Benchmarks

library(tinycodet)
#> Run `?tinycodet::tinycodet` to open the introduction help page of 'tinycodet'.
loadNamespace("bench")
#> <environment: namespace:bench>
loadNamespace("ggplot2")
#> <environment: namespace:ggplot2>

 

Introduction

Some effort has been made to ensure the functions in ‘tinycodet’ are well optimized. The string related functions, for example, are about in the same order of magnitude in terms of speed as the stringi functions they call.

Here some speed comparisons are given, using the ‘bench’ package.

 

stri_locate_ith

stri_locate_ith() has about the same performance as stri_locate_all():

n <- 5e4
x <- rep(paste0(1:50, collapse = ""), n)
p <- "1"
i <- sample(c(-50:-1, 1:50), replace=TRUE, size = n)
bm.stri_locate_ith_vs_all <- bench::mark(
  "stri_locate_ith" = stri_locate_ith_fixed(x, p, i),
  "stringi::stri_locate_all" = stringi::stri_locate_all(x, fixed = p),
  min_iterations = 500,
  check = FALSE
)
summary(bm.stri_locate_ith_vs_all)
ggplot2::autoplot(bm.stri_locate_ith_vs_all)

#> # A tibble: 2 × 6
#>   expression                    min   median `itr/sec` mem_alloc `gc/sec`
#>   <bch:expr>               <bch:tm> <bch:tm>     <dbl> <bch:byt>    <dbl>
#> 1 stri_locate_ith            15.4ms   16.7ms      59.4     781KB     21.5
#> 2 stringi::stri_locate_all   14.4ms   16.1ms      61.6     391KB     14.4
#> Loading required namespace: tidyr

 

strcut

The strcut_ functions from ‘tinycodet’ have somewhat similar performance as the boundary functions provided by ‘stringi’:

n <- 1e5
x <- rep("hello", n)
i <- sample(1:3, n, replace = TRUE)
loc <- stri_locate_ith(x, i=i, regex="a|e|i|o|u")
bm.strcut <- bench::mark(
  "strcut_loc" = { strcut_loc(x, loc) },
  "strcut_brk" = { strcut_brk(x, type = "", tolist = TRUE) },
  "stringi::stri_split_boundaries" = {
    stringi::stri_split_boundaries(x, type="character")
  },
  min_iterations = 500,
  check = FALSE
)
ggplot2::autoplot(bm.strcut)

 

Row/columns-wise re-ordering

re-ordering rows/columns using the %row~% and %col~% operators provided by ‘tinycodet’ is a faster than doing so using traditional loops:

tempfun <- function(x, margin) {
  if(margin == 1) {
    for(i in 1:nrow(x)) x[i,] <- sort(x[i,])
    return(x)
  }
  if(margin == 2) {
    for(i in 1:ncol(x)) x[,i] <- sort(x[,i])
    return(x)
  }
}

mat <- matrix(sample(seq_len(n^2)), ncol = n)
bm.roworder <- bench::mark(
  tinycodet = mat %row~% mat,
  "base R" = tempfun(mat, 1),
  min_iterations = 250
)
ggplot2::autoplot(bm.roworder)

bm.colorder <- bench::mark(
  tinycodet = mat %col~% mat,
  "base R" = tempfun(mat, 2),
  min_iterations = 250
)
ggplot2::autoplot(bm.colorder)