Various mathematical routines with SIMD acceleration (SSE/AVX/NEON) in the form of a compact C library. Designed without any OS portability in mind, tested only on Linux. Can work on BSD/Darwin after some potentially minor adaptation. Supports Android build / bionic.
- Conversion between int16_t, int32_t and float
- Parts of BLAS levels 1, 2, 3 with a completely different API (e.g., matrices, vectors, scalars)
- 1D convolution and correlation with best approach detection (naive, overlap-save, FFT)
- 1D peak detection
- sin, cos, log, exp (delegated to AVX mathfun and NEON mathfun)
- 1D and 2D normalization
- 1D decimated and stationary (undecimated) wavelets
./autogen.sh
mkdir build && cd build
../configure
make -j$(getconf _NPROCESSORS_ONLN)
make test
make install DESTDIR=...
By default, this library makes use of FFTF.
You can pass --disable-simd-fftf to configure to skip building dependent features.
Copyright © 2013 Samsung R&D Institute Russia