( ESNUG 459 Item 3 ) -------------------------------------------- [12/14/06]
Subject: ( ESNUG 457 #1 ) Columbus extracts SPEF in 2 days; DEF in 1 hour
> e. Built-in Extraction. Sequence's Columbus is a stand alone product,
> which recently has also become an option for CoolTime. It extracts
> tremendously optimized or highly accurate SPEFs directly from GDS for
> all macro and leaf cell power rails. Signals can also be extracted
> for signal based EM analysis.
From: Kishore Gottimukkala <kgottimukkala=user domain=giquila spot mom>
Hi, John,
We're a Magma house that uses Sequence CoolTime and Columbus for the final
signoff of our chips. We've used both for about 15 months now. They are
now part of our standard flow.
We chose them based on Columbus' accuracy plus we felt it was one of the
de facto RC extraction industry standards. The Columbus SPEF extractor can
handle both LEF/DEF and GDSII input. The largest module we've run through
CoolTime was around 800 K gates. Its run time for the DEF extraction for
800 K logic gates + register files was 40 minutes. For our top level runs
we use PNM models generated by CoolTime.
CoolTime's biggest strengths to be its stability and it was very easy to
integrate in a Blast Fusion environment. We wanted to run CoolTime with
the same settings as we had in our place and route tool. Initially we had
tried Columbus SPEF extraction using GDSII, but it took about 2 days for
the SPEF! We later found that it only took 1 hour to generate it using
DEF. So we had to set all the options related to DEF.
It took us 2 weeks to set up our environment with CoolTime. We started
with SDC creation which is compatible with CoolTime. First, we export the
SDC from Blast Fusion and feed it to CoolTime. We then set global variables
for crosstalk enabled timing & glitch analysis. CoolTime then automatically
creates the run file for timing analysis and glitch analysis based upon our
place and route settings.
We used CoolTime to validate our 4 M gate design + 2 MB of memory in TSMC
90G chip and were very satisfied. We used a 100 psec setup and hold margin
on our 400 Mhz clock and did not uncover any problems. It's a good tool.
- Kishore Gottimukkala
GiQuilla Corp. Santa Clara, CA
Index
Next->Item
|
|
