al davis wrote:
> On Monday 08 October 2007, Stephen Williams wrote:
>>  I think we are particularly
>> keen on this as we in gEDA feel there is a dumbing down of
>> the CS/EE curriculum by the big tool vendors.
> 
> It's tempting to think of the "big tool vendors" in our case as 
> Synopsys, Cadence, and Mentor.  They are not the problem.
> 
> Well ... they might be a little because some professors insist 
> on teaching it even though it is way to hard to use for 
> beginners.

you mean a 1,000,000 page manual spanning 500 volumes that reads like 
"The enable super secret parameter stepping button enables stepping of 
super secret parameters" is overwhelming for beginners?

I don't have a problem using those tools in schools if you're actually 
designing a real chip.  But they're the wrong tools for an introduction 
to simulation.  Too easy to spend all your time on the mechanics of 
"pick this menu choice, now click here, now fill in this form" sort of 
like a class on ms-excel instead of learning what is actually more 
useful in practice.  The fill in the form gui stuff really can be left 
as an exercise for the engineer.

> The biggest dumbing down is by accepting the MS-windoze way of 
> doing things, which you can do on any system.  Typing a command 
> is completely foreign, even if the command is only two letters.  

I have to agree 100% with Al here.  I consider the gui only way of 
driving simulators to be harmful to students.  I couldn't even begin to 
count how many times I've had some question or other about what was 
going on with a big simulation and the answer came fairly easily with a 
long ugly line of awk code or a big pipeline of awk, sed, and other 
standard unix commands.  Even if you use a high dollar simulator which 
comes with a GUI environment, you can never ignore whats going on and if 
you do, you will limit yourself.  Also despite what some big CAD 
companies might like you to believe, my experience has been that a GUI 
simulator environment is ok for small stuff but just not good enough for 
complex things.

> Another part of problem in our case is the demo-ware that comes 
> with the texts.  The "free" (meaning non-free) versions of 
> Spice make me boil over.
> 
>>From an educational viewpoint, they offer nothing that the true 
> free version of Spice doesn't offer.  Worse, they take a lot 
> away, and they are very hard to learn, very time consuming to 
> teach.  It is nearly impossible to go beyond the magic black 
> box level.

also agreed.  Working with netlists is not hard and it is very 
educational and has real payoffs in the real world outside of education.

> Gnucap offers some real advantages from an educational 
> viewpoint.  It is interactive, lets you play with the circuit, 
> and shows stuff that Spice doesn't.  I can teach enough of it 
> to get started in a lot less class time than is required for 
> Spice or a GUI-Spice.

Yep.  And especially since all simulators are different in terms of 
their exact feature set and input syntax its not like you're missing out 
on something important by picking one over another.  And, despite all 
the differences, most are basically similar. If you know how to work 
with a gnucap netlist, then a berkeley spice-3f5 or hspice or spectre 
netlist is a simple matter of syntax.


-Dan


_______________________________________________
geda-dev mailing list
[email protected]
http://www.seul.org/cgi-bin/mailman/listinfo/geda-dev

Reply via email to