Re: [oc] async/sync reset.

[Rudolf Usselmann <rudi@asics.ws>]

On Thu, 2003-04-10 at 20:30, cyrano@nerim.net wrote:
> Rudolf Usselmann <rudi@asics.ws> a écrit :
> 
> > On Thu, 2003-04-10 at 07:20, nico wrote:
> > > In most design guides, reset must be asynchronous to avoid hasard at
> > 
> > Oh really ?!  What hazard would that be ?
> > 
> 
> Mostly, the behavior of the output of the chip, during the power up and the beginning of the clock.
> 
> > > power up. But in manual of Synopsys Design Compiler, they incourage the
> > > use of synchronous reset.
> > 
> > Async resets are very dangerous and very difficult
> 
> :) Funny. It's seems that it is another religious war. In the lab where i work, sync reset are forbiden.

Not really a religious war, just something that people
tend not to think about at all and just apply some "design
guide". There are very good reasons to do async reset and
there are very good reasons to do sync reset. Async reset
is a bit trickier to do correctly than sync reset (IMHO,
anyway).

> > to get right (specially if all you do is believing
> > some "guidelines" how to do things right).
> > 
> > Avoid using async resets unless you absolutely have
> > too AND fully understand the implication of it and
> > how to do it right.
> > 
> > Assertion of Async reset is not a problem it's the
> > de-assertion of it that has to be done right. Even
> > if you do it right many design tools have problems
> > with async reset. Static timing analysis (part of
> > Synopsys Design Compiler) is on of those tools.
> > 
> 
> So like an async set of the reset but  sync unset of the reset ?

Exactly. You assert async reset asynchronously but de-assert
it synchronously. This is probably one of the most comon
mistakes when using async reset. 

> > 
> > > So, what do you use and why ?
> > > 
> > > nicO
> > 
> > Here is a link to an excellent white paper discussing
> > all the pros and cons of Sync/Async reset.
> > 
> > http://www.sunburst-design.com/papers/CummingsSNUG2002SJ_Resets_rev1_1.pdf
> > 
> 
> i will read it thanks.

Enjoy !

> nicO


Regards, 
rudi
------------------------------------------------
www.asics.ws   - Solutions for your ASIC needs -
FREE IP Cores  -->   http://www.asics.ws/  <---




--
To unsubscribe from cores mailing list please visit http://www.opencores.org/mailinglists.shtml