Module simd_kernels 

SIMD-optimized kernel implementations for high-performance computation

These kernels use explicit SIMD intrinsics for critical hot paths.

Structs

KernelConfig

Configuration for SIMD kernel selection

KernelDispatcher

Kernel dispatcher that selects optimal implementation

Functions

batch_rolling_mean

Batch rolling operations for multiple series

batch_rolling_std

Batch rolling std for multiple series

cumsum_simd

Vectorized cumulative sum using chunked processing

dot_product

Vectorized dot product for correlation calculations

ema_simd

Vectorized EMA with better cache utilization

mean_simd

Vectorized mean

rolling_corr_simd

Rolling correlation with optimized computation

rolling_mean_simd

SIMD-optimized rolling mean using chunked processing

rolling_std_simd

SIMD-optimized rolling standard deviation

sum_simd

Vectorized sum