Closed christophe0606 closed 3 years ago
In the README, a custom arm_math.h
is mentioned.
I was interested in reading the source code to see if a similar approach can be used to integrate CMSIS-DSP in a ceedling-based testing framework, but it looks like it's missing from CMSIS/DSP/PythonWrapper/cmsisdsp_pkg/src folder.
@maugar This custom arm_math.h is no more needed. (I'll update the README).
Now you can use the standard one. If you compile with MSVC it will work. It you use another compiler you may define __GNUC_PYTHON__
This is needed to redefine following macros and select C implementation of some intrinsincs like __SSAT
__ALIGNED(x)
__STATIC_FORCEINLINE
__STATIC_INLINE
LOW_OPTIMIZATION_ENTER
LOW_OPTIMIZATION_EXIT
IAR_ONLY_LOW_OPTIMIZATION_ENTER
IAR_ONLY_LOW_OPTIMIZATION_EXIT
you may define
GNUC_PYTHON
This information was very useful, thanks 🙏
I think it's worth sharing that, as of today, support for GNUC_PYTHON
define is implemented in develop branch and I was using the master branch.
@maugar Right ! It is a recent change and not yet in the main branch.
is there any chance that this will make it on PyPi ?
@LukeGary462 At some point yes. But I need to improve the setup.py script.
Dear CMSIS-DSP users,
I have recently pushed an experimental Python wrapper for CMSIS-DSP which is compatible with Numpy and which can be used on any laptop/desktop.
You can build your signal processing flow with scipy then progressively replace some part of it with CMSIS-DSP and experiment with saturation, rounding effects until you have something ready for a board.
More details are in the README in folder DSP/PythonWrapper. I am not a Python expert and the installation procedure can be improved.
Also it is a very first release and I have not tested all the functions in the wrapper although the main ones (like biquad, fir, FFT …) were tested.
I hope it will be useful and I am waiting for your feedback.