pub fn conv_3x3_simd(
input: &[f32],
kernel: &[f32],
output: &mut [f32],
in_h: usize,
in_w: usize,
in_c: usize,
out_c: usize,
stride: usize,
padding: usize,
)Expand description
SIMD-accelerated 3x3 convolution with automatic architecture dispatch