Lpc2000

> 6. One thing I would say is that for very small filters (e.g. 
> the 4-tap IIR mentioned below), we've found it more efficient to move
the
> data through the filter, rather than use pointers, as in:
> 
> reg[3] = reg[2];
> reg[2] = reg[1];
> reg[1] = reg[0];
> reg[0] = new;
>
> Hope this is of some help.

This is exactly what I did.  I think you come out ahead even with more
taps, especially if you use inline asm and LDM/STM to do the actual
move.

Also, it doesn't bridge the performance gap, but ARM is a 32-bit fixed
point processor, compared to most fixed point DSP's I know about which
are all 16- or 24-bit.  The extra bits are nice for IIR and other
applications where dynamic range can make a real difference.