Module simd 

SIMD-accelerated wavelet transforms.

The DWT Haar 2-tap kernels dispatch through pulp, which runtime-selects the best available ISA (AVX2 / AVX-512 on x86_64, NEON on aarch64, scalar elsewhere) from a single SIMD-generic implementation. The MODWT streaming kernels use the same pulp substrate for their vectorizable dot-product path.

Functions

dwt_portable_simd_f64

Scalar fallback implementation for portable SIMD tests.

dwt_simd

Generic SIMD DWT dispatcher — routes to the type-specific entry point that computes in T’s own precision domain.

dwt_simd_f32

SIMD DWT for f32 signals — see dwt_simd_f64 for routing semantics.

dwt_simd_f64

SIMD DWT for f64 signals.

modwt_convolve_single

Auto-dispatching single-point convolution.

modwt_convolve_single_portable

Portable (non-SIMD) single-point convolution for streaming MODWT