[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

6/8 bits imm instructions ( was Re: Rep:Re: [f-cpu] statistics of direct indexing usage)



On Wed, 8 Jan 2003 15:27:48 +0100 (CET)
devik <devik@cdi.cz> wrote:

> > Is it what you wanted to know ?
> >
> > >>><mode=chieur>
> > Not really. :) It's an interresting number however. The idea was
> > better to add the 6 bits immediat adressing mode rather than
> > _replace_ the 8 bits mode.
> >
> > By using 8 bits immediat, you can't use some flags combinaison. So
> > what is the impact in the generated code (how many instruction would
> > like to use this particular combinaison?).
> 
> uh uh, can you give me please some examples of such instruction ?

It's the problème of bit 12-13 (manual 0.2.5). 

It's use in the scan instruction to bit-reverse the input or negate it.
It used as negate sign flag of the MAC instruction. All of this flag
can't be used in the 8 bits immediat of such fonction. So is a pain or
not ?

> or you wanted to have BOTH 6bit and 8bit immediates and see which
> will be used ? I'm a bit lost - can't still understand .. :)

We must take into account that if not respecting fields boundary in
the instruction word will complicate the decoder, and make the ISA not
orthogonal. But in other hand, 8 bit is fewer than 6 :)

Using both could be too much. Maybe loadconst could do the job.

nicO

> devik
> 
> *************************************************************
> To unsubscribe, send an e-mail to majordomo@seul.org with
> unsubscribe f-cpu       in the body. http://f-cpu.seul.org/
> _____________________________________________________________________
> Envie de discuter en "live" avec vos amis ? T_l_charger MSN Messenger
> http://www.ifrance.com/_reloc/m la 1_re messagerie instantan_e de
> France
*************************************************************
To unsubscribe, send an e-mail to majordomo@seul.org with
unsubscribe f-cpu       in the body. http://f-cpu.seul.org/