( DAC 04 Item 30 ) --------------------------------------------- [ 02/09/05 ]

Subject: Tanner L-Edit & T-SPICE

CHEAP IS GOOD -- Shamelessly promoting a "we're cheap, but good" philosophy,
Tanner prides itself in being the low cost alternative to all the other
high priced custom layout tools.  Tanner has a fanatical base of lots of
tiny customers with small EDA budgets.  To Tanner, small is beautiful.  And
to the small customers, Tanner is beautiful.  (Awww...  Ain't love grand?)


    I have done numerous designs using L-Edit the last couple of years.
    Everything from a simple high-voltage controller to a 2.5M transistor
    Monolithic Active Pixel Sensor.  This latter chip (a 3mm x21mm monster)
    was the first time I ran into serious problems -- the most serious
    being a limitation in S-Edit of only 998 ports on a symbol, which meant
    that I couldn't do full LVS (am being told will be fixed in a future
    release).

    Tanner Strengths:

        - Windows based
        - Easy to use
        - Good initial starting libraries
        - Bug free

    Tanner Weaknesses:

        - Windows based
        - doesn't scale well to very large designs
        - no support/library for IBM SiGe process

    We are also using Cadence in the lab, though I myself have not been
    using it extensively (am leaving it to the young engineers) and loathe
    the Cadence look and feel.  We are currently doing a design in SiGe
    and the layout learning curve has been painful.  The design was
    supposed to have been done months ago -- due in part to bugs and the
    very steep learning curve involved.

        - Gary Varner, University of Hawaii


    I thought L-Edit was very easy to use in my latest chip design of about
    3,000 transistors.  I can compare it only to Paragon.

    There are a number of things L-Edit it does that are not in Paragon.
    Scaling is probably the most important of these.  We laid out our design
    using scalable rules because at first we weren't sure which process or
    which foundry we would use, and then adjusted it in native rules to meet
    the chosen foundry's native DRC.  This would have been impossible in
    Paragon: there, they assume you're in native rules from the start and
    make minimal provision for those who see a benefit in switching from
    scalable to native (or from 0.5 to 0.35).

    I'm not sure if Paragon has these features, since I ended up doing the
    bulk of the layout in L-Edit, preferring a number of GUI features, but
    these are some things I liked most: the array feature of cell placement
    was very easy to use and modify arrays.  This is particularly useful for
    active contact arrays, but also for exact placement of many flip-flops,
    etc. I didn't use the schematic driven layout, but it is there if one
    wants to use it.  I decided (probably incorrectly) that my design was
    too simple to use it.  I've seen Thuong U demonstrate it, and I'd
    probably use it on my next scratch design.

    The one area of L-Edit that Paragon MIGHT have an edge is LVS.  I say
    MIGHT, because we never got Paragon's LVS up and running in time to be
    useful (it couldn't recognize our capacitors or diodes).  If we only had
    transistors in our layout, Paragon apparently has the feature of
    displaying the extracted schematic as a schematic, and not just a net
    list, which should make tracking down bugs easier.  I didn't actually
    try this feature - that is what was claimed.  I saw a demo that sort of
    demonstrated that claim.   We tried to get this up and running in the
    last weeks before tapeout, but the ext deck never quite worked properly,
    so we didn't use it.

    If there is one feature I would recommend to L-Edit, it is to provide
    some visual representation of the extracted SPICE file, so that one can
    see, for example, that this whole set of logic has CLK mixed up with
    CLKB, or whatever.

    Over all, L-Edit outperformed all the naysayers in our group, who began
    with the assumption that only "Cadence compatible" systems were worth
    working with.  Our chief critic ended up concluding that L-Edit was much
    better than he originally thought (he is quite familiar with all of the
    tools you mentioned, and he picked Paragon as the L-Edit alternative).
    He plans to continue to use L-Edit at least for the scalable -to- native
    conversion feature.  (Next layout will be in scalable, even though we
    have landed the process.  It's much easier to do the layout with a
    lambda grid than with a 0.05 um grid.)

        - Mark Zdeblick of Proteus Biomedical


    I've seen Silvaco's product and it seemed very clunky and still in
    development.  Not ready for prime time.  I've seen Mentor Graphics, and
    while it is very nice, its cost is out of our grasp.  There also seems
    to be an IT maintenance issue with such a complex system as MG, because
    you can't just load it on a PC and start working.  Not keen on their
    method of licensing either.

    Tanner is well suited to what we do.  Our designs are strictly analog
    with less than 400 transistors per layout, sometimes less than 100.
    Autorouting is out of the question for us; it all has to be done by
    hand.  We don't need the most cutting edge or complex layout tool to
    handle our designs, so we haven't felt a strong push to migrate to
    another platform.  We haven't used Tanner's simulator or schematic
    capture, although we've heard that the former is pretty weak and the
    latter is pretty good.

    Tanner has its share of weaknesses.  We've run into:

      1) we've wrestled with output plots for a long time, mostly because
         Tanner uses a 3rd party plot utility which isn't entirely
         integrated into their package.
      2) LVS error reports are very cryptic and difficult to decipher.
      3) the wiring tools are rudimentary and are missing nice features
         that more sophisticated packages offer.
      4) setting up the DRC rules is pretty clunky and could use refinement

    I could come up with other stuff, but those are some of the basic
    issues.  But in general, it's a reasonably good design tool.

        - David Curry of That Corp.


    My experience has been limited to Virtuoso, Silvaco Expert, and L-Edit.

    I used Virtuoso in school, so naturally it was my first and my most
    preferred tool.  When I started working at AnalogicTech, I had my first
    run-in with L-Edit.  I wasn't excited about starting work with L-Edit
    because of things I had heard in regards to it.  However, I soon
    discovered that L-Edit was amazing in its simplicity; I was doing my
    first layout the second day I was here, and it was a big block, nothing
    small.  L-Edit is laid out amazingly well; the menu options and all
    other features are well named and placed in a logical location.  I
    personally think that anyone who has worked with any layout tool will
    find it very easy to get adjusted to L-Edit.  Its UPI (User Programmable
    Interface) is very robust; it offered me a chance to write a really
    elaborate device builder.  The interpreted C is well laid out and
    Tanner offers library functions that take advantage of all the power
    and flexibility offered by the tool.

    L-Edit has one major problem, and that is the fact that it crashes at
    least once every day.  Some layout designers, including myself, have
    lost a few hours of work at a time, but all of this can be avoided by
    saving on a regular basis.  We've all learned to do that, and since
    then, the crashes have become just a slight annoyance (and I'm sure
    Tanner will clean everything up over time).

    Another important point: Tanner's support is amazing.  Their responses
    are very quick, I have a person who I communicate with at Tanner who
    knows me and the company, and provides us with aid and information
    quickly.  I started working as a layout designer only a year ago, so
    I have not had the chance to experience other tools; although I
    have had a short experience with Silvaco Expert (and I think that short
    experience was too long as it was).  I was very disappointed with
    Expert.  Its interface looks like a cheap imitation of Virtuoso's.  My
    opinion of the tool isn't very high.  The tool and all of its features
    are laid out very poorly; menu items are not self- explanatory and data
    input, like editing the properties of objects, require a lot of time to
    figure out.  I was very disappointed with the tool and would be very
    annoyed if I had to work with it.

    To wrap it up:

     - Virtuoso & L-Edit have been wonderful tools to work with.
     - L-Edit is by far the quickest to learn and pretty useful/productive.
     - Two thumbs down for Silvaco's Expert.

    Again I want to say that this is based on my personal experience with
    L-Edit, Virtuoso, and Silvaco Expert.

        - Marius-Paul Dumitrean of AnalogicTech


    I hear the Tanner stuff is pretty hokey, but have not looked at it
    myself.  All the design kits seem to be based on Cadence.  I used to
    like Magic in my younger days, and it's free.

        - [ An Anon Engineer ]


    Well, I used L-Edit for our mask layout.  I used to use Cadence and
    was very skeptical on converting to L-Edit.  I started playing with the
    software and then Tanner offered a training one day and I decided to
    attend it.  Found that to be very, very helpful.  There were features
    that L-Edit had that seemed user friendly than Cadence.  But then there
    were also things that Cadence can do and L-Edit can't.  As I used L-Edit
    more, I really got a feel of it and liked it a lot.  Also, as we upgrade
    to newer versions, most of the features that Cadence are now available
    in L-Edit.  I look at it as this:

       Cadence = stickshift (lots of manual work, Unix-based and not
                 everyone knows Unix!)

        L-Edit = automatic (easy, user friendly and Windows based!)

    Now, I won't even go back to using Cadence.

        - Hazel Gutierrez of International Rectifier


    My overall Tanner experience has been excellent.  I had about one
    year of experience with Cadence at Honeywell and then used an older
    version of Cadence here at NVE for a while.  We actually used ICED for
    a few years and I was one of the people to pursue Tanner as a layout
    tool.  I started using Tanner L-Edit with v 8 and have been impressed
    with Tanner's ability to improve their tool significantly with each
    release.  Many software companies sell you their software and then
    don't update it anymore.

    I have been fortunate to be part of Tanner's Beta testing program which
    maybe has made my customer support experience very good as well.  I
    have found when I communicate software bugs or request enhancements to
    them, that they listen.

    Their software is very affordable as well.  Maintenance can be costly
    but you get what you pay for.

    The only real complaint I have is in Tanner documentation.  I have
    tried to keep up with whatever documentation they have generated and
    have found sometimes things aren't as clear or as forthcoming as they
    could be.  I have suggested that they do more tutorial type docs.
    I know they make money on their training classes so they may not be as
    willing to give you "training" type information in the documentation.
    For the most part, Tanner documentation is OK, but could be better.

    On a scale of 1 - 10, I'd give Tanner L-Edit an 8 or 9 depending on
    my mood that day.

        - Joan Lange of NVE Corp.


    When using the Avanti toolset, I was only concerned with tweaking,
    and/or reviewing an existing design.  In my current situation, I am
    using Tanner to do a full custom, ground up.  Here is my breakdown:

    Tanner Strengths:
    1. $$, small company, no budget, its cheap and gets the job done.
    2. extraction of cells/transistors is logical and transformable.
       i.e,, when we need to create a little hack to the normal
       standard cell extraction, we can usually find a quick way to do it.
    3. customer support is excellent, usually solving our problem
       quickly; although the solution is usually we did something wrong,
       which was not obvious why it was wrong.
    4. SPICE simulation is nicely integrated and easy to deal with.
    5. cell development is smooth.

    Tanner Weaknesses:
    1. object/layer selection.  not sure what it uses to identify what
       line/wire/layer we are trying to select, but it is frustrating. 
       My understanding from talking with the company is that the tool
       is primarily used in cell development -- so our large design is
       showing some of the weaknesses it exhibits when designs get larger.
    2. (which is more related to #1), moving something you didn't intend to
       move.  At some point, you click, think you are moving the two objects
       you have selected, only to find out, you are moving three.... one of
       which is completely off the screen!  It does have a little info in
       the corner to tell you "Hey, stupid, you are moving three objects",
       but since the third object is off the screen, you just don't get the
       visual feedback to tell you something is not right.
    3. This is a bug I presently have submitted, and after two years of
       use, it is the first real bug I have found.  It is having trouble
       generating an SDF after PR on a standard cell design created from a
       Cadence BuildGates EDIF netlist.  This is the first time I have used
       the standard cell PR tools.  Hopefully, this is another bug where we
       did not do something proper, but at this point it is not looking like
       it.  A side note, after this error, the working database was
       corrupted beyond use.  Glad I had a back up.
    4. SPICE Waveform viewing, arg, I wish it was better/more robust/more
       configurable.  ...one of those, "If I could take the Modelsim
       waveform viewer, and make it read in this output"

    When comparing Tanner to Avanti, it is quickly obvious where the hundred
    of thousands of dollars go in Avanti.  However, for those dollars, I
    would expect something light years ahead of the Tanner Pro tool suite,
    which is not the case.  Tanner doesn't do everything, and it doesn't
    have the quick shortcuts to do it, but it does get the job done for a
    small company with a very tight budget.

        - Greg Alkire of PicoDyne, Inc.


    How does Tanner compare to Cadence tools?  I think Tanner will catch up
    with Cadence eventually.  But for the price, it can't be beat!

    Upside:

    1. PC based - can take it anywhere and work at home if you are on a
       deadline.
    2. L-Edit does an excellent job of place & route of standard cells.
    3. Excellent documentation - practical.
    4. Best customer support.
    5. Importing/exporting GDSII files is a breeze.
    6. The tool remembers the setup, and any cell created maintains the
       setup like design rules, display layers, technology file settings.
    7. Does a great job of taking EDIF files to place & route.
    8. Sophisticated L-Edit tool for production level designs.

    Downside:

    1. Does not allow cell rows to flip over so power busses may be shared,
       creating a larger core area than Cadence's Silicon Ensemble.
    2. May be nice to be able to have unlimited Boolean functions for
       derived layers versus just 2 or 3.
    3. T-SPICE program not as intuitive as OrCad's P-SPICE.  With T-SPICE,
       you have to create a new page for most simulations - DC, AC, etc.
    4. T-SPICE doesn't have all the models that OrCad supplies.  Only
       generic transistor models are available for edit.  Would be nice to
       have models for specific op amps like the LM108 or LM741 and other
       components.

    Overall, I'd give Tanner Tools an A+!!!  Excellent tool & customer
    support.  I'm still learning a lot from it.  L-Edit is a very
    sophisticated layout tool which can stand against the best of them.
    And how can you complain when the price is so affordable?

        - Mike Fujita of Defense Microelectronics Activity


    I use L-Edit for very non-traditional applications (printable,
    polymer-based electronics), so I never use many of the features
    customed designed for the silicon world.

    That being said, I love the program.  I find it intuitive, robust,
    cleverly designed, and easy to use.

        - Geoff Nunes of DuPont


    We selected Tanner because it is intuitive and simple to use (it takes
    15-20 minutes for me to show anyone who has experience with a drawing
    program).  Its hierarchal/cell structure makes it easy to produce
    patterns with a wide variety of permutations (for example, sequential
    arrays of annular rings that increase in size by increments of 10%);
    this is very important for researchers who aren't sure exactly what
    dimensions will work for their experiments and want to explore them
    all.  L-Edit allows all angles and curves/arcs -- particularly
    important for our non-semiconductor device researchers who make up
    the majority of our lab member base -- and whose files rarely follow
    Manhattan geometries.

    L-Edit allows us to export limits on file settings -- the number of
    segments defining an arc probably being the most important (since files
    with curves and arcs explode in size when you fracture them for mask
    writing.)

        - Mary Tang of the Stanford Nanofabrication Facility


    We are looking forward to seeing capability to both place and optimize
    designs with multiple power domains and multiple cell types flat.
    Cadence, Magma, and Synopsys are promising this for the new year.

    We presently use Tanner tools for custom layout.  It works OK.

        - [ An Anon Engineer ]


    Among the others I am only familiar with Tanner L-Edit.  It is not in
    the same league, but it is cheap, and does the job OK for smaller
    modules.  For full chip layout it is struggling.

        - [ 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)