Sequential data table benchmark V2

library(data.table)

args <- commandArgs(trailingOnly=TRUE)
n <- if (length(args) > 0) as.integer(args[1]) else 1000000
m <- if (length(args) > 1) as.integer(args[2]) else ceiling(n / 100)
set.seed(107)
td <- data.table(val=rnorm(n), id=sample(m, n, replace=TRUE))

start <- proc.time()[3]
td[, means := mean(td$val[-.I]), by = id]
elapsed <- proc.time()[3] - start

res <- td$means
cat("Sequential data table version:\n")
cat(sprintf("Rows: %d, Unique IDs: %d\n", length(td$id), length(unique(td$id))))
cat(sprintf("Elapsed time: %f\n", elapsed))

library(digest)
cat(sprintf("MD5 hash: %s\n", digest(res)))