pub const FP8_QUANT_HOPPER_GROUPS_PER_CTA: u32 = 8;Expand description
2026-09-25: K-groups one Hopper CTA covers: 128 threads / 16 threads per
group, as HQ_GROUPS_PER_CTA in the .cu. The kernel derives its group
span from gridDim.y, so this value sets only the grid’s Y extent.