using C++ within R

June 8, 2014
By

(This article was first published on Dan Kelley Blog/R, and kindly contributed to R-bloggers)

Introduction

Quite often I write which(...)[1] to find the first element of a vector matching some condition. One has to wonder whether that’s wasteful, though, since there is no need to do any tests once one works. I decided to try using C++, using Rcpp, to see if speed advances could be made.

Procedure

1
2
3
4
5
6
library(Rcpp)
library(microbenchmark)
cppFunction("\n            int firstZero(IntegerVector x) {\n                int nx = x.size();\n                for (int i = 0; i < nx; ++i) {\n                    if (0 == x[i]) {\n                        return i+1;\n                    }\n                }\n                return 0; // means none found\n            }")
x <- rep(1, 10000)
x[seq.int(500, 10000)] <- 0
microbenchmark(firstZero(x), times = 1000L)
## Unit: microseconds
##          expr   min    lq median    uq   max neval
##  firstZero(x) 17.17 18.16  19.05 19.29 738.6  1000
1
microbenchmark(which(0 == x)[1], times = 1000L)
## Unit: microseconds
##              expr   min    lq median    uq   max neval
##  which(0 == x)[1] 31.74 33.26  33.95 35.99 740.1  1000

Results

The C++ method was nearly twice as fast. However, other tests (with different vector lengths, different fractions zeroed-out, etc) showed nearly identical times for the two methods.

Conclusions

In light of variations in test results, and the added complexity of including C++ code in an R program, I advise carrying out data-tailored benchmarks before deciding to use Rcpp.

Note that the test does not account for the time to compile the C++ program, which can outweigh time savings in small problems. However, this is irrelevant because one shouldn’t be worrying about optimization in small problems anyway, and large problems will likely involve package generation, which means that the C++ compilation will be done as the package is being built.

Resources

To leave a comment for the author, please follow the link and comment on his blog: Dan Kelley Blog/R.

R-bloggers.com offers daily e-mail updates about R news and tutorials on topics such as: visualization (ggplot2, Boxplots, maps, animation), programming (RStudio, Sweave, LaTeX, SQL, Eclipse, git, hadoop, Web Scraping) statistics (regression, PCA, time series, trading) and more...



If you got this far, why not subscribe for updates from the site? Choose your flavor: e-mail, twitter, RSS, or facebook...

Comments are closed.