pub fn select_tier(m: u32, present: [bool; 5], exact_m: bool) -> Option<usize>Expand description
PURE tier decision: index into W4A16_BATCHM_WIDTHS of the narrowest
tier that both COVERS m rows and is present in the loaded target, or
None when this family cannot serve m (caller falls back to the tile
GEMMs / the wide tiers).
present[i] is whether W4A16_BATCHM_WIDTHS[i] resolved. exact_m is
exact_m_tiers_enabled, threaded in rather than read here so both
polarities are testable without touching process env or a latched
OnceLock.
Narrowest-that-covers, not exact-only: a target missing a tier must still
dispatch, and the next wider tier is bit-identical at the same m (same
template, MAX_M-independent per-row FMA chain) — only slower.