Speex dev - Aug 2008 - fir_mem16,iir_mem16 and filter_mem16 optimisations

Hi!

I have some questions about that functions: fir_mem16, iir_mem16 and
filter_mem16.
Filtering is very slow on TI DSP, and i want to optimise it. 
Can somebody give me formulas which discribe work of this filters?
Or any suggestions about how to transform code for better performance.

I going to implement this functions in assembler, but it is hard to do without
full understanding how functions work.

??????? ??????? a ?crit :
> I have some questions about that functions: fir_mem16, iir_mem16 and
filter_mem16.
> Filtering is very slow on TI DSP, and i want to optimise it. 
> Can somebody give me formulas which discribe work of this filters?
> Or any suggestions about how to transform code for better performance.
> 
> I going to implement this functions in assembler, but it is hard to do
without full understanding how functions work.
These are direct-form II transposed filters. There's actually two ways
to compute them. For an alternate way, have a look at the commented
version of filter_mem2() completely at the bottom of filters_bfin.h. The
only thing you won't need from that one are the shifts by SIG_SHIFT
because the inputs and outputs of filter_mem16() are already 16 bits.

	Jean-Marc

-----Original Message-----
From: Jean-Marc Valin <jean-marc.valin at usherbrooke.ca>
To: ??????? ??????? <altersoft at mail.ru>
Date: Sat, 02 Aug 2008 07:54:34 -0400
Subject: Re: [Speex-dev] fir_mem16,iir_mem16 and filter_mem16 optimisations
> 
> ??????? ??????? a ?crit :
> > I have some questions about that functions: fir_mem16, iir_mem16 and
filter_mem16.
> > Filtering is very slow on TI DSP, and i want to optimise it. 
> > Can somebody give me formulas which discribe work of this filters?
> > Or any suggestions about how to transform code for better performance.
> > 
> > I going to implement this functions in assembler, but it is hard to do
without full understanding how functions work.
> 
> These are direct-form II transposed filters. There's actually two ways
> to compute them. For an alternate way, have a look at the commented
> version of filter_mem2() completely at the bottom of filters_bfin.h. The
> only thing you won't need from that one are the shifts by SIG_SHIFT
> because the inputs and outputs of filter_mem16() are already 16 bits.
> 
> 	Jean-Marc
> 
I still in doubt.
You could tell in more detail about filter_mem16()? 
I can't find comments in filter_mem2().
I think, filter_mem16 is a IIR filter, and const spx_coef_t *num, const
spx_coef_t *den is a input and output signal coefficients, but i am not sure...
is it right?
But there is only one coefficients array in iir_mem, and logic is not clear for
me.
I feel, fir_mem16() and iir_mem16() is a two parts of one IIR filter. fir_mem16
multiplies input signal, and iir_mem16 doing that with output...
So, if we call fir_mem16() and then iir_mem16() with corresponding parameters it
will be equivalent to filter_mem16()?

I'm sorry for a stupid questions, i am newbie in digital filtering.
May be filter_mem2() will work better, but i am trying to understand approach.
For me is interesting not only result in better performance of code, but i want
to get some aducational effect.

Thanks a lot )) 
If it is interesting for you, i can send you optimized functions after i shall
finish with it.