RE: [fpu] FPU operations

[Damjan Lampret <damjanlampret@yahoo.com>]

> Well, I thought it might be simpler if we would stretch the basic FP
> OPs to take all equal amount of time. So if fadd only needs 2 cycles

If fadd needs two then two it shall be. It really isn't important to be
all the same. More important is that they are finished as soon as
possible.

> 
> No, No, No !  I don't know how this came about, but I have never
> meant
> to suggest that the execution time depends on the input OPERANDS. I

You didn't. I was just careful to cover all possibilities (Jamil was
asking something in this direction). So I am saying CPU wouldn'y know
IF ... Basic clk cycles don't have to do with operands content
everything should be ok.

> >Result vs zero. Is it greater, lesser or equal (actually LT(GT) and
> EQ
> >are enough to always know GT(LT)).
> 
> No, that is not quite correct. If one operand is a NAN or INF, it is
> undetermined.
> 
Sure. You can include 'undetermined' also since it is one of the status
signals. Anyway 'undetermined' isn't evaluted by branch insns but
always triggers exception (is my interpretation right: SNAN triggers
exception and QNAN doesn't?).

> To compare against zero, should be easy, I could see how this would
> be
> very fast. Lets document this somewhere and implement it ...
> I'm thinking this would be a second comparator that would look at the
> intermediate result.
> 

Yeah. Lets.

> No, then the FPU would need to know about timing. I think it's
> simpler
> if the CPU handles all of that.
> 

I said it is ok. ;-)

--damjan



__________________________________________________
Do You Yahoo!?
Kick off your party with Yahoo! Invites.
http://invites.yahoo.com/