( ESNUG 589 Item 02 ) --------------------------------------------- [04/07/21]
Subject: Sawicki on Siemens Avatar PnR vs. Innovus, ICC2, and Fusion Compiler
Virtual DAC'20 Troublemakers Panel
Cooley: Joe, Sawicki. Why did Mentor buy Avatar? Aren't you walking
into a huge battle? Didn't you learn from Sierra Olympus?
Where is the upside for Mentor in this? I don't get it. This is
a bloodbath between two elephants and you're a mouse walking in.
Sawicki: Well, we definitely learned some things having bought Sierra a
while back. One is you can't just have one feature. Sierra had
a really good optimization engine that we did not successfully
follow through with, with a full place and route solution.
I'd emphasize in this market one of the things that does make it
attractive... there's the fact that there are two big gorillas
-- but there also is the fact that customers really like having
multiple PnR tools in this space.
We've seen over and over again where even on the same chip you'll
have people using different PnR tools on different blocks because
one PnR tool happens to give them better results. They want to
always be matching one PnR tool versus the other.
Historically, it was common for people to have three PnR tools.
So, it's in that context that we looked at Avatar.
And why we bought it was pretty simple. We brought it in, did
some benchmarking -- had some 3rd party help on this. And for
a company that had been one of the almost... I think that ATOP
was first or second in the parade of companies Synopsys has
taken to their knees with their legal strategy.
And that one is fascinating -- you've got to remember they [SNPS]
took them [Atoptech/Avatar] out on a copyright violation.
Cooley: Yeah it wasn't a technology fight, it was just dirty lawyer crap.
92% of EDA users say Justice NOT served in SNPS-Atoptech case
Sawicki: Kind of amazing. But even after going through all the trials and
tribulations of trying to stay alive in this market, while
everyone was nervous about their [Atoptech/Avatar] survival...
Their technology benchmark was right up there with best-in-class.
And that's with almost nobody in the company.
3 users benchmark Avatar Aprisa vs. Innovus/ICC2 at 7nm
We closed on Monday (August 24, 2020). We'll finish up some
releases on our existing Nitro tool suite. And we will double
the size of that organization as soon as that last release is
done, and we put those people on board. And we intend on adding
significantly more people next year.
So, the bet is simple. This market likes having a lot of tools.
Avatar has an architecture that has been able to put together
results that are outstanding with a minimum engineering
environment.
Between our [Siemen EDA's] stability as customers look at this,
and our ability to scale resources to increase scope and scale,
this is something that we're going to be able to do very well on
financially. So, yeah, we've learned some things.
Cooley: Aart has huge market share from sheer inertia. But there's a
lot of times that you use all three PnR tools at the same time
and block by block basis.
And then they just go "Which one generates the best block?"
That's the most common way. It's very rare that a company says,
"We're going to be going all Synopsys" or "We're going to be
going all Cadence" or "We're going to be going... all anything".
Users mix and match. Because the blocks are so huge that... a
block today was a chip in my youth.
Sawicki: A block today was one of those little 74F series things in your
youth. (laughter)
Cooley: How do you counter ... what Anirudh was saying is at the nodes at
5nm and 3nm ... things are integrating together. You're getting
synthesis integrating with timing and ECOs and everything like
that. You [Avatar] are still going to be a PnR point tool. How
do you compete in that environment?
Sawicki: Well, I think that there is certainly something to be said for
the integration. But I don't think it is the most compelling
action point right now. I do think that the Avatar technology
comes with some really incredible, integrated detailed synthesis.
And that as we work with our customers to serve their needs in
the PnR market we'll look at how we can leverage that and what
our plans will be to respond to that.
You've got to keep in mind, this is a very small company we've
bought -- but that has some amazing technology we'll be able to
leverage.
I can tell you the customer base when we went out and told them
we did this, they did not say, "Oh why are you even talking to
me? Where's your synthesis solution?".
Our customers said, "Thank you for buying this. This Avatar
stuff is incredible. We really want to work with Avatar and
now we know we can because it belongs to Mentor."
[ Editor's Note: Did you catch this? What Sawicki is very
politely saying here is "even with a small R&D staff that
was under siege, Avatar PnR kicked ass in benchmarks
against both Aart's and Anirudh's PnR offerings.
Now imagine how formatible Avatar PnR will be with Siemens
Mega lawyers keeping Aart's lawyers at bay and with Siemens
Mega $$$ funding Avatar R&D!" Whoa! - John ]
Cooley: So Anirudh's strategy and Synopsys' strategy is they're going to
marry everything. Synthesis, timing, ...
Sawicki: Well, I don't think that's block versus full chip. That's
whether or not you're doing those blocks or top-level chip with
the same synthesis and place & route engine. I think people will
continue to do block-level design at 5nm and 3nm, that's not
going to change.
Our strategy is that, certainly for the short-to-medium term, as
we ramp this product and scale it in the market, we will be
viable as a netlist-to-GDS2 solution, while we work on what we're
going to respond to in the longer term for any synthesis needs.
Cooley: What's going to happen with Badru and the Nitro group. Is that
going to be gone or what?
Sawicki: Nitro team will augment the Avatar team.
Cooley: Is it going to merge into one product?
Sawicki: No, we're going to merge the teams, not the products.
Cooley: Okay, cool. Thank you.
---- ---- ---- ---- ---- ---- ----
Related Articles:
Avatar Aprisa benchmarks vs. Innovus/ICC2 at 7nm is Best of 2019 #7b
Avatar/AtopTech's big comeback in digital PnR is #4a "Best of 2018"
Chi-Ping Hsu answers the 6 big technical doubts about Avatar PnR
Synopsys layoffs means ICC2 rewrite is unknown for 3 to 4 years out
92% of EDA users say Justice was NOT served in SNPS-Atoptech case
Join
Index
Next->Item
|
|
