Re: [sv-bc] Minutes: Nov 8 SV-BC CC

Doug Warmke writes --
> Currently I lean toward x's for hardware design friendliness.

I agree. In synthesis, you're going to get x's for the initial
value of a static enum variable, even if you use an explicit
declaration initialization.

If the LRM says the variable should get magically initialized to
some other value, synthesis has to issue a warning about a potential
mismatch between simulation and synthesis, unless it can somehow
detect that the variable will not actually be read until it has
a stable value. So, as a practical matter, unless the LRM is
changed, users will have to get such warning messages on every
static enum variable, even if they studiously avoid declaration
initializations.

The use model I'd recommend would be to begin with the default base
type of int while working out the design, then sometime before
final synthesis move to a smaller 4-state base type that reflects
what one wants in hardware.

>If we go with x's for 4-state enum value initialization,
>then I would prefer to see inconsistency with 2-state initialization
>than to see 2-state enum variables forcibly initialized to all 0's.

I agree. Two-state should make it easy to explore the design space.
Four-state and two-state are for different purposes, so why do they need
to be 'consistent'?

-- Brad

> -----Original Message-----
> From: owner-sv-bc@eda.org [mailto:owner-sv-bc@eda.org] On
> Behalf Of Maidment, Matthew R
> Sent: Monday, November 08, 2004 1:37 PM
> To: sv-bc@eda.org
> Subject: [sv-bc] Minutes: Nov 8 SV-BC CC
>
> The minutes from the Nov 8 CC have been posted here:
>
> http://www.eda.org/sv-bc/minutes/sv-bc_04_11_08.txt
>
> Per my action item, previous meeting minutes have also
> been updated here:
>
> http://www.eda.org/sv-bc/minutes/
>
>
> Matt
> --
> Matt Maidment
> mmaidmen@ichips.intel.com
>
>
>
Received on Tue Nov 9 12:02:39 2004