Re: gEDA-dev: Last call for input on SOW for Linux Fund....

[John Doty <jpd-at-NOSPAMnoqsi_dot_com>]

On Dec 3, 2008, at 5:26 PM, Dan McMahill wrote:

> John Doty wrote:
>> On Dec 3, 2008, at 7:40 AM, Svenn Are Bjerkem wrote:
>>
>>> Ken Kundert knows what he is talking about
>>
>> I don't think so. He thinks mixed-signal design is like computer
>> programming. But it's not: to do it well you have to start from the
>> *physics* of what you're trying to accomplish.
>
> I guess thats why a huge number of complex commercial chips are  
> verified
> using the simulator that Ken designed?

Of what sort? Can you name an actual chip? There Verilog-AMS culture  
seems to be about making extravagant claims which can never be  
evaluated. The physical reality is always hidden in the fog. But  
simulation isn't the issue here anyway: it's the claim that the  
utility of Verilog-AMS goes beyond simulation.

>   He certainly has his view point
> about things that you may disagree with, but to say he doesn't know  
> what
> he is talking about is pretty far off base.

I see no evidence that he understands the process of *designing*  
mixed-signal circuits. The critical parts of this process are applied  
physics, not computer science, but his book shows no understanding of  
this elementary fact.

Simulation supports design, but it is not the same thing. And in  
mixed-signal design, the requirements (noise, dynamic range, power)  
generally take you straight to the critical low-level circuits, not  
to high level interfaces. The high level circuitry is the scaffolding  
that supports the critical stuff. It's not like computer science. In  
software design, you have much more implementation freedom, so you  
can let interfaces drive implementation.

For example, I often work with signals where the noise variance is  
proportional to signal amplitude. Now, when this happens, you can  
help your dynamic range (something I always seem to be fighting for)  
by processing the square root of the amplitude. And in the digital  
domain, this is easy: the algorithm to compute stable, accurate  
square roots fast has been known for two millennia. Therefore, you  
can make a top level decision that you'll encode amplitudes by their  
square roots, and design top down, without trouble showing up when  
you hit the level where you have to take square roots.

This doesn't work with analog. Analog square rooters have stability,  
accuracy, and dynamic range problems themselves, so this approach fails.

You can write computer programs from the top down in ignorance of the  
physical implementation of the computer. You cannot do this  
effectively with mixed signal design, because the physics of the low- 
level parts is profoundly important.

>
>> I wonder what tools Paul Horowitz uses these days. *There's* a guy
>> who understands mixed signal disign and can write clearly about it.
>
> I wonder if Paul Horowitz does complex chip designs.  The reality  
> of it
> is that simulation tools have never been that useful or in my opinion
> critical for the sorts of circuits in his book.

But what Horowitz understands is that the physical fundamentals are  
keys in the mixed-signal domain. Kundert does not seem to understand  
this.

>   But for the IC industry
> which is really where Verilog-{HDL,A,AMS} are used, simulation  
> tools are
> critical and those 3 in particular fill a critical need.

But what got me started on this thread is Al's claim that the  
capabilities of Verilog-AMS go beyond simulation to supporting a  
general-purpose interchange format. This seems a pretty wild claim:  
where's at least a toy example? Kundert makes an equally unsupported  
claim: that software-oriented top-down design is a sensible approach  
for mixed-signal circuits.

>>
>> Kundert's book is what Al got me to buy. It's a sales pitch, not a
>> textbook: the signal to noise ratio is very low. Someone who has a
>> problem to solve doesn't want a bunch of hype on the advantages of a
>> tool: they want a roadmap that will allow them to find a solution.
>
> 12 pages of background and context

Read like sales pitch to me.

> 20 pages of sales pitch
>> 200 pages of reference material spelling out the language

A *programming* language. Where are the circuits? It's all sales  
pitch: it's not a designer's guide at all.

>
> doesn't sound like bad SNR to me.
>
> When I can't quite remember the name of a function or the order of the
> arguments that is the book I reach for.

OK, so you use it as a reference manual for a programming language  
for simulation. That's not what Kundert is selling: he wants you to  
be designing your circuits with it. And it's not what Al is selling:  
he wants you to represent your schematics and layouts with it. All  
this selling...

>   In contrast I have not reached
> for the Horowitz book since sophomore or junior in college 2  
> decades ago.

What do you read when you need to refresh your knowledge of the  
physical foundations?

>
> But... unless you have coughed up the money for the simulator its  
> not so
> useful.  So my question for you is do you have access to Verilog-A or
> Verilog-AMS so that you have been able to try them out on a real  
> design?
>     If you have used the tools, how did you learn about them?  If you
> haven't used the tools, I question how useful of a review you can give
> of a book that teaches the tools.

I learned about them from Al. I'm interested in Al's objectives. Al's  
a smart guy: I thought he might be on to something. But I don't see  
where Verilog-AMS has any substance beyond simulation: it's not a  
tool for creating or representing designs for other purposes.

>
>
> If you don't use tools like these, do your designs have large mixed
> signal components that are not easily broken apart?

No, in fact my designs are highly modular. But they don't start from  
top down: that would be insane.

>   By that I mean, do
> you have blocks that contain some subcells which are prohibitive for
> simulating at the transistor level but yet you have to have a model of
> them in place during a simulation because they are part of a feedback
> loop?

For my IC work, not prohibitive, but sometimes difficult.

>   What tools do you use to handle that complexity?

Oh, I use various techniques. Analog computer simulations in SPICE  
are easy and not subject to physical limitations. Mathematica is a  
powerful tool because it allows mixtures of symbiolic and numerical  
analysis, so, for example, I can find the time constants for critical  
damping easily.

>   How do you
> ensure that your million dollar mask set

Million dollar mask set? That's got to be nearly all digital, no?  
Maybe a dusting of mixed signal around the edges. But you're changing  
the subject: the issue isn't whether Verilog-AMS is good for digital  
simulation, it's the intemperate claims that it's good for design and  
layout.

> is going to produce silicon
> that is at least functional and doesn't have outright wiring errors or
> fundamental screw-ups in the block diagram?

SPICE. And the only time I've had that (partially) fail was when the  
mask set the layout contractor sent to MOSIS was different from what  
they used to extract the netlist for my final sims. A major screwup,  
but "garbage in, garbage out". A better simulator wouldn't have helped.

>   How do you include the
> complex behavior of one block where 20 dB analog accuracy is  
> sufficient
> in a simulation where 100 dB of analog accuracy is need for another
> block?  It's one thing to realize you need to buy a different op- 
> amp for
> your board or add 15 blue wires.  It is another thing entirely when  
> you
> find you just took a many month schedule hit and had to come up with
> another huge pile of cash for masks because you only did block level
> simulations on your chip.

It's a balancing act. I've cut back some on sims for my more recent  
designs because the time to set them up and interpret the results was  
costing more than a 40 chip run.

>
> Those are the sorts of real world problems which are solved every day
> using Verilog-{HDL,A,AMS}.  It is not hype or vaporware.

You claim that, but like all Verilog-AMS claims I've seen I cannot  
verify what you say. And anyway, it's irrelevant. The hype that  
really bugs me is the claims that its capabilities go far beyond  
simulation.

John Doty              Noqsi Aerospace, Ltd.
http://www.noqsi.com/
jpd@noqsi.com




_______________________________________________
geda-dev mailing list
geda-dev@moria.seul.org
http://www.seul.org/cgi-bin/mailman/listinfo/geda-dev