skip gforce() setup when j is only .N

[m-muecke]

On small tables data.table seems to be a bit slower for DT[, .N, by = g] than dplyr. AFAICT, this is due to overhead from gforce() that is not needed. An alternative approach would've been returning a GForce_N element in the output list of .optimize_gforce() and checking with that.

master

library(data.table, warn.conflicts = FALSE)

set.seed(1)
n = 1e7
bench::press(
  groups = c(10, 1e6),
  {
    DT = data.table(g = sample.int(groups, n, TRUE), v = runif(n))
    df = as.data.frame(DT)
    bench::mark(
      data.table = DT[, .N, by = g],
      dplyr = dplyr::count(df, g),
      check = FALSE,
      iterations = 20,
      filter_gc = FALSE
    )
  }
)
#> Running with:
#>    groups
#> 1      10
#> 2 1000000
#> # A tibble: 4 × 7
#>   expression  groups      min   median `itr/sec` mem_alloc `gc/sec`
#>   <bch:expr>   <dbl> <bch:tm> <bch:tm>     <dbl> <bch:byt>    <dbl>
#> 1 data.table      10  37.33ms   38.1ms    22.8     268.8MB    22.8 
#> 2 dplyr           10  22.75ms   26.3ms    38.7      80.9MB     9.68
#> 3 data.table 1000000 107.77ms  112.4ms     8.01    297.6MB     8.01
#> 4 dplyr      1000000    3.67s     3.8s     0.260   219.7MB    13.9

^{Created on 2026-10-13 with reprex v2.1.1}

PR

library(data.table, warn.conflicts = FALSE)

set.seed(1)
n = 1e7
bench::press(
  groups = c(10, 1e6),
  {
    DT = data.table(g = sample.int(groups, n, TRUE), v = runif(n))
    df = as.data.frame(DT)
    bench::mark(
      data.table = DT[, .N, by = g],
      dplyr = dplyr::count(df, g),
      check = FALSE,
      iterations = 20,
      filter_gc = FALSE
    )
  }
)
#> Running with:
#>    groups
#> 1      10
#> 2 1000000
#> # A tibble: 4 × 7
#>   expression  groups      min   median `itr/sec` mem_alloc `gc/sec`
#>   <bch:expr>   <dbl> <bch:tm> <bch:tm>     <dbl> <bch:byt>    <dbl>
#> 1 data.table      10   17.8ms   18.5ms    52.5      39.9MB    18.4 
#> 2 dplyr           10   22.4ms   23.7ms    39.1      80.9MB    23.5 
#> 3 data.table 1000000   94.5ms   97.6ms    10.1      68.7MB     4.02
#> 4 dplyr      1000000     3.7s     3.9s     0.255   219.7MB    14.6

^{Created on 2026-10-13 with reprex v2.1.1}

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 99.02%. Comparing base (1d87418) to head (4b729cd).
⚠️ Report is 1 commits behind head on master.

Additional details and impacted files

@@           Coverage Diff           @@
##           master    #7920   +/-   ##
=======================================
  Coverage   99.02%   99.02%           
=======================================
  Files          88       88           
  Lines       17381    17383    +2     
=======================================
+ Hits        17211    17213    +2     
  Misses        170      170           

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.