Dennis wrote:
> SO HOW is the blazerouter?
> just curious

I have the Blaze router as well as Spectra. Unfortunately, I never had the
time over the past year or two that PADs released the new  router software
to tinker with the router.  Almost all of the PADs designers I know in our
area have bailed from using PADS and joined the Cadence bandwagon.   I know
Altium is preparing to release their new router and I read their autorouter
whitepaper with interest.  Here are my  observations about routers in
general.   I worked with PHD math major who walked around with a coffee mug
most of the time sharing his paradoxes of math algorithms and equating them
to real world applications.   He once told me the boring mathematical theory
how to stack oranges in the most efficient manner.  Yes there was a formula
to it. There was also the stock boy with a 6th grade education working in
the produce section of a grocery store who already knew without  a proof
that the most efficient method was to stack them side by side and  build the
orange pyramid in the isle.
Years later, I read Dr Cooper and Dr. Chang applied the same theory into
developing the  Spectra autorouter.   Can the basic algorithm be improved
on? The answer is no.  Can Spectra be improved, the answer is yes. What is
lacking with current router technology is the ability to recognize patterns.
Specta's approach is route point to point regardless if the pattern repeats
itself 100 times.     I am guessing Altiums router will have the
intelligence to recognize real patterns.
For example  Take a backplane with three  3 row connectors pins 1 -96 of J1
connected to same pins of J2 and J3.   No router in the world can route
three connectors, straight,  on the least amount of layers, using no vias.
Sure Spectra can , if you program every connection like it was line of code,
but to hit GO and route....forget it.   It does no better than Protel's
autorouter.    This is the simplest of routing task yet it cant be rotund or
stacked like the oranges, efficiently .     The most efficient method is
obvious to you the designer because you can visually see there are
connector  as objects , each identical, representing a pattern.    I have
waited  a long time in anticipation for someone to write code that
recognized real patterns.

While I am on this subject of pattern recognition.....Altium if you are
reading this.....give us something that will recognize signal (neg) and
signal (pos) as a patterned pair. Or ability to automatically
identify/correlate and pair  positive and negative signals from the netlist.
This may  require setting   a standard naming conventions  for  signal names
on schematics.    Differential paired routing has taken  over in high speed
design and is a must for any new router.

I have my checkbook open and am waiting for the new router.    If 99SE is
any indication of how good these guys in AU  can write software then the new
router should shake the industry .

Mike Reagan

* * * * * * * * * * * * * * * * * * * * * * * * * * * * * *
* To post a message: mailto:[EMAIL PROTECTED]
* To leave this list visit:
* Contact the list manager:
* Forum Guidelines Rules:
* Browse or Search previous postings:
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * *

Reply via email to