`:=` vs 'set' in data.table

DT_comp_set.R

require(data.table)

set.seed(1L)
DT1 <- data.table(x=sample(1e7), y=as.numeric(sample(1e7)), z=sample(letters, 1e7, TRUE))
DT2 <- copy(DT1)

val <- runif(1e7)

# 'set' seems faster when adding 1-column
# =======================================

# on 1 column
system.time(set(DT1, i=NULL, j='bla', value=val)) # ----------- 0.195 secs
system.time(DT2[, bla := val]) # ------------------------------ 0.235 secs

# ':=' seems faster when adding 10-columns (due to for-loop? move to "C" for 'set'?)
# ==================================================================================

# 'set' ---------------------------------- 1.896 secs
cols <- paste("bla", 1:10, sep="")
system.time({
  for (j in cols) {
    set(DT1, i=NULL, j=j, value=val)
  }
})

# ':=' ----------------------------------- 1.645 secs
system.time({
    DT2[, (cols) := val]
})