For mid range keys, XXH3 uses a Mum-hash variant. */

* At very short lengths, there isn't enough input to fully hide secrets, or use * the entire secret. * * There is also only a limited amount of mix

* This is a stronger avalanche, * inspired by Pelle Evensen's rrmxmx * preferable when input has not been previously mixed */

* Frees an aligned pointer allocated by XXH_alignedMalloc(). Don't pass * normal malloc'd pointers, XXH_alignedMalloc has a specific data layout. */

* Malloc's a pointer that is always aligned to @align. * * This must be freed with `XXH_alignedFree()`. * * malloc typically guarantees 16 byte al

* In XXH3_scalarRound(), GCC and Clang have a similar codegen issue, where they * emit an excess mask and a full 64-bit multiply-add (MADD X-form).

! * A polyfill for POWER9's vec_revb(). */

* bench_hash_internal(): * Benchmarks hashfn repeateadly over single input of size `size` * return: nb of hashes per second */

* @return: hashID (necessarily between 0 and NB_HASHES) if present * -1 on error (hname not present) */

The wyrand PRNG that pass BigCrush and PractRand

wrapper for pthread interface

@return : 0 on success, 1 on error */

POOL_thread() : * Work thread for the thread pool. * Waits for jobs and executes them. * @returns : NULL on failure else non-null. */

returns time span in microseconds */

! @ingroup XXH3_family */

! @ingroup XXH3_family */

* A bit slower than XXH3_mix16B, but handles multiply by zero better. */

! @ingroup XXH3_family */

! @ingroup XXH3_family */

! @ingroup XXH3_family */

* It's important for performance to pass @p secretLen (when it's static) * to the compiler, so that it can properly optimize the vectorized loop. *

* It's important for performance that XXH3_hashLong is not inlined. */

! @ingroup XXH3_family */

! @ingroup XXH3_family */

! @ingroup XXH32_family */

! @ingroup XXH32_family */

! @ingroup XXH32_family */

! @ingroup XXH32_family */

! @ingroup XXH32_family */

! * @internal * @brief Processes a 64 byte block of data using the scalar path. */

! @ingroup XXH3_family */

! @ingroup XXH3_family */

* It's important for performance that XXH3_hashLong() is not inlined. */

* It's preferable for performance that XXH3_hashLong is not inlined, * as it results in a smaller function for small data, easier to the instruction