Re: [f-cpu] registers

[Mohamed Ali Kilani <mohamed.kilani@m4x.org>]

Salut cedric,

Je viens de debarquer y a pas longtemps dans la planete F-CPU et j'ai 
commence par lire le manuel. J'ai recupere un Snapshot et j'ai regarde 
ou en etait les differentes EUs.

Je suis interesse par aider sur l'EU de la division entiere.
J'ai vu que t'as deja designe une unite qui gere au niveau 8 bits 
unsigned. J'ai vu aussi que tu obtenais le resultat en 7 cycles.

Je voulais savoir, quelles etaient les contraintes qui sont imposees au 
niveau timing, perf, taille a l'implementation etc.

J'ai pas mal de documentation de cours anciens que j'ai pris sur de 
l'arithmetique hardware. Je vais essayer de voir ce que je peux en sortir.

A bienot

Dali

Cedric BAIL wrote:
>>[You]
>>Nowadays we too often think about 64 bits register length. so 64 ==
>>the biggest int number. But in fact, the most interresting vector size is
>>256 bits.
> 
>  
> 
>>As you can see, using a register of 256 for storing an integer of 64
>>bits is a big waste ! Look at our programmation api. We always have 3
>>or 4 pointers in the register set, so 3/4 of the register will never be
>>used 95% of the time, what a waist, don't you think ?
> 
> 
>>{I}
>>Okay, single integer can be only in 64-bit even if physically they are
>>256 bit. So you prefer to split and use a different register-bank instead.
>>The trouble if we consider using vector integer registers are much more
>>underused than single integer registers, well we still waste memory
>>indeed for 95% of the time wherever the 3/4 of 256-bit for registers are
>>physically
>>located :).
> 
> 
> And perhaps more, because you always need to save all your register, so
> you have 256*63 + 64*63 = 20160 (2520 o => 2,5 Ko) instead of 
> 256*63 = 16128 (2016 o => 1,9 Ko). So you loose more place and you increase
> needed move when you work with SIMD.
> 
> 
>>Sparc V8 arch use 32 windows register (>100 registers but only 32 seeing
>>in a given time). When you add the fpu you receive 32 registers
>>dedicated to the fpu (the 32 fpu regiter are 32 bit but you could access
>>it by to for 64 bit double).
> 
> 
>>i have ask to our expert why adding new register set and not use the
>>integer register bank. I had in mind the f-cpu approach.
> 
> 
>>This answer was quite simple : "to use more register" !
> 
> 
> I think it's not the real reason. Because when you have 64 registers for
> int and float you can produce better code with less data in memory than with
> 32 registers for int and 32 registers for float. In a normal application
> you don't use a lot of float (and SIMD), so you can't use them for int (or 
> scalar). In fact by dividing the registers bank, you waste capacity from
> my point of view.
> 
> A+
>   Cedric
> *************************************************************
> To unsubscribe, send an e-mail to majordomo@seul.org with
> unsubscribe f-cpu       in the body. http://f-cpu.seul.org/
> ______________________________________________________________________
> Etudiant: Wanadoo t'offre le Pack eXtense Haut Débit soit 150,92 euros
> d'économies ! Clique ici : http://www.ifrance.com/_reloc/mail.etudiant 
> 



*************************************************************
To unsubscribe, send an e-mail to majordomo@seul.org with
unsubscribe f-cpu       in the body. http://f-cpu.seul.org/