Re: [avr-libc-dev] fplib and MUL

[Joerg Wunsch]

As Peeter Vois wrote:

> 1) to use MUL commands for the CPU's that support it.

Sounds like a good idea.

> 2) to avoid division operations for log_2(x) calcultaion

This one, too...

> 3) Instead of division with constant to use multiplication for
> example by calculation of log_10(x),

...as well as this one.

> Some more ideas will come as I do this optimisation. My question now
> is that is someone doing something in this area?

Alas, that's the weak point.  We currently lack a person actively
maintaining fplib.  Almost all of the more serious open bug reports
affect fplib.  We're still seeking a volunteer here...


As Björn Haase wrote:

> > As first I will go and try
> > to make the multiplication and look if the device groups avr-gcc
> > entirely have or have not MUL instruction available.

> ... Summing up: All of the avr5 devices are
> assumed to have hw multipliers.

Actually, avr4 and avr5 do have the hw multiplier.  The difference
between both is that devices which only have/need RJMP/RCALL form
avr4, while those that need/have JUMP/CALL are avr5.

For preprocessor tests, the compiler sets the cpp macro __AVR_ARCH__
to the respective value, so the existence of a hardware multiplier can
(currently) be tested using

#if __AVR_ARCH__ >= 4
...
#endif

Note that the current scheme needs to be refined some day, as we
recently saw the first devices that don't fit anymore, and had to be
``downgraded'' into the next lower class: ATtiny13 and ATtiny2313.
Both have some features of avr4 cores (word instructions) yet don't
have a hardware multiplier, so they are effectively avr2.5 or
something like that...   I guess the same will apply to all future
ATtiny devices that appear on the market.  Currently, GCC cannot use
their ability to handle word instructions.

-- 
cheers, J"org               .-.-.   --... ...--   -.. .  DL8DTL

http://www.sax.de/~joerg/                        NIC: JW11-RIPE
Never trust an operating system you don't have sources for. ;-)