Expand description
SIMD-accelerated wavelet transforms for x86/x86_64 architectures
This module provides highly optimized SIMD implementations of wavelet transforms using AVX2 instructions when available, with automatic fallback to SSE2 or scalar.
Functionsยง
- dwt_
simd - Generic SIMD DWT dispatcher - delegates to type-specific implementations
- dwt_
simd_ f32 - dwt_
simd_ f64 - Fallback implementations for non-x86 architectures
- has_
avx2 - Check if AVX2 is available at runtime
- has_
sse2 - Check if SSE2 is available at runtime
- modwt_
convolve_ single - Auto-dispatching single-point convolution - uses SIMD if available
- modwt_
convolve_ single_ portable - Portable (non-SIMD) single-point convolution for streaming MODWT