Module kernels 

INT8 Quantized Kernels Module

Provides SIMD-optimized INT8 kernels for:

• 2D Convolution (standard and depthwise)
• Matrix multiplication (GEMM)
• Dot product operations

Architecture Support

• x86_64: AVX2 with _mm256_maddubs_epi16 and _mm256_madd_epi16
• aarch64: ARM NEON with vmull_s8 and vpadalq_s16
• wasm32: WebAssembly SIMD128 with i8x16 operations

Modules

int8_avx2

AVX2 INT8 SIMD Kernels for Quantized CNN Operations

int8_neon

ARM NEON INT8 SIMD Kernels for Quantized CNN Operations

int8_wasm

WASM SIMD128 INT8 Kernels for Quantized CNN Operations