( ESNUG 455 Item 4 ) -------------------------------------------- [06/29/06]

Subject: Users on Apache PsiWinder

TIRE KICKERS -- PsiWinder is still at the evangelical sales stage where
people are trying to figure out what it is and if they really need it or
not.  No biggie there; it was introduced only last year.

 "In your view, how does the new (DAC'05) PsiWinder rate?  What are its
  competitors?  Is it a useful tool or a waste of money?"


    PsiWinder is an interesting tool.  It allows us to take dynamic
    IR-drop VDD/VSS waveform, at each node in the circuit, into
    SPICE-like simulation to get a more realistic timing analysis
    for clock jitter and skew.  I don't know of any other tool that
    can even do it.

        - [ An Anon Engineer ]


    We used PsiWinder to analyze the concurrent impact of SI and power
    integrity for both the clock tree and critical paths.  It has been very
    useful in helping us get accurate delay, slack and skew information
    before tapeout.  Since our power integrity data comes from RedHawk, it
    doesn't take extra time to prepare it for PsiWinder.

    PsiWinder is based on Apache's NSPICE; we chose it for its accuracy and
    performance.  When we compared its results with HSPICE, the difference
    was within 2%.  Our design is 0.13 um.

    Our clock tree information

                         levels    # buf/inv    # leaf node
                  CLK1     11         781          40,269
                  CLK2     10         316          15,714
                  CLK3     12        1296          76,767
                  CLK4     12        1004          58,096
 
    HSPICE (golden) vs. PsiWinder NSPICE

               diff.   diff.   diff.     max.    min.    ave.
               max.    min.    ave.      ratio   ratio   ratio
        CLK1   60 ps   42 ps   51.4 ps   1.84%   1.30%   1.61%
        CLK2   51 ps   25 ps   38.5 ps   1.35%   1.77%   0.91%
        CLK3   69 ps   23 ps   43.9 ps   2.01%   0.68%   1.29%
        CLK4   67 ps   18 ps   40.1 ps   1.93%   0.53%   1.18%

    For the timing path report, PsiWinder generates a PrimeTime incremental
    report and appends a modified timing delay (based on their SPICE
    simulation result) directly after the PT delay number.  We can easily
    see the differences using this report.

    It generates a file which clearly lists detailed information about every
    net and cell (delay, total cap, coupling cap, slew, power drop, and
    ground bounce value).  Because PsiWinder is integrated with RedHawk, we
    can trace the path directly in RedHawk's GUI environment.

    PsiWinder has a convenient way to generate SPICE deck and analyze PI and
    SI noise effects on timing.  It generates information on "input
    sensitization", "vector generation", and "noise signal alignment through
    iterative simulation", which would be time consuming to create manually.

    A PsiWinder drawback is that Apache still lacks a STA engine to analyze
    whole timing paths like Synopsys PrimeTime, so we need to choose which
    critical paths to analyze, which can be difficult.  Their STA engine is
    not able to accurately interpret the timing impact combining by PI & SI
    noise effect.  Even so, PsiWinder's results are usually consistent with
    HSPICE and PrimeTime.

        - [ An Anon Engineer ]


    We use PsiWinder to check critical paths.  It is as if our Focused Ion
    Beam (FIB) ASIC timing change was done after the issue was found in the
    production of the chip.  This saves a lot of money and also brings up
    the mass production yield with first pass success.  To use PsiWinder,
    your SPEF must contain the base cap and the net node to net node
    coupling capacitance for accurate SPICE modeling.  The best switching
    scenario can be created using the statistical algorithm and RedHawk's
    timing-aware function by means of a PrimeTime report for the critical
    paths.  Once the Vdd and Vss waves are reviewed as accurate, and the
    values applied to each cell in the critical path, your timing margin
    can be reviewed and adjusted with SPICE accuracy.  It should one day
    replace the need for a FIB ECO timing modification.

        - Lief Erickson of VikingStarTech, Inc.


    I did not try it yet, but it seems to be a powerful tool as it links
    power integrity, with signal and timing integrity in a more intuitive
    way.

        - [ An Anon Engineer ]


    Will try it if I had a chance.

        - [ An Anon Engineer ]


    No comment yet.

        - [ An Anon Engineer ]


    Not used yet but we are planning on doing it

        - [ An Anon Engineer ]


    Seems like a useful tool but don't have it inhouse.

        - [ An Anon Engineer ]


    Although we don't use PsiWinder, our introduction to it piqued our
    interest.  We'd be even more interested if it eventually morphed into
    a full-design PI-SI analyzer.

        - [ An Anon Engineer ]


    Didn't used it.

        - [ An Anon Engineer ]


    We adopted PsiWinder for PI & SI concurrent analysis on clk tree and
    critical paths.  It helped us get accurate delay, slack and skew
    information before project tapeout.

        - [ An Anon Engineer ]


    Dunno

        - [ 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)