AVX-VNNI Int8 Inference

Demonstrate AVX-VNNI (VPDPBUSD) for Int8 inference acceleration on Intel Meteor Lake+ CPUs. Detects VNNI capability at runtime, compares Int8 vs FP32 throughput, and measures quantization error.

Device:

cargo run --example simd_avx_vnni_int8_inference --release

Key concepts: AVX-VNNI detection, symmetric int8 quantization, quantization error analysis, GOPS benchmarking.