( SNUG 04 Item 18 ) ---------------------------------------------- [08/11/04]

Subject: Synopsys Power Compiler & PrimePower, Sequence PowerTheater

CUT OFF AT THE PASS:  Outside of IR-Drop tools, there's a separate subset
of chip power analysis/optimizers that work at the RTL level.

       Dataquest FY 2002 RTL Power Implementation Market (in $ Millions)

       Synopsys Power Compiler  ############## $13.7 (81%)
         Sequence PowerTheater  ### $3.3 (19%)

Synopsys dominates this segment.  Just to be fair, Synplicity, Magma, and
Cadence all also say they have some sort of RTL power but they're bundled
into their RTL synthesis packages.  That is, they don't sell them
separately, so they don't report any numbers in this catagory to Dataquest.
And for the veterans here, the Sequence PowerTheater tool is the old
Sente WattWatcher.

On the user side, people like Power Compiler but dislike PrimePower.


  12.) Do you use "power tools" like Synopsys Power Compiler and PrimePower?
       How do they fare vs. Sequence Power Theatre, Synplicity Iota, Apache
       RedHawk & SkyHawk?


    We use both Power Compiler and PrimePower.  Power Compiler is simple and
    smooth to use.  PrimePower is totally a mess.  The ASIC vendors have
    been changing the support of power analysis tool.  There is no standard
    library for the power calculation and analysis.  Why are Cadence and
    Mentor giving up this market?  There will be more and more mobile
    devices in the market.  We need the tools to design low power circuits.

        - Edmond Tam of Global Locate, Inc.


    I would like to use Power Compiler and PrimePower, but we cannot
    afford them; especially as long as I can manually achieve betters
    result than them.

        - Marcello Vena of Xignal Technologies AG


    We use Power Compiler extensively and although its fiddly to use, it
    gives good results.  PrimePower was evaluated, but didn't do anything
    more than our in-house scripts.  Haven't touched the others, although
    I did look at Sequence Power Theatre in the past and it wasn't
    anything wonderful.

        - [ An Anon Engineer ]


    We use Power Compiler for automatic clock gating insertion (along with
    designer-created clock gating at "functional" level) and leakage power
    recovery.  Quite slow if you don't use the shortcut (swapping low and
    high Vt cells).  I have been disappointed by PrimePower because I can't
    just feed it a netlist and get a quick (quicker than with Power
    Compiler) leakage power report, the tool requires some activity
    annotation to run (unless gross oversight on my part).  Oh well.

        - [ An Anon Engineer ]


    Don't use power tools.   I looked at PrimePower a long time ago (about
    4 years I think).  The big problem was that gathering the power stats
    during simulation caused the simulator to run out of memory.

        - [ An Anon Engineer ]


    Use Power Compiler for fine grain clock gating but now integrated flow
    with backend, so pretty much useless most of the time.  Looking at
    Apache, motivated by AR concerns and need for dynamic power analysis.

        - [ An Anon Engineer ]


    We kind of used PrimePower and Power Compiler.  We do a lot of up-front
    analysis in the design for power instead of counting on tools.

        - Haiming Jin of Intel


    We've used Power Compiler when it was rolled under the hood of DC.
    Overall I was quite happy with Power Compiler.  I'm not sure about how
    well it compares with silicon though.  The only painfull part is
    re-compiling simulator executables to record switching activity.  

        - Gord Allan of Carleton University (Canada)


    As mentioned above Ambit BuildGates Extreme seemed to do a better job at
    reducing the power than Power Compiler in simple test run by the AEs on
    our code.  Sequence PowerTheatre is more for pre-netlist work and can
    do some optimization.

        - Brett Warneke of Dust Networks


    We use Power Compiler and Apache RedHawk.  Power Compiler can optimize
    based on power goals, but it seems to be primarily useful for simple,
    first-order reports and for inserting clock gating cells.  RedHawk seems
    to be the best at static/dynamic power analysis, particularly at finding
    peak power (current) draws in a design.  One issue with all the power
    tools is that we don't have a mature flow feeding voltage drop data into
    STA and/or into P&R tools.  These flows are still in their infancy.

        - Andrew Bell of PMC-Sierra, Inc.


    Using Power Compiler and PrimePower.

        - [ An Anon Engineer ]


    Don't use them.  We use "manual" clock gating in RTL to reduce power.

        - Juan Carlos Diaz of Agere


    No.

        - [ An Anon Engineer ]


    Power Compiler is all we use.

        - Massimo Scipioni of STmicroelectronics


    Only have used PrimePower and the past tool PowerMill (now NanoSim).
    No comparsion data against other tools.  If the designer understands
    and knows how to expect the worst power situations, and has generated
    vectors accordingly, these tools have been quite useful for me.  Quite
    necessary to ensure power grid EM and IR requirements are met.  An
    important input source for VoltageStorm for example...

        - [ An Anon Engineer ]


    Power Compiler and Magma BlastRail are equivalent.

        - [ An Anon Engineer ]


    Apache RedHawk

        - David Fong of S3 Graphics


    Definitely use Power Compiler for mixed-Vt optimizations and now for
    power optimization in general.  PrimePower somewhat.  We abandoned
    Sequence tools a couple of years ago in favor of PrimePower.  Never
    tried Synplicity Iota (no engineering bandwidth available).

    Desperately need dynamic power analysis - Redhawk currently only
    solution of interest.  No engineering with which to address at this
    time, though.

        - [ An Anon Engineer ]


    We use Power Compiler.  I would not like to have power optimization
    as a point tool.  My attitude to these is "how hard can that be", so I
    assume the results won't diverge a whole lot.  I prefer a tool that
    doesn't add a lot of complexity to my flow.

        - [ An Anon Engineer ]


    We have developed in house tool which gives fine results for us.

        - [ An Anon Engineer ]




 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)