Function dot_i8_batch 

pub fn dot_i8_batch(
    query: &[i8],
    vectors: &[i8],
    scales: &[f32],
    dim: usize,
    results: &mut [f32],
)

Compute dot products for a batch of vectors with dequantization.

Computes: result[i] = dot(query, vectors[i * dim..(i+1) * dim]) * scales[i]

Arguments

• query - Query vector (i8)
• vectors - Flattened database vectors (i8, n_vec × dim)
• scales - Per-vector dequantization scales
• dim - Dimension of each vector
• results - Output dequantized dot products

Panics

Panics if buffer sizes are inconsistent