Module vector_ops 

SIMD-accelerated vector operations for CPU fallback paths

Provides vectorized element-wise add, multiply, scale, dot product, and reduction operations. Architecture-specific implementations are selected at compile time via cfg(target_arch), with a scalar fallback for unsupported platforms.

Functions

vector_add_f32

Element-wise addition: c[i] = a[i] + b[i]

vector_dot_f32

Dot product: sum(a[i] * b[i])

vector_mul_f32

Element-wise multiplication: c[i] = a[i] * b[i]

vector_reduce_sum_f32

Sum reduction: sum(a[i])

vector_scale_f32

Scale every element: c[i] = a[i] * scalar