[BUG] Neon performs worse than C

[hanyin-arm]

Hi, when we build and run your hello-neon sample app on my Pixel 2 (Snapdragon 835) and Pixel 9 (Tensor G4), both show a worse performance using Neon as shown below

Upon some initial investigation, our SIMD experts are worried about this sample code's FIR filter's Neon algorithm being sub-optimal, thus not leveraging Neon's potentials on performance boost.

Please let us know your thoughts and we can follow up either in public or private (e.g. via my @arm.com email)
Thank you and happy holidays 🎄