Functions§
- bernoulli_
compare_ ⚠avx2 - Compare 32 random bytes against an unsigned threshold and return bit mask.
- dot_
f64_ ⚠avx2 - Dot product of two f64 slices using AVX2 FMA.
- fused_
and_ ⚠popcount_ avx2 - Fused AND+popcount over packed words using AVX2 for the AND stage.
- fused_
xor_ ⚠popcount_ avx2 - Fused XOR+popcount over packed words using AVX2 for the XOR stage.
- hamming_
distance_ ⚠avx2 - Hamming distance between two packed bitstream slices using AVX2.
- max_
f64_ ⚠avx2 - Maximum of f64 slice using AVX2.
- pack_
avx2 ⚠ - Pack u8 bits into u64 words using AVX2 movemask.
- popcount_
avx2 ⚠ - Count set bits in 64-bit words using AVX2.
- scale_
f64_ ⚠avx2 - Scale f64 slice in-place: y[i] *= alpha, using AVX2.
- softmax_
inplace_ ⚠f64_ avx2 - In-place softmax using AVX2 for max, sum, and scale steps.
- sum_
f64_ ⚠avx2 - Sum of f64 slice using AVX2.