topic Re: How to PowerQuad to implement powf function? in LPC Microcontrollers

How to PowerQuad to implement powf function?

Amendes — Thu, 20 Oct 2022 13:48:45 GMT

Hi,

Powerquad has the functions: ln(x) e^(x) and e^(-x). So I suppose it is possible to implement the powf function using Powerquad, I found it strange that NXP did not make a macro of it in the Powerquad API or commented on this possibility in the application note. Are there any restrictions on the range of values of x that make it impossible to implement the powf function using powerquad?

Re: How to PowerQuad to implement powf function?

Julián_AragónM — Mon, 24 Oct 2022 17:13:04 GMT

Hi @Amendes,

PowerQuad specializes in matrix operations, FFT, convolutions and more complex algorithms for Signal Processing.

Currently there is no implementation plans for the pow and powf functions in PowerQuad. I will ask with the development team for the compatibility with both functions.

The library "math.h", however, supports both pow and powf if you need to implement them in a project.

Best regards, Julian

Re: How to PowerQuad to implement powf function?

Amendes — Mon, 24 Oct 2022 17:45:02 GMT

Hi Julian,

I don't expect powerquad to be able to do the powf function directly. I didn't understand why NXP didn't create an API to do powf using ln(x) and e^x. I got a LPC55S69 kit and tested these functions, and I noticed that the only limitation is in the function e^(x) , where x cannot be too big, so that e^(x) is bigger than 2^32. Even so, it is very impressive

Then it is possible to implement powf, with the limitation that x^y is less than 2^32.

Re: How to PowerQuad to implement powf function?

frank_m — Tue, 25 Oct 2022 06:08:18 GMT

The (vendor-independant) ARM DSP-Lib used to provide an implementation.
Have you checked ?

It is a while since I worked with it.