Use Arm Neon equivalent instructions to Intel VNNI

[amitdo]

... for int dot product.

[amitdo]

The equivalent for VNNI VPDPBUSD seems to be USDOT.

https://developer.arm.com/documentation/ddi0596/2021-12/SIMD-FP-Instructions/USDOT--by-element---Dot-Product-with-unsigned-and-signed-integers--vector--by-element--

There are some variants: SDOT, SUDOT, UDOT.

[amitdo]

CC: @robinwatts

[amitdo]

Arm's USMMLA instruction seems even better.

https://developer.arm.com/documentation/ddi0596/2021-12/SIMD-FP-Instructions/USMMLA--vector---Unsigned-and-signed-8-bit-integer-matrix-multiply-accumulate--vector--

I wonder if Intel has an equivalent instruction.