( DAC'17 Item 9 ) ------------------------------------------------- [06/01/18] 

Subject: Cadence Voltus, Ansys RedHawk, SeaHawk, Tempus, Primetime, Dorado

LAST YEAR WAS BAD FOR APACHE/ANSYS: For the Ansys/Apache folks, DAC'17 in
Austin, TX was a crappy time for them.
      
The problem was John "Jolly" Lee was trying to wow the EDA world with his
still newish SeaHawk "Big Data" tool...  I'll let the Cheesy Must See List
write-up explain the situation...

 4.) Ansys Gear SeaHawk is John Lee's R&D guys bringing those much talked
     about Big Data into the RedHawk IR-drop/EM franchise.  (ESNUG 554 #1)
     It doesn't change RedHawk itself -- SeaHawk guides RedHawk for best
     QoR.  Claims are SeaHawk can reduce die size by 5%.  The gotcha is
     SeaHawk/RedHawk is getting inconsistant results.  (ESNUG 563 #10)
     (booth 647)  Ask for John "Jolly" Lee.  Freebie: stuffed bulldog

     Ansys Apache RedHawk is full-chip/3d-IC power integrity analysis and
     sign-off, transients, simultaneous switching noise package/PCB with
     distributed processing.  Scalable to 16-32 machines (128-256 cores).
     "500M insts with 8B resistors while keeping flat simulation accuracy"
     Vector-based and vectorless.  Clock jitter.  TSMC 16/10/7nm FinFET.

     NEW! -- RedHawk-SC claims "IR-drop in 6 hours on a 1 billion gate
     chip on a 16G machine" and "does 1000 scenarios overnight".  Says
     its 2 years old, so it's a rebranding of earlier SeaHawk/SeaScape.
     Maybe they fixed the inconsistency/accuracy problems now?

         - from "Cooley's Cheesy Must See List for DAC 2017"

Those damned inconsistancy problems!

But that was last year.  Since in the survey respones and from the rumor
mill, I didn't get a lot of complaints about them, I'm guessing that the
SeaHawk plus RedHawk inconsistancy problems are solved -- or that users
have just bailed on the SeaHawk altogether and stuck with their tried and
true RedHawk.

The problem is I only got one positive user reponse about Apache/Ansys:

  "I like the ICC2/Redhawk integration.  It's unusual to see two EDA
   competitors work together like that."

And that's it.  Nothing about SeaHawk.  Nothing about have better RedHawk
is.  Just a "it works with ICC2" comment.  No news is good news?

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

MORE INNOVUS/VOLTUS ANSWERS: On the flip side, for the 2nd year more user
talk was on using Innovus with Voltus, or Innovus with Voltus and Tempus.
That is, they were talking about pairing two or three tool in the same EDA
tool flow as being the "Best of" for 2017.

NOT the tools themselves, but the pairing and threeways of the tools.

Why is this?  Aniridh's CDNS sales army has been in a full frontal assault
against Apache/Ansys by chatting up the advantages of getting IR-drop
closed inside an Innovus PnR flow.
(click to enlarge pic)
That's 93% fewer victims.  And 26% less IR-peak.  All within Innovus PnR.
Sexy stuff for chip designers doing 16/14/10/7nm. ... and apparently their
message is catching on.

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

      QUESTION ASKED:

        Q: "What were the 3 or 4 most INTERESTING specific EDA tools
            you've seen this year?  WHY did they interest you?"

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

    Cadence Voltus will give Ansys-Apache a run for their money.

    Last year Voltus wasn't quite ready for signoff, but this year we are
    using it for signoff on a number of devices and not just IR, but
    chip-package interaction as well.  

    It's displaced Ansys-Apache and as they continue investing and offer a 
    compelling solution more integrated with P&R tools.  Over time, this
    Voltus-Innovus combo will clearly take the market.  

    At the end of the day, Ansys is proud of their tools but we feel they're
    using revenue to invest elsewhere and not continue improvement of their
    IR-drop tools.

    Voltus' focus needs to be around usability.  It's good technology.
    But hard to use.

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

    We've used Cadence Voltus inside Innovus and Virtuoso

    From what we've experienced, Voltus' performance gain is relative to 
    design size.  

        -  Most small designs won't benefit much from distributed mode; 
           plus the multi-threaded solver in Voltus is very efficient, 
           so single machine performance is already great.  

        -  The larger designs can benefit from multi-machine distributed
           configuration.  The gain quickly saturates -- however, the 
           distributed mode enables us to run designs that cannot fit on
           single machine.

        -  Using distributed mode, Voltus can support very large designs.
           So far, it can handle all our designs without any problems.

    We have successfully run Voltus and displayed the results within 
    Innovus.  The two tools share the same GUI frontend, which is a bonus 
    for designers.  We don't currently use the ECO/optimization.

    Cadence informed us they are fully certified down to 7nm and is working
    with TSMC, Samsung, and GF for their under 7nm nodes.  So far, we have
    not encountered any problems related to foundry support.

    Voltus-FI has a GUI interface build-in to Virtuoso which the designers 
    like.  A user can run the tool and display the results all within the 
    familiar Virtuoso platform.  On the Voltus end, the latest build does 
    have Sigrity commands built in, but the feature set seems to be very 
    primitive, and looks like it's still work in progress.

    Cadence seem to be gaining share in the digital market; it's no longer 
    an EDA company feeding on Virtuoso.

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

    We use Cadence Voltus as a standalone tool.  For us Voltus outperformed 
    Ansys Redhawk in terms of both runtime and accuracy.

    Seems this is a combination of both Redhawk stagnating a bit since the
    Ansys acquisition and Voltus improving.

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

    Cadence Voltus  

    Voltus is the best standalone power analysis and repair tool available,
    especially when integrated with Innovus and Virtuoso.  

    This combination saves days over manual solutions.

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

    Cadence Voltus

    Voltus seems to be doing better than Redhawk (Ansys/Apache) for power 
    sign off.  

    I've heard it's fast, accurate, and has the largest capacity on the 
    market.
  
        ----    ----    ----    ----    ----    ----   ----

    Voltus is good.

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

    Voltus + Innovus

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

    I like the ICC2/Redhawk integration.

    It's unusual to see two EDA competitors work together like that.

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

    Tempus + Voltus:

    Tempus and Voltus built on the same code base, i.e. truly integrated.

    This is unique.  It's the only integrated signoff for STA and signoff
    power analysis -- critical for 16nm or lower with low supply voltages,
    where you can't just rely on standalone STA and power signoff anymore.

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

    Tempus 

    At 7nm, Tempus right now is the better sign-off timing tool, with
    higher accuracy and much better turnaround time than PrimeTime.  

    Tempus' ECO integration with Innovus works very well.

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

    Tempus

    Tempus is the same or better as PrimeTime.  Once the brain trust left 
    Synopsys, Cadence really started to make progress on signoff -- probably
    5 years back Cadence when was "good enough" for signoff.  Now they are
    solid in signoff.  We've had no major issues on correlation -- in fact
    better on advanced nodes than it was even 5-10 years back in ETS days.  

    Advanced waveform propagation and PBA analysis are very nice.  

    EMIR integration is not really not a concern for us, as there are too 
    many other uncertainties which are more significant and not modeled 
    either (e.g. simultaneous input switching).

    Cadence R&D needs to continue to focus on TAT and throughput.  Not just
    being able to distribute but being able to run fast single thread since
    signoff is usually 10s-100s of scenarios that need to run in parallel.  
    We can't really dedicate 16/32-cpu machines to each one without 
    incurring a huge IT cost.

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

    Tempus is riding Innovus' coattails.

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

    Tempus TSO will do a break even and will start competing with Dorado.

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

    For Innovus users, Dorado is in trouble if Tempus TSO takes off.

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

    We live and breathe PrimeTime here.

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

    Primetime

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

    We like the Primetime SIGs at DAC

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


Related Articles

    Anirudh and Sawicki on iffy Apache IR-drop #'s vs. Voltus/Innovus
    A second CDNS Voltus-DP vs. ANSS Gear RedHawk-DMP user benchmark
    The Nvidia stealth benchmark of CDNS Voltus vs. ANSS Gear RedHawk
    CDNS Innovus/Voltus, Apache Redhawk, ICC/ICC2, MENT Nitro-SoC


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 1991-2025 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)