Skip to main content

nvfp4_gemv_zero_heap

Function nvfp4_gemv_zero_heap 

pub fn nvfp4_gemv_zero_heap(
    input: &[f32],
    packed_matrix: &[u8],
    block_scales: &[u8],
    global_scale: f32,
    rows: usize,
    cols: usize,
    out: &mut [f32],
) -> Result<(), Nvfp4Error>
Expand description

Compute Matrix-Vector multiplication $y = A \cdot x$ where $A$ is in NVFP4 format.

Zero-heap hot path: unpacks weights into stack scratch [f32; NVFP4_BLOCK_SIZE] per block, computing the dot product incrementally without heap allocation.