Hi all,
I recently found that adding my code to a simple example where the audio input
is looped back to the headphones changes the output, which I suspect is caused
by the fact that the code I use takes too long to run. One reason this may be
caused is the fact that I run multiple filters, which are all implemented using
floating data types. Before attempting a fixed point implementation however, I
figured I could perhaps compute how long it takes my code to run.
I'm currently searching for some form of documentation that could perhaps
help me figure out how many cycles it takes the module I'm using (PEP5416)
to perform floating type multiplies and adds. If anyone knows of any such
document, I would be most grateful to learn of it.
Otherwise, if anyone has any other suggestions to alleviate this problem, I
would also appreciate suggestions. The only possible fixes I know of are fixed
point implementation or configuring the codec and DSP clocks to a higher
frequency. Thanks.