// is reached and keep reusing the same source string - that
	// should therefore be always resident in the L1 cache - until we
	// have completed the construction of the result.
	// This yields significant speedups (up to +100%) in cases where
	// the result length is large (roughly, over L2 cache size).
	const chunkLimit = 8 * 1024
	chunkMax := n
	if chunkMax > chunkLimit {
		chunkMax = chunkLimit / len(b) * len(b)

            AVX512_BF16, AMX, etc.
            ([Intel Cascade Lake](https://www.intel.com/content/www/us/en/products/platforms/details/cascade-lake.html)
            and newer CPUs.)
            *   [Example performance speedups.](https://medium.com/intel-analytics-software/leverage-intel-deep-learning-optimizations-in-tensorflow-129faa80ee07)
        *   For older CPUs, oneDNN optimizations are disabled by default.

      totalCount += count;
      return this;
    }

    AvlNode<E> add(
        Comparator<? super E> comparator, @ParametricNullness E e, int count, int[] result) {
      /*
       * It speeds things up considerably to unconditionally add count to totalCount here,
       * but that destroys failure atomicity in the case of count overflow. =(
       */
      int cmp = comparator.compare(e, getElement());

     */
    // NB: yes, this is surprisingly high, but that's what the experiments said was necessary
    // Raising this number slows the worst-case contains behavior, speeds up hashFloodingDetected,
    // and reduces the false-positive probability.
    static final int MAX_RUN_MULTIPLIER = 13;

    /**