( ESNUG 412 Item 5 ) -------------------------------------------- [05/22/03]
Subject: ( SNUG 03 #3 ) Well, At Least One Customer Liked Synopsys LEDA
> LEDA just jumped up with 0 pos and 6 neg votes. Something's up there.
> It doesn't have many users, so 6 negs makes LEDA really stand out here.
>
> - from http://www.deepchip.com/items/snug03-03.html
From: [ Richard Gere's Lost Gerbil ]
John,
You have to understand why LEDA has a bad rap though. No one want's the
code police coming down on them, and certainly since that was added to my
duties I've lost a lot of friends. :) Even now there are still pockets of
resistance against LEDA, but all new ASIC code here is being run though it
and we really have found late end bugs because of LEDA. Anyways if you
can't keep me anonymous, please don't post this.
We use LEDA and have been successful with it. We first started using LEDA
about a year ago and at the time there was severe resistence as it was much
slower than Verilint and produced incredible volumes of data that were
simply impossible to sift through. LEDA also crashed a lot and to run the
GUI seemed to require our biggest machines.
Anyways in the last year our benchmarks are at least 5x faster and the rules
themselves are behaving more consistently. The big problem is it looks so
simple to run... Hours after you have LEDA installed you can start sifting
through error logs and finding issues, that you think it's simple. However
LEDA did not start to become a value-add to us until we sat down and
carefully defined several corporate-wide policies. Once we did this, logs
that previously had 10,000 violations were reduced to a few hundred serious
issues and debugging became much simpler. Since then we caught 2 late-stage
bugs (on designs where we brought LEDA in late) that would of resulted in
reduced functionity if they hadn't been corrected and generally produced
more readable and reusable Verilog.
Consider me a vote for LEDA!!!
- [ Richard Gere's Lost Gerbil ]
|
|
