( ESNUG 448 Item 5 ) -------------------------------------------- [11/11/05]
Subject: Cooley Messed Up the Star-RCXT Question in the Synopsys Survey
> 12. Aart claimed in his keynote that Star-RCXT is now "5X faster"; in
> your experience is this real or B.S.? How does Star-RCXT, Cadence
> Simplex Fire&Ice, Mentor xCalibre, and Sequence rank in your eyes?
From: Michelle Meier <mmm=user domain=synopsys spot gone>
Hi, John,
Our R&D folks and even some customers were very surprised when they read
in your SNUG 2005 survey that Star-RCXT was supposed to be 5X faster in
2005. 20X in 2 years would be pretty amazing. (In SNUG 04 #13, Star-RCXT
was 4X faster in 2004.)
I checked and re-checked Aart's SNUG 2005 keynote and finally found that
"5X faster RC extractor" on page 91 you read. The reference you found is
for the built-in extractor in Astro 2004.6 which is not Star-RCXT. This
extractor in Astro is optimized for implementation and trades faster speed
against accuracy. It's correlated to within 5% of Star-RCXT. The
difference is that Star-RCXT is sign-off, which requires deadly accuracy.
John you made a mistake here in this survey question.
- Michelle Meier
Synopsys, Inc. Sunnyvale, CA
P.S. For the record, Star-RCXT in 2005 has 2X higher capacity and its
accuracy is now 0.2 fF for 65 nm. It's not any faster but it
correlates within 5% of Raphael NXT and Quickcap NX.
Editor's Note: Oops. My bad. I'll have to own this mistake when I
compile the stats for the "Synopsys vs. its Rivals" survey. - John
Index
Next->Item
|
|
