mean_)
)
M <- matrix(rnorm(1500L), nrow=30L);
microbenchmark(rowMeans(M),
apply(M, 1L, mean),
rowApply0(M, mean),
rowApply1(M, mean),
rowApply2(M, mean),
rowApply3(M, mean),
rowApply3(M, mean_),
rowApply4(M, mean_)
)
*/
$ RcppScript rowApply.cpp
> library(microbenchmark)
> options(digits = 3)
> mean_ <- function(.) .Internal(mean(.))
> M <- matrix(rnorm(15), nrow = 3)
> identical(rowMeans(M), apply(M, 1, mean))
[1] TRUE
> identical(rowMeans(M), rowApply0(M, mean))
[1] TRUE
> identical(rowMeans(M), rowApply1(M, mean))
[1] TRUE
> identical(rowMeans(M), rowApply2(M, mean))
[1] TRUE
> identical(rowMeans(M), rowApply3(M, mean))
[1] TRUE
> identical(rowMeans(M), rowApply3(M, mean_))
[1] TRUE
> identical(rowMeans(M), rowApply4(M, mean_))
[1] TRUE
> microbenchmark(rowMeans(M), apply(M, 1, mean), rowApply0(M,
+ mean), rowApply1(M, mean), rowApply2(M, mean), rowApply3(M,
+ mean), rowAppl .... [TRUNCATED]
Unit: microseconds
expr min lq median uq max neval
rowMeans(M) 6.89 9.28 10.30 11.5 22.2 100
apply(M, 1L, mean) 64.11 68.62 71.26 87.7 133.5 100
rowApply0(M, mean) 66.97 71.78 74.74 90.3 831.4 100
rowApply1(M, mean) 25.88 27.77 30.01 36.9 55.7 100
rowApply2(M, mean) 25.37 27.83 30.94 36.5 43.5 100
rowApply3(M, mean) 25.36 27.06 29.23 34.3 41.9 100
rowApply3(M, mean_) 6.37 7.75 8.78 10.6 25.4 100
rowApply4(M, mean_) 6.45 7.82 8.41 10.2 14.8 100
> M <- matrix(rnorm(1500), nrow = 30)
> microbenchmark(rowMeans(M), apply(M, 1, mean), rowApply0(M,
+ mean), rowApply1(M, mean), rowApply2(M, mean), rowApply3(M,
+ mean), rowAppl .... [TRUNCATED]
Unit: microseconds
expr min lq median uq max neval
rowMeans(M) 10.7 14.4 15.1 15.7 21.1 100
apply(M, 1L, mean) 350.6 362.0 371.2 382.7 2225.8 100
rowApply0(M, mean) 635.8 655.2 672.3 716.8 2595.1 100
rowApply1(M, mean) 225.1 233.8 239.7 249.8 2212.1 100
rowApply2(M, mean) 221.3 229.4 236.6 260.2 329.0 100
rowApply3(M, mean) 217.0 223.9 229.1 247.7 2077.2 100
rowApply3(M, mean_) 34.3 38.2 39.3 41.1 47.0 100
rowApply4(M, mean_) 34.1 36.7 38.0 39.3 47.1 100
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment