At 08:32 AM 3/16/01 -0500, Mike Reagan wrote:
>Thank you Mr. Lomax,
>For you input so eloquently expressed.   I would have eliminated the parts
>of the netlist, however it would have been counter productive to goals.

Ultimately, of course, one would have to load the entire net list. But that 
process might be reserved for overnight. Of course, if it is going to 
crash, none of this would be of any use.

[...]
>For placement, I
>generally eliminate VCC and Gnd nets from the netlist because they get in
>the way visually.

You can, of course, turn off VCC and GND net display. But those nets will 
still cause some operations to take much more time.

>   During the course of my design, I will import the netlist
>two more times. Once before it goes to the router and one last time before I
>perform final DRCs and gerber files.    My macros end up very clean, no
>hits, no runs, no errors.

I follow the same process. Always reload the net list before sending the 
job out. There should be no macros, except that if there are multiple pads 
(more than one pad for a given pad name), there will, at current rev, be 
some remove net macros; they are easy to recognise. Hopefully, Protel will 
fix this bug.

Why do this? Protel allows you to edit net assignments. It is not 
impossible that one could inadvertently edit one. Reloading the net list is 
the only way to discover this kind of error. There is also the multiple pad 
bug, if one has such creatures, and there can be other reasons to reload as 
well.

>[...]
>I think there is some limitation to Protel's design size which we are
>unaware of.   On medium to large design sizes to about 8000-9000 nets,
>Protel slows down but doesn't stall out as I mentioned.

I don't call those medium to large designs, I call them very large designs. 
The largest design I've ever encountered in my entire career -- so far -- 
was 8000 pins. Pins, not nets. I didn't do it -- impossible with Tango -- 
but I vaguely recall that some company did the job for about $20,000. An 
8000 net board is at least twice as large as that.

That was pre-surface mount; the board was physically quite large. It 
becomes feasible to have more components nowadays, but most companies 
partition designs instead of trying to do one big board. It's more 
flexible. And one does not choke the CAD systems.....

>    The design I am
>working on right now has 500 less components than a similar design I
>struggled with last month, however the net count is about 6000 more.

So, say, 14,000 nets, thus *at least* 28,000 pins. Plus bypass caps. In the 
old language, this is 1750 IC equivalents. Yes, this is a large board among 
large boards.

>   I am
>afraid to think what is going to happen on my next import.  I hate to use 98
>because I am using features only available on 99....more planes, design
>rules . etc.    Maybe I will throw it on a computer overnight and see if
>ever loads.  As I mentioned, I was posting this to draw fire and attention
>to the issue, which has not been resolved thru a two releases and 6 service
>packs.

It would be useful to know the practical limits for Protel. None are stated 
anywhere that I have seen. But I am sure there are actual absolute size 
limits, because of the number of bits available in certain pointers, etc., 
and, though that number might be quite large, there is also the practical 
limit that if it takes two months to load the net list, the engineers might 
get a bit antsy.

The practical limit is not an easy number to determine, because it depends 
not only on the program but also on the processor speed, memory, and other 
variables as well. But what is our experience. Has anyone else successfully 
done a design as large as the one which is causing Mr. Regan such problems?

I once did some tests with Tango by multiplying up a design until it got 
too unwieldy. It would not be difficult to do this with Protel.

[EMAIL PROTECTED]
Abdulrahman Lomax
P.O. Box 690
El Verano, CA 95433

* * * * * * * * * * * * * * * * * * * * * * * * * * * * * *
* To post a message: mailto:[EMAIL PROTECTED]
*
* To join or leave this list visit:
* http://www.techservinc.com/protelusers/subscrib.html
*                      - or email -
* mailto:[EMAIL PROTECTED]?body=leave%20proteledaforum
*
* Contact the list manager:
* mailto:[EMAIL PROTECTED]
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * *

Reply via email to