Enable vectorization on arm when compiled with -mfpu=neon
Tested on my raspberry pi. I did need to add -mfpu=neon
to the compile flags but otherwise this worked smoothly. SciPy test suite passes and multidimensional transforms show a speedup.
Tested on my raspberry pi. I did need to add -mfpu=neon
to the compile flags but otherwise this worked smoothly. SciPy test suite passes and multidimensional transforms show a speedup.