( ESNUG 495 Item 9 ) -------------------------------------------- [11/17/11]

Subject: CDNS bought Azuro because Synopsys/Magma can't do multi-mode CTS

> Then in 2007-2008, Azuro had Toshiba and NXP publically endorse their
> PowerCentric clock tree synthesis (CTS) and gate-level clock gating tool.
> But after that initial great start, the Azuro folks never got the word
> out about any successes in any *independent* EDA venue.  They stupidly
> seemed to think that putting up canned "user" quotes on Azuro.com and
> press releases on the business wire was sufficient!  D'oh!
>
>     - from http://www.deepchip.com/wiretap/110713.html


From: [ I Am Legend ]

Don't care about their marketing, but on the tech side Azuro finds a lot of
useful skew yet still keeps your clock trees low power.  Their tool also
understands skew and clock-gate-enable timing and MCMM and OCV derates.

It's not just fine tuning on a few reg-to-reg paths, it builds all your
clock trees from scratch and does incremental logic optimization (sizing,
placement, etc.)

On a fully taped out chip I've seen a 50% power reduction in one 120 K gate
block and about 30% power reduction in another 1.4 million gate block by
using Azuro.  (Our baseline flow is Cadence.)  These 2 blocks are multiply
instantiated and so made up about 80% of our die-area so these results were
a big deal for us.

    - [ I Am Legend ]

         ----    ----    ----    ----    ----    ----   ----

From: [ I Am Number Four ]

Please keep me anonymous.

We have used Azuro's PowerCentric CTS tool for many years on many designs.

Aside from the savings in dynamic power, the clock trees produced are high
quality, low latency and the GUI is very powerful for debugging and
visualization.  Their Rubix tool is compelling when extra performance is
required with an added benefit of better IR drop.

    - [ I Am Number Four ]

         ----    ----    ----    ----    ----    ----   ----

From: J-F Vidon <jean-francois.vidon=user domain=stericsson not mom>

Hi John,

Evaluated PowerCentric last year.  Used it on two 800 K inst blocks.

We integrated PowerCentric into our EDI P&R toolset flow by simply replacing
the entire Encounter ClockTree Synthesis step with it.  We had to use DEF
and netlists to exchange data in/out Azuro.

Cons:

 - DFT and gate-level simulation needed updating to cope with "more"
   complex clock-gate enable structure.

 - I also had to create a UPF as CPF wasn't supported at that time.

 - Our P&R flow "more complex", as we had to share data between 2 tools.

But it was worth it for a 15% dyn power reduction vs. our baseline Cadence
EDI clocktree synthesis flow.

Timing correlation and equivalence-checking were hassle-free.

Thanks to their reactive FAE and R&D, integration only took a few months.

Since Cadence acquired Azuro, I look forward a more integrated flow.

    - Jean-Francois Vidon
      ST-Ericsson                                Austin, TX

         ----    ----    ----    ----    ----    ----   ----

From: [ Elf Boy ]

Hi, John,

We had 2 designs that couldn't do multi-mode CTS in neither Talus nor ICC.
Both the SNPS and LAVA tools each required mucho hand-holding and custom
scripting for each mode -- not a viable solution for us.

What finally closed our 2 designs (full P&R and real MM CTS) was a combined
Encounter with PowerCentric flow.

From where we sit, it made perfect sense for Cadence to buy Azuro.

    - [ Elf Boy ]

         ----    ----    ----    ----    ----    ----   ----

From: [ My Name Is Earl ]

Hi, John,

Keep me DEEPLY ANONYMOUS.  :)

You missed the mark on your Azuro guess, though you were right about their
marketing.

You did a whole write-up on Azuro and you don't mention Rubix?  Rubix, not
PowerCentric, is the game changer.

Cadence with this one little acquisition will become THE dominant force in
layout and timing closure for the next 5 years.  Rubix is, without a doubt,
the best CCOPT tool ever developed-light years ahead of what everyone else
has.

I say this as the engineer who, completely skeptical at first, staked my
whole career and reputation on obtaining this tool after running it on some
of the toughest benchmarks we have in house.  I have truly never seen
anything like it, aside from our own in-house tool -- which is now defunct
because we didn't want to be an EDA company.  (No money in it.)

    - [ My Name Is Earl ]

         ----    ----    ----    ----    ----    ----   ----

From: [ Call me Ishmael ]

Keep me anon, OK?

On fast designs here (not in my group), Azuro has been able to get their
designs to run faster using Rubix for CTS.  It optimizes the datapath
and clock tree simultaneous.

Enclosed is the Rubix/CCOPT whitepaper they sent us to explain it.

    - [ Call me Ishmael ]

         ----    ----    ----    ----    ----    ----   ----

 Editor's Note: This Rubix/CCOPT whitepaper is #65 in Downloads.  - John

Join    Index    Next->Item






   
 Sign up for the DeepChip newsletter.
Email
 Read what EDA tool users really think.


Feedback About Wiretaps ESNUGs SIGN UP! Downloads Trip Reports Advertise

"Relax. This is a discussion. Anything said here is just one engineer's opinion. Email in your dissenting letter and it'll be published, too."
This Web Site Is Modified Every 2-3 Days
Copyright 1991-2024 John Cooley.  All Rights Reserved.
| Contact John Cooley | Webmaster | Legal | Feedback Form |

   !!!     "It's not a BUG,
  /o o\  /  it's a FEATURE!"
 (  >  )
  \ - / 
  _] [_     (jcooley 1991)