Re: [f-cpu] F-CPU architecture...

[Yann Guidon <whygee@xxxxxxxxx>]

hi,

Rares Marian wrote:

> Just a bizarre thought:
> What if instead of moving data, the registers reassigned their own addresses so that data would appear to have moved?
>  
you mean : "register renaming" ?
in FC0, this would almost kill the remaining performance ...
This would add one or two stages at the beginning of the pipeline,
thus doubling the penalty of a branch.
So we'll have to include branch prediction etc....
and i don't count other things like all the register renaming mechanisms
which would consume even more silicon.
Now, balance that with the performance gain
which i estimate is near zero for FC0.

YG


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