Re: [Iverilog-devel] Cosimulation with Icarus

[Kevin C. <iv...@gr...>]

I think this is the old code -

http://www.v-ms.com/download/ic-cap-i386-2009-06-03.tbz

---

One of the problems with a Verilog/SPICE simulation is that it is at the two extremes of modeling circuits, it helps to consider various levels of modeling in between, and separating out concepts like strength and certainty from values (1/0) - also potential & flow on the analog side.

ADMS is a bit of an abortion, if you want to do Verilog-A style modeling, it's probably better to translate to C++ and use something like GINAC -

https://www.ginac.de/

[ as an aside I do believe in software patents, Verilog descriptions of ICs are software, algorithms are inventions. ]

Simulators that don't do speculative stuff are "conservative" and don't back-track. However, back-tracking is not too hard if you are only dealing with FSMs (RTL/gate-level), although I've never been convinced it has much value.

Kev.

On 01/21/2018 05:13 AM, Felix Salfelder wrote:
> On Sun, Jan 21, 2018 at 01:29:14AM -0800, Kevin Cameron wrote:
>> I handed over working code (still have it around somewhere), got it back
>> converted to a "plugin" (unnecessarily), and not working.
> Hi Kevin.
>
> I have never seen the working (non-plugin?) version of this code. If you
> insist that it worked, and still have it around, then may I please ask
> you to share it? Beat, in particular could be interested.
>
> I presumably only know the code after you "got it back". I was not even
> around at earlier times. As the "conversion to plugin" usually does not
> affect functionality, your claim of having "working code" again pricks
> up my ears. and i must say, i am (naturally) skeptical.
>
> The code that I have seen tried to link (i.e. "cosimulate") two separate
> event queues. is that not the idea behind the "working code"? Now in
> reply to Beat, you write "simulation as co-simulation [it] generally
> doesn't go well"...
>
> if cosimulation does not go well (which i believe is true), it looks
> more promising to implement a tgt-$simulator for Icarus. this is
> off-topic in a cosimulation thread. but examples exist, these just dont
> fully work and need work.  tgt-$simulator would turn a verilog module
> into a component the $simulator can pick up and run.
>
> there's no doubt that Icarus is the right tool, also gnucap-adms should
> be ported to Icarus, eventually. there should be just one that does all
> of ams. certainly, even tgt-xyce would be pretty neat, if somebody
> cares to do it.
>
> On Sat, Jan 20, 2018 at 04:00:42PM -0500, Beat Arnet wrote:
>> And is there a way to save all simulation states in Icarus in order to
>> revert a run back from the future?
> It is possible, but (was) pretty hard to do. i gave up on it, when it
> did not seem feasible without forking, either way. Or not worth it.
>
> best
> felix
>
> ------------------------------------------------------------------------------
> Check out the vibrant tech community on one of the world's most
> engaging tech sites, Slashdot.org! http://sdm.link/slashdot
> _______________________________________________
> Iverilog-devel mailing list
> Ive...@li...
> https://lists.sourceforge.net/lists/listinfo/iverilog-devel
>