Function simd_gemv_cache_optimized_f32 

pub fn simd_gemv_cache_optimized_f32(
    alpha: f32,
    a: &ArrayView2<'_, f32>,
    x: &ArrayView1<'_, f32>,
    beta: f32,
    y: &mut Array1<f32>,
)

Cache-aware matrix-vector multiplication (GEMV) using unified interface