r_benchmark

data_read.R

dta <- palmerpenguins::penguins
ids <- sample(nrow(dta), 1e7,  replace = TRUE)
dta <- dta[ids,] |> dplyr::mutate(across(where(is.factor), as.character))

tmpf_csv <- tempfile(pattern = '.csv')
data.table::fwrite(dta, tmpf_csv, nThread = 10)
tmpf_parquet <- tempfile(pattern = '.parquet')
arrow::write_parquet(dta, tmpf_parquet)
tmpf_qs <- tempfile()
qs::qsave(dta, tmpf_qs, nthreads = 10)
tmpf_fst <- tempfile()
fst::write_fst(dta, tmpf_fst)

n_iterations <- 20
res <- bench::mark(
        'readr'      = readr::read_csv(tmpf_csv, show_col_types = FALSE, progress = FALSE),
        'data.table' = data.table::fread(tmpf_csv, nThread = 10),
        'arrow'      = arrow::read_csv_arrow(tmpf_csv),
        'parquet'    = arrow::read_parquet(tmpf_parquet),
        'qs'         = qs::qread(tmpf_qs, nthreads = 10),
        'fst'        = fst::read_fst(tmpf_fst),
        check = FALSE,
        iterations = n_iterations
)
res |> subset(select = c('expression', 'min', 'median', 'mem_alloc'))

#### AMD RYZEN 5 3600 (6c, 12t)
#   expression      min   median mem_alloc
#   <bch:expr> <bch:tm> <bch:tm> <bch:byt>
# 1 readr         3.66s    3.69s  611.88MB
# 2 data.table 959.47ms 987.13ms   558.6MB
# 3 arrow      310.33ms 319.15ms  118.64MB
# 4 parquet    283.04ms 303.39ms    1.08MB
# 5 qs            1.52s    1.53s  495.91MB
# 6 fst        758.77ms 764.96ms  495.92MB

#### AMD RYZEN 9 5900X (12c, 24t)
#   expression      min   median mem_alloc
#   <bch:expr> <bch:tm> <bch:tm> <bch:byt>
# 1 readr         2.22s    2.25s   612.5MB
# 2 data.table  580.9ms    584ms  556.05MB
# 3 arrow         1.41s    1.54s  118.64MB
# 4 parquet    627.74ms  640.3ms    1.08MB
# 5 qs          917.3ms 938.59ms  495.91MB
# 6 fst        702.07ms 712.91ms  495.96MB

lapply(paste0('tmpf_', c('csv', 'parquet', 'qs', 'fst')), unlink)
rm(list = ls())
gc()