( ESNUG 475 Item 7 ) -------------------------------------------- [09/18/08]

Subject: Magma FineSim, Nascentric, Berkeley, Z-Circuit

THE SPICE OF LIFE: In the SPICE world, the user mindshare was equally
split between the year old Magma FineSim simulator and BDA's SPICE related
tools, and Nascentric optimizing its SPICE to run on Nvidia GPU's
(engineers liked the idea.)  In the "Lies" section of this report, you
find one engineer dissing Mentor ADiT's numbers.  Not good.

Magma SiliconSmart and Z-Circuit did well in lib characterization, too.


    "What were the 1 or 2 or 3 INTERESTING specific tools that you
     saw at DAC this year?  WHY where they interesting to you?   
     (If any were under NDA say it and I'll keep you anon on them.)"

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

  At the Magma booth I saw presentations on FineSim SPICE and FineSim Pro.
  They are tools we are following since several months so I was already
  aware of them and their features.  I went there to meet Magma people and
  to look at the product road map and future enhancements.

   - FineSim SPICE is promising technology in terms of SPICE accuracy
     with a high simulation speed.  It's good on large transistor
     level IP blocks and analog macrocells.  It is probably the only
     one on the market fully supporting multi thread and multi process.

     We had very good benchmark results with respect to the pure SPICE
     engines and fast SPICE engines used at maximum accuracy.

   - FineSim Pro is a fast SPICE simulator.  It is good in terms of
     speed and accuracy but sometimes performance of competitors are
     better.  It also supports multi threading and multi processing
     but the performance improvement is not as good as FineSim SPICE
     because it's more difficult to split computation in a fast SPICE
     approach than in pure SPICE single matrix approach.

  I think that FineSim SPICE today is a candidate to replace SPICE-like
  simulators on IP blocks and analog macrocells when you want SPICE
  precision on a large transistor level block.  It can be used also in
  characterization flows for standard cell and analog libraries or memory
  characterization.

  FineSim Pro is good, but one competitor that I can't name is beating it
  on our benchmarks.  It needs to improve on capacity and behavioural
  support.  Built-in flash cell model is not supported as well yet.

      - Pierluigi Daglio of STmicroelectronics

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

  (ii) Magma Finesim Pro:  For companies that are interested in circuit
  simulation of large, custom designs, Magma showed Finesim Pro, a
  simulator that provides full Hspice accuracy (applicable to blocks that
  need high accuracy) and Fast Spice performance (for high capacity, full
  chip simulations) within a single kernel.  The tool touts an engine with
  native parallelism for higher efficiency and capacity that scales with
  the number of processing nodes.

      - [ An Anon Engineer ]

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

  Nascentric: Nvidia GPU 128 processor accelerated SPICE on "cheap"
         hardware is way nerd cool.  Dunno about accuracy/performance
         of tool - it's just cool.  Yes Gauda claimed to do some OPC
         on there as well.

  DAC's technical program was better this year.

      - [ An Anon Engineer ]

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

  Most interesting 2 - Nascentric's parallel hardware SPICE acceleration.

  I'm not really an analog user, but the implementation interests me.  I'm
  not sure that GPUs are the way to go (maybe yes, maybe no) for assorted
  types of parallel HW acceleration, but hats off to them for being the
  first one in the water.

      - [ An Anon Engineer ]

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

  Nascentric OmegaSim was also very interesting and we shall be giving it
  a try sometime.  Pricing model is very interesting but might be a
  challenge for managers trying to plan a budget.  Some work to do on
  that.

      - Andrew Cole of Foveon Inc.

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

  I have to say that Berkeley Design Automation's (BDA) Analog FastSPICE
  was the most interesting tools I've seen, both in what they claim and
  what we've observed here personally.  The numbers they claim are NOT
  fantasy.  They are actually real, and every time internally a skeptical
  design engineer runs their tool on their toughest design, that engineer
  comes away impressed.

  Magma's "turbo" SPICE segfaults regularly on those same designs.

  Now if only BDA could make a mixed-signal tool to compete with Mentor's;
  their existing solution has all the same issues that Spectre Verilog has.

      - [ An Anon Engineer ]

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

  Unfortunately due to lack of time I was not able to attend any official
  Magma presentations.  Therefore, I obtained information about Magma tools
  only from my discussions with Magma people and its customers.

  I was impressed with accuracy, capacity and performance claims of Magma
  SPICE-level circuit simulator "FineSIM".  Their distributed processing
  option is an advantage for simulating large analog and mixed-signal
  circuits with silicon-accurate results.

  I also was impressed with Magma "SiliconSmart" library characterization.
  It claims good performance and high accuracy for creating models of
  nanometer standard cells and IOs.  I like its support for a wide
  variety of industry standard formats.

      - Vladimir Zolotow of IBM

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

  I only attended Magma's library characterization demo.

  They seem to provide all the expected characterizations (timing, power,
  noise) in the standard syntaxes (.lib, Verilog, IBIS), including various
  statistical models, although this was pretty standard for all of the lib
  char tool vendors.  (They said they would support ECSM power & noise
  modeling by end of year.)

  They've got a built-in SPICE simulator engine, "FineSIM", you can use for
  the characterization or an external SPICE simulator.  (FineSIM obviously
  has internal hooks to make use of it for characterization more efficient,
  but would probably require qualification with company-mandated SPICE
  flows.)

  Along those lines, you can buy extra licenses to run concurrent sessions
  of FineSIM, or you can do concurrent sessions of whatever external SPICE
  simulator you normally use (I think it was without licenses).

  I don't have much more detail than that, like general price levels or
  specific functionality; my focus was to see if the various tools were
  interesting enough for our company to evaluate further.  For any more
  detailed information, like accuracy, capacity or benchmark results, we
  would have to actually do an evaluation.

      - Kim Flowers of Lattice Semi

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

  Z-Circuit Automation - they have a new F/F characterization engine with
  a real algorithmic development that lets you get layout and power supply
  related timing models in under a minute vs 1+ hour in the past.  This
  has been a stumbling block for the characterization industry since the
  Meta-Software days (1980's).  With the low power methodologies now in
  place, supply and load accurate timing of the retention flops and the
  load flops are the most critical feature of "multi-mode" restart timing.

      - [ An Anon Engineer ]
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)