24 threads ( 6 on each of 4 nodes)
-------------------------------------
  WRITE: bw=33.6GiB/s (36.1GB/s), 1433MiB/s-1507MiB/s (1503MB/s-1580MB/s), io=96.0GiB (103GB), run=2718-2858msec

32 threads (8 on each of 4 nodes)
-------------------------------------
  WRITE: bw=37.8GiB/s (40.6GB/s), 1248MiB/s-1369MiB/s (1309MB/s-1436MB/s), io=124GiB (133GB), run=2991-3282msec

48 threads (12 on each of 4 nodes)
-------------------------------------
  WRITE: bw=34.8GiB/s (37.3GB/s), 741MiB/s-1010MiB/s (777MB/s-1059MB/s), io=192GiB (206GB), run=4054-5525msec