( ESNUG 588 Item 4 ) ---------------------------------------------- [09/19/19]
Editor's Note: Every time I publish, I get pelted with these short
reader comments about that day's news. Since they're not often full
emails, but more quick reactions, I call them "snarkies". - John
Subject: 56% of users doubt Anirudh's secret "new" Clarity matrix solver
A SERIOUS THREAT: When you do 10X - 20x types of speed/memory/price
breakthroughs (like what Anirudh's Clarity is showing here) -- that's
when EM customers comfortable with their olde reliable Ansys HFSS will
sit up and take notice; and even consider changing over to Clarity.
(click on pic to enlarge image)
|
In technical terms CDNS Clarity has/is:
- new compact data architecture -- uses 12x to 14x less memory
- cloud friendy -- manages latency & scalabiliy, fault tolerant
- new algorithm -- secret sauce that doesn't approximate physics
and is blazing fast
- deep hooks into Virtuoso/ADE and Allegro and Sigrity
- auto partitioned -- no segmenting needed!
- massive multi-threading, massively distributed
- uses many small cheaper and disparate machines, elastic compute
- Anirudh goes total war on Ansys mothership at CDNlive'19
DON'T WANT TO GET BURNED: On the business side, the users naturally welcome
a new competitor for the obvious reasons that competition brings down tool
pricing while at the same time increases tool quality.
Users' pre-DAC "snarkies" on Anirudh's Clarity to take on ANSS HFSS news:
doubt secret
"new" matrix
algorithm: :###################################################### 56%
distrubuted
is NOT a new
algorithm: :##################### 21%
competition
is good: :################ 16%
HFSS will
catch up: :############ 12%
this story
woke up Ajei: :######### 9%
less memory
lowers costs!: :######## 8%
going cloud
is good: :####### 7%
benchmark
detail needed: :###### 6%
doubt auto-
partitioning: :##### 5%
doubt
scaling: :#### 4%
meshing
questions: :## 2%
other
comments: :######### 9%
But the user comments reacted VERY NEGATIVELY to the claim that some unknown
secret "breakthrough" matrix solver algorithm was significantly responsible
for the far better runtime results. It's not a personal dis on Anirudh as a
man nor as a techie; it's just that engineers are heavily schooled to never
accept anything on faith ever. They don't want to get burned because they
didn't take the time to fully explore how something did (or didn't) work.
"Trust, but verify."
- Ronald Regan, U.S. President (1911 - 2004), an old
Russian proverb he often used in nuclear disarmament
discussions with the Soviet Union.
---- ---- ---- ---- ---- ---- ----
USER DISTRUST OF SECRET "NEW" MATRIX SOLVER ALGORITHM
There's something off about a magic new algorithm that gets
blazing fast results and uses no approximations. Sounds an
awful lot like magic beans to me.
---- ---- ---- ---- ---- ---- ----
Until I know the deep details on those benchmarks, I don't
trust them worth the PowerPoint slides they're written on.
---- ---- ---- ---- ---- ---- ----
I distrust all allegedly new algorithms until I see them
fully vetted first.
---- ---- ---- ---- ---- ---- ----
A matrix solver I don't understand is a matrix solver I don't
use.
---- ---- ---- ---- ---- ---- ----
My Ansys AE thinks that Cadence Marketing might be confusing
algorithmic speed-ups due to massive parallelism through
distributed computing with a brand new matrix solution.
He says totally new matrix solvers are extremely rare.
---- ---- ---- ---- ---- ---- ----
Multi-threaded isn't a new algorithm.
---- ---- ---- ---- ---- ---- ----
Distributed, multi-threaded is good, but it'd not a new
secret sauce algorithm.
---- ---- ---- ---- ---- ---- ----
I suspect once HFSS gets distributed up and running, that
lead that Clarity has might evaporate.
---- ---- ---- ---- ---- ---- ----
It's quite easy to make boisterous yet ultimately claims
about a new algorithm that no one is allowed to see.
---- ---- ---- ---- ---- ---- ----
The only secret sauce I trust is the secret sauce that's on
a Big Mac, which is mayo mixed with pickle relish, only because
I know what's in that secret sauce.
---- ---- ---- ---- ---- ---- ----
Until I hear more details about this alleged new algorithm, I
see Clarity is a rebranding of their failed 3D EM tool.
---- ---- ---- ---- ---- ---- ----
What's the skinny on this "breakthrough" new matrix solver???
---- ---- ---- ---- ---- ---- ----
I looked up the U.S. Patent Office applications. So far, there
is nothing filed by Cadence nor Anirudh Devgan regarding any
new matrix solving algorthms!
This means Anirudh's new matrix solution is only known within
the walls of Cadence R&D -- and that no outside academics have
done a formal proof that his algorithm actually works, much
less how accurate it is.
Caveat emptor!
---- ---- ---- ---- ---- ---- ----
If this is all about Cadence R&D finally getting 3D-EM to be
distributed and in the cloud, Aniridh's claim of a new algorithm
is going to be very embarrassing for him.
---- ---- ---- ---- ---- ---- ----
HFSS FEM is an approximation. A good approximation, but still
an approximation.
Clarity is supposedly faster and not an approximation ???!
I have a bridge in New York to sell you if you want.
---- ---- ---- ---- ---- ---- ----
something is wrong if it's secret
---- ---- ---- ---- ---- ---- ----
My contacts tell me there is no new algorithm. Cadence marketing is
just calling the gains from Clarity being massively distributed and
in the cloud is a "breakthrough algorithm".
---- ---- ---- ---- ---- ---- ----
Something is amiss whenever an EDA vendor hides the fundamental
way how their tool works.
---- ---- ---- ---- ---- ---- ----
I don't trust secret new algorithms.
---- ---- ---- ---- ---- ---- ----
HFSS added multi-threading in parts of it a few years ago.
Even Anirudh's own benchmarks show HFSS using 32 CPU machines.
---- ---- ---- ---- ---- ---- ----
I have a lot of respect for Anirudh's technical prowess. Once he
gives a rough outline of how his breakthrough algorithm works, then
and only then will I trust my circuits to it.
---- ---- ---- ---- ---- ---- ----
need more details on this matrix solver
---- ---- ---- ---- ---- ---- ----
Cloud and distributed may be new, but not a new algorithm.
---- ---- ---- ---- ---- ---- ----
massive distributed =/= breakthrough algorirthm
---- ---- ---- ---- ---- ---- ----
in technical matters, i never rely on anything secret
---- ---- ---- ---- ---- ---- ----
---- ---- ---- ---- ---- ---- ----
---- ---- ---- ---- ---- ---- ----
CDNS ANIRUDH CLARITY VS. ANSS AJEI HFSS
If Anirudh has the balls to take on HFSS in a full frontal
assault, I say good luck with that!
---- ---- ---- ---- ---- ---- ----
I think all your scoop did, John, was to wake up Ajei's HFSS
R&D team to clean up their mess 6 to 8 months earlier than if
Ansys mgmt was left to their own resources.
---- ---- ---- ---- ---- ---- ----
For Cadence to go against a heavily entrenched foe like Ansys
is either utter folly or something very very clever.
Until I use Clarity myself, I can't tell which yet.
---- ---- ---- ---- ---- ---- ----
God knows Ansys R&D needs a good swift kick to get off their
butts and start seriously revamping HFSS. Too much deadwood.
---- ---- ---- ---- ---- ---- ----
This will be a coup for everyone if Anirudh takes even 5% of
the HFSS market in the first year. Customers get a new choice
and Ajei will have to start re-investing in HFSS R&D.
---- ---- ---- ---- ---- ---- ----
Yay, Anirudh! Competition is good!
---- ---- ---- ---- ---- ---- ----
If Clarity is faster than HFSS, and is better integrated into
Virtuoso, that could make it a serious contender.
---- ---- ---- ---- ---- ---- ----
Lots of CDNS mktng claims there. An awful lot.
---- ---- ---- ---- ---- ---- ----
I predict HFSS will be in AWS and Azure within 9 months.
---- ---- ---- ---- ---- ---- ----
That's a smart move by Anirudh. A logical adjacency to grow into.
Now the question is can he pull it off?
---- ---- ---- ---- ---- ---- ----
Odd niche for Cadence to go into, but if they can make money
there, great.
---- ---- ---- ---- ---- ---- ----
Your 24 to 36 month estimate is wrong. I think this article will
cut that to 12 to 18 months before Ajei catches up woth Anirudh.
---- ---- ---- ---- ---- ---- ----
You're doing drama there, John.
If CDNS takes $5 million, or even $10 million from ANSS, that's
hardly noticable. ANSS makes $1 billion in revenue each year.
That $5 million is 0.5% of ANSS revenue.
Barely TOTAL WAR by any measure.
---- ---- ---- ---- ---- ---- ----
Instead of going after first place HFSS, wouldn't it be more prudent
for Anirudh to go after second place (Dassault) CST instead?
---- ---- ---- ---- ---- ---- ----
---- ---- ---- ---- ---- ---- ----
---- ---- ---- ---- ---- ---- ----
BENCHMARKING THE IMPACT OF MEMORY USE
Whoa! Less memory cuts overall costs by ~20X!!!
---- ---- ---- ---- ---- ---- ----
Who knew cutting memory use had such a big effect on the
bottom line?
---- ---- ---- ---- ---- ---- ----
Good analysis on how less memory usage impacts overall cost.
Well done.
---- ---- ---- ---- ---- ---- ----
Can you get deeper details on those 3 user memory benchmarks?
We'd love a Cooley report on any of them.
---- ---- ---- ---- ---- ---- ----
I never knew that memory use made any real difference before.
---- ---- ---- ---- ---- ---- ----
---- ---- ---- ---- ---- ---- ----
---- ---- ---- ---- ---- ---- ----
AUTO-PARTITIONING / AUTO-SEGMENTING / SCALING / MESHING
Auto-partitioning here is like auto-deductions from my paycheck.
---- ---- ---- ---- ---- ---- ----
No hand segmenting needed? Riiiight...
---- ---- ---- ---- ---- ---- ----
Never trust a partition that you didn't do yourself.
---- ---- ---- ---- ---- ---- ----
I'm curious on how Clarity does meshing.
---- ---- ---- ---- ---- ---- ----
Linear scalability up to what? All I see is 32 CPUs on those
benchmarks.
---- ---- ---- ---- ---- ---- ----
What cloud does Clarity run on today? AWS? Azure? IBM?
---- ---- ---- ---- ---- ---- ----
What the difference between Clarity, HFSS, CST, AWR, ADS?
---- ---- ---- ---- ---- ---- ----
tight hooks to Virtuoso is a big plus
---- ---- ---- ---- ---- ---- ----
---- ---- ---- ---- ---- ---- ----
---- ---- ---- ---- ---- ---- ----
OTHER READER COMMENTS
Good scoop, John. That's why I read you.
---- ---- ---- ---- ---- ---- ----
Aart is still reeling from his ICC2 disaster even today.
---- ---- ---- ---- ---- ---- ----
ANSS has an operating profit margin of 45%??!!
I gotta buy me some ANSS!
---- ---- ---- ---- ---- ---- ----
Related Articles:
SCOOP! -- Anirudh goes total war on Ansys mothership at CDNlive'19!
Ansys CEO Ajei Gopal cleverly acquires Helic as #8 "Best of 2018"
Hogan and Anirudh on Ansys ANSS Redhawk/SeaScape vs. CDNS Voltus
Anirudh and Sawicki on iffy Apache IR-drop #'s vs. Voltus/Innovus
Cooley's open letter to ANSS Ajei Gopal to join Troublemaker Panel
Join
Index
Next->Item
|
|
