Skip to main content

vector_add_cpu

Function vector_add_cpu 

Source
pub fn vector_add_cpu(a: &[f32], b: &[f32]) -> Vec<f32>
Expand description

The pure-Rust floor of the cascade. Infallible and correct.

Length rule: the output is as long as the SHORTER input (zip stops there). The GPU path enforces equal lengths and errors on mismatch; to keep the two twins returning the same thing, callers should only ever pass equal-length slices. Given equal lengths, this and the GPU kernel agree exactly.