( DAC'14 Item 4 ) ----------------------------------------------- [12/05/14]
Subject: Mentor BDA AnalogFastSpice and Solido were #4 tools at DAC'14
WHAT'S NEXT, RAVI?: After much legal drama, U.S. District Court in Oakland
dismissed with prejudice CDNS vs. BDA in January (ESNUG 535 #7). SPICE
users cheered (ESNUG 535 #9). In March, MENT acquires BDA (ESNUG 538 #2).
  
"... all agree that this is a merger where Ravi Subramanian and
his BDA guys will be given free reign to build a killer custom
division within MENT to directly compete against SNPS in SPICE
and Cadence in SPICE, AMS, and custom."
- from http://www.deepchip.com/items/0538-06.html
So naturally every Virtuoso and SPICE user at DAC'14 was trying to corner
Ravi Subramanian to ask him "what's next?". Heck, BDA was the very first
question in the DAC Troublemakers Panel even! (ESNUG 543 #1)
For BDA AFS rivals, engineers were curious about Cadence's new Spectre XPS.
SOLIDO KICKS ASS: Last year there was a big 5-way fight at DAC'13 in Austin
about exactly who had the best SPICE variation tool:
"But ours is better..."
- Amit Gupta, CEO of Solido, about the four other
SPICE variation vendors at the Austin DAC'13.
"But ours is better..."
- Harald Neubauer, CEO of MunEDA, on the four other
SPICE variation vendors at the Austin DAC'13.
"But ours is better..."
- Zhihong Liu, CEO of ProPlus, about the four other
SPICE variation vendors at the Austin DAC'13.
"But ours is better..."
- Firas Mohamed, CEO of Infiniscale, on four other
SPICE variation vendors at the Austin DAC'13.
"Huh? Oh. No, really, OURS is better..."
- John Stabenow, Cadence marketing, on four other
SPICE variation vendors at the Austin DAC'13.
Quoted directly from http://www.deepchip.com/items/0528-02.html
Not one SPICE user said anything about ProPlus, Infiniscale nor Cadence's
SPICE variation as their "Best of DAC" tool this year -- leaving a mess of
users gushing about Solido and one lone user liking MunEDA Wicked.
SURVEY QUESTION #1:
"What were the 3 or 4 most INTERESTING specific EDA tools
you saw at DAC this year? WHY did they interest you?"
---- ---- ---- ---- ---- ---- ----
Ravi reassured us that we didn't have to switch to Eldo.
---- ---- ---- ---- ---- ---- ----
Our most important task at this DAC was to corner the CEO of BDA
to find out which tools were slated to die at Mentor.
---- ---- ---- ---- ---- ---- ----
We use BDA ACE/Cadence ADE-L. Mentor plans to keep supporting ACE.
---- ---- ---- ---- ---- ---- ----
AFS rocks!
---- ---- ---- ---- ---- ---- ----
Ravi's roadmap won't be changing much post-acquisition. Good for us.
---- ---- ---- ---- ---- ---- ----
AFS Mega. 100 M+ elements fast SPICE for memories.
---- ---- ---- ---- ---- ---- ----
We use BDA AFS and Cadence Spectre in ADE. Glad the lawsuit is done.
---- ---- ---- ---- ---- ---- ----
Hope Ravi doesn't lose focus on SPICE with his new duties at MENT.
---- ---- ---- ---- ---- ---- ----
BDA AFS
---- ---- ---- ---- ---- ---- ----
Spectre is accurate but not fast.
Spectre-APS is fast but not accurate.
AFS is both fast and accurate.
---- ---- ---- ---- ---- ---- ----
AFS house. 5X faster than Spectre.
---- ---- ---- ---- ---- ---- ----
Cadence Spectre XPS. 10X faster with 1/3 the memory use of FineSim.
---- ---- ---- ---- ---- ---- ----
We plan to check out Spectre XPS back in our office.
---- ---- ---- ---- ---- ---- ----
Do you have any Spectre XPS benchmarks, John?
---- ---- ---- ---- ---- ---- ----
We do std cell. Synopsys HSPICE/HSIM.
---- ---- ---- ---- ---- ---- ----
SNPS LAVA FineSim
---- ---- ---- ---- ---- ---- ----
Solido Variation Designer.
I started using Solido for our RF devices for our near field wireless
communication chips in 2014. It took about 2 hours of training over
a couple of days. Does Monte Carlo and performance sensitivity
against PVT corners.
I used it to test and optimize other circuit parameters instead PVT.
I basically use Solido to sweep bias voltages and transistor sizes
all at the same time to find optimal performance - that way I don't
need to run all combinations,
There can be up to 1000 combinations of Size and Bias choices to try,
but with Solido, I only need to run 100-200 to do the optimization.
We keep our SPICE simulations simple, so it's only 30 seconds a run.
I met my performance spec with low power consumption from battery.
Plus the analysis took much less time than the method I was using
before, where I would have to run all the combinations using brute
force or optimize each parameter individually.
I was able to shorten a 1 week effort to only 2 days.
---- ---- ---- ---- ---- ---- ----
We use Solido Fast PVT. It runs ~3X faster than regular PVT and
presents the results in very user friendly GUI. You sort the
results, investigate particular runs, create graphs.
From using Fast PVT we are now not able to guess the WC, but
really get it from simulations.
We also use Solido to margin sensitive to variation elements like
Level-Shifters, SA, Bandgap, etc.
---- ---- ---- ---- ---- ---- ----
I had a look at Solido's Variation Designer at DAC.
The tool looked very promising. Our goal is it to reduce the number
of necessary Monte-Carlo runs without losing verification coverage.
If the tool works like presented, it could save design time, plus
some computing resources and license costs.
---- ---- ---- ---- ---- ---- ----
We use Solido Variation Designer for design sensitivity. It helps us
figure out which devices in our design impact circuit performance
the most. Solido has distributed pre- and post-processing. Overall
performance was over 25x faster.
---- ---- ---- ---- ---- ---- ----
We use Solido for high sigma analysis.
It's good for information where Monte Carlo analysis just cannot get
answers due to excessive runtime.
---- ---- ---- ---- ---- ---- ----
We are trying to get a handle on variation in general. Right now we
just run PVT corners, if that.
The Solido demo was well done and we plan to look into it further.
The idea of tool doing quick efficient variation analysis is appealing.
---- ---- ---- ---- ---- ---- ----
The DAC Solido discussion went well.
There are difficult pain points when running Monte Carlo simulations on
analog data. Solido's has a GUI integration into CDNS, and reduces the
SPICE simulations you run without sacrificing worst case scenarios.
I would also like to see their demo for a special case specification.
---- ---- ---- ---- ---- ---- ----
Solido Variation Designer
From the demo I had a good impression of their GUI.
---- ---- ---- ---- ---- ---- ----
MunEDA Wicked
We do WCD Monte Carlo. Our tech guy doesn't like HSMC. Too big
of a relative sample error. Says WCD is better.
---- ---- ---- ---- ---- ---- ----
Related Articles:
SCOOP -- Judge dismisses CDNS vs. BDA lawsuit "with prejudice"
So far public opinion is strongly anti-Cadence on CDNS vs. BDA
BDA tells CDNS "IN YOUR FACE" with new analog characterization
Join
Index
Next->Item
|
|
