Re: [f-cpu] FPU Adder....

[Nicolas Boulay <nico@seul.org>]

Le Mercredi 31 Décembre 2003 02:45, Michael Riepe a écrit :
> Nicolas Boulay wrote:
> [...]
>
> >>>"En" signal  must not be in the process sensitivity list of a clocked
> >>>process.
> >>
> >>Why not?
> >
> > There is no use of it. A synchrone process could only wake up by the
> > clock and by the asynchronous reset. So a modification of "En" wake up
> > the process. So you use more simulation time. Sometimes you could also
> > have very strange side effect du to delta cycle.
>
> Hmmm... why do some tools complain about an `incompletely specified
> sensitivity list' if you skip it, then?  IIRC, Synopsys DC was one of them.
>

Sensitivity list define the signal to be watched to "reexecute" the process 
during a delta cycle.(each process are reexecute until nothing change to jump 
to the next time stamp). So each signal whitch are read must be in the 
sensitivity list. But in synchronous process with asynchronous reset, things 
happen only on the change of the clock and on the reset. So it's only a 
slowdown of the simultion.

It could be much more dangerous in combinatory process. All read signal must 
be in the sensitivity list. If one miss, the change will not be taken into 
account in the simulation. Because the synthetiser never mind about 
sensitivity list, there could have a mismatch between simulation and the 
netlist generated.

So some compiler check for sensitivity list. MTI does it, but stupidely : it's 
ignore synchronous process.

nicO

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

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