Subject: FW: Fwd: Implicit FSMs style with mutiple clocks
From: J. Bhasker (jbhasker@Cadence.COM)
Date: Fri Dec 14 2001 - 11:20:21 PST
-----Original Message-----
From: owner-siwg@eda.org [mailto:owner-siwg@eda.org]On Behalf Of Jonas
Nilsson
Sent: Friday, December 14, 2001 11:41 AM
To: siwg@vhdl.org
Cc: Vinaya Singh
Subject: Re: Fwd: Implicit FSMs style with mutiple clocks
I agree with Vinaya.
The great thing with an implicit FSM is that the control is embedded
into the dataflow description, which in many cases can make the code
much more readable.
It's even possible to mix the two, embedding local "implicit" state
into the explicit states in an ordinary FSM, to describe mutlicycle
operations in an ordinary FSM. This works fine in synthesis tools too.
Implicit FSMs are great for some purposes, but they should of course
not be used in cases where a "normal" explicit FSM would be better.
Regards,
Jonas Nilsson
Vinaya Singh wrote:
> I agree that implicit FSM is not a great way to write a controller. If
you have
> state transition diagram, the best way to code is explicit machine.
>
> However, 'implicit FSM' is great way to model, multi-cycle
data-paths,
> DSP algorithms etc. That is the data path devices, where control is
implicit and
> embedded into it.
>
> Any voice from design community on this.
-- Jonas Nilsson, HARDI Electronics AB Derbyvagen 6B, SE-212 35 MALMO, SWEDEN Phone: +46-40-59 29 00 Fax: +46-40-59 29 01 E-mail: jonas@hardi.se WWW: http://www.hardi.se
This archive was generated by hypermail 2b28 : Fri Dec 14 2001 - 11:26:46 PST