- perf: slightly improve codegen for
hash_valid_kmers_simdby avoidingzip. (Instead just call.next()manually.) - doc: update readme to link to
ensure_simddocs. - misc: cargo update
- perf: precompute initial hash values.
- perf: more
inline(always) - perf: Avoid slow
u32x8::splaton NEON. - perf: bump to packed-seq 4.2.0
- feat: Use
PackedNSeqwrapper type introduced bypacked-seq4.1.0. - feat: Add
scalarfeature that disabled thepacked-seqerror when native compilation is disabled and no SIMD intrinsics are found. - perf: Add
inline(always)totable_lookupintrinsic for better perf with thin-lto.
- Copy from
simd-minimizers - Add
KmerHashertrait; makeka parameter of the hasher itself. - Add generic to
NtHashto rotate more than 1.