Skip to main content

put_f64s

Function put_f64s 

Source
pub fn put_f64s(out: &mut Vec<u8>, v: &[f64])
Expand description

Append a float column’s wire bytes (little-endian, every NaN canonical) in one pass: each element is loaded, tested and stored once, the memory traffic of a copy. On an AVX2 host the pass runs four elements per instruction (an unordered compare and a blend).