FIR Convolution algorithm error (DSP-120) about esp-dsp

[TD-er]

If you're really into speed optimizations, then you may also want to look at lines like these:

for (int n = fir->pos; n < fir->N ; n++) {

Using ++n in a loop is faster.
Make as much as possible const in the invariant.

Thus:

const int fir_N = fir->N;
for (int n = fir->pos; n < fir_N; ++n) {

[dmitry1945]

Hi @Davidliujx,
We've changed the fir->coeffs[coeff_pos--] to the fir->coeffs[coeff_pos++] for speed optimization reason. The implementation In this way is just faster.
But, for this, the order of your coefficients should be reversed.
I will add the comment to the header about coefficients order.
Thank you,
Dmitry

[dmitry1945]

@TD-er yes, will add this to the list (it's not only FIR feature).

Thank you,
Dmitry

[Davidliujx]

thanks for your all reply