Home > Community > Forums > Digital Implementation > postRoute in EDI13.2

Email

* Required Fields

Recipients email * (separate multiple addresses with commas)

Your name *

Your email *

Message *

Contact Us

* Required Fields
First Name *

Last Name *

Email *

Company / Institution *

Comments: *

 postRoute in EDI13.2 

Last post Wed, Mar 5 2014 9:59 PM by ajay01. 12 replies.
Started by BackerShu 01 Mar 2014 11:21 AM. Topic has 12 replies and 1627 views
Page 1 of 1 (13 items)
Sort Posts:
  • Sat, Mar 1 2014 11:21 AM

    • BackerShu
    • Not Ranked
    • Joined on Wed, Dec 23 2009
    • Posts 7
    • Points 140
    postRoute in EDI13.2 Reply

    Hello All,

    I am trying update the script for DEI13.2, and have some issue while doing postRoute optimization. The old script line is as following:

    optDesign -postRoute

    The following error shows up in the report:

    **ERROR: (ENCOPT-7016): use setDelayCalMode -engine [feDC | signalStorm] -SIAware false before running optDesign -postRoute -si [-hold].

     

    I changed the command to:

     setDelayCalMode -engine feDC -SIAware false

     optDesign -postRoute -si  

    I got two errors shown as following: (similar results while using signalStorm option.)

    **ERROR: (ENCOPT-7071):    The distributed optimization engine is obsolete and has been replaced by the GigaOpt optimization engine which is default in this release. Please update your scripts to remove any explicit settings of 'setDelayCalMode -engine feDc' or 'setDelayCalMode -engine signalStorm' so you get the default Advanced Analysis Engine (AAE) delay calculator to enable the GigaOpt engine.
    **ERROR: (ENCSYT-6692):    Invalid return code while executing 'encounter.tcl' was returned and script processing was stopped. Review the following error in 'encounter.tcl' then restart.

    Taking that  AAE is enable by default, I tried directly

      optDesign -postRoute -si 

    This time I got two errors  

     **ERROR: (ENCOPT-7016): use setDelayCalMode -engine [feDC | signalStorm] -SIAware false before running optDesign -postRoute -si [-hold].
    **ERROR: (ENCOPT-6080): AAE-SI Optimization can only be turned on when the timing analysis mode is set to OCV.
     To run optimization while taking into account on chip variation, set an MMMC environment with the right operating condition.

    Looks like I cannot do -si analysis without turning on the OCV. I checked in the user guide, and turned on the OCV mode as following (I have set a BC lib and a WC lib in the operating conditions):

    setAnalysisMode -analysisType onChipVariation -cppr both
    optDesign -postRoute -si

     I got the following errors:

    **ERROR: (ENCOPT-7016): use setDelayCalMode -engine [feDC | signalStorm] -SIAware false before running optDesign -postRoute -si [-hold].

     **ERROR: (ENCSP-2002):  Density too high (100.0%), stopping detail placement.
     **ERROR: (ENCSP-9022):  Command 'refinePlace' completed with some error(s).

    The first error shows up again! The last two errors are new, and I guess they are related to the OCV mode, which I might not used properly.

     To summarize, I have two questions:

     1) Does the   **ERROR: (ENCOPT-7016)  for the old script mean NO postRoute optimization is done? 

     2) As suggested, AAE delay calculator is by default, why does the **ERROR: (ENCOPT-7016): show up for the old script?

     3) How important is it to run   optDesign -postRoute -si? Is it a common practice to run it in EDI13 or newer level?

     4) Can I do  optDesign -postRoute -si option without turning on OCV mode?

     5) If I have to turn on the OCV mode to run optDesign -postRoute -si, what is the correct way to do it? (I tried the example script in user guide, but I got some new errors as shown above.)

     

    Sorry for so many questions, I am not very familiar to this flow. Any comments are highly appreciated.

     

     

     

    • Post Points: 20
  • Sun, Mar 2 2014 9:01 PM

    • ajay01
    • Top 500 Contributor
    • Joined on Tue, Feb 12 2013
    • ahemdabad, Gujarat
    • Posts 32
    • Points 415
    Re: postRoute in EDI13.2 Reply

     Hi ,

    I am also using EDI 13.2 cadence encounter..:

    below thing you need to do:

     setDelayCalMode -engine aae -SIAware false

    optDesign -postRoute

     Please use above setting.

    Answer of your summar:

    1.not done.

    2.13.2 version using gigaopt for that you need to give aae engine.

    3.no dont use it will take care automatic.

    4.yes.

    5.better to do  optDesign -postRoute only..

    if any error then point me.

    Thaks,

    Ajay

    • Post Points: 20
  • Mon, Mar 3 2014 11:10 AM

    • wally1
    • Top 50 Contributor
    • Joined on Mon, Aug 4 2008
    • Bellevue, WA
    • Posts 151
    • Points 2,940
    Re: postRoute in EDI13.2 Reply

    Post-route optimization with GigaOpt optimizes base and SI delay simultaneously but it requires OCV mode or else the SI push-out/pull-in delays will not be correct. To do post route optimization with SI run:

        setDelayCalMode -engine aae -SIAware true

        setAnalysisMode -analysisType onChipVariation -cppr both

        optDesign -postRoute

    Regards,

    Brian

    • Post Points: 20
  • Mon, Mar 3 2014 9:05 PM

    • ajay01
    • Top 500 Contributor
    • Joined on Tue, Feb 12 2013
    • ahemdabad, Gujarat
    • Posts 32
    • Points 415
    Re: postRoute in EDI13.2 Reply

     Hi Brain,

    Thanks for correction.

    In my next run i will also take care and check result.

    But can you explain  SI push-out/pull-in delays in detail???

    Thanks,

    Ajay

    • Post Points: 35
  • Mon, Mar 3 2014 10:01 PM

    • BackerShu
    • Not Ranked
    • Joined on Wed, Dec 23 2009
    • Posts 7
    • Points 140
    Re: postRoute in EDI13.2 Reply

     Thank Ajay and Brian,

     

    I actually tried both ways you guys suggested. The previous error information went away. There are two new error information about the density I think is related to the fillers. I think I can ignore them. Please let me know if I am wrong. Error information is as following:

    [03/03 18:21:09    60s] Starting refinePlace ...
    [03/03 18:21:09    60s] **ERROR: (ENCSP-2002):  Density too high (100.0%), stopping detail placement.
    [03/03 18:21:09    60s] Type 'man ENCSP-2002' for more detail.
    [03/03 18:21:09    60s] [CPU] RefinePlace/total (cpu=0:00:00.0, real=0:00:00.0, mem=1876.1MB) @(0:01:01 - 0:01:01).
    [03/03 18:21:09    60s] Total net length = 4.068e+03 (2.307e+03 1.761e+03) (ext = 3.647e+02)
    [03/03 18:21:09    60s] **ERROR: (ENCSP-9022):  Command 'refinePlace' completed with some error(s).

     

    Another followed question, I actually got the same error information as shown above for both commands you guys suggested. So my question is, how can I differentiate which command is used? or in other words, is there any place in the result I can find the difference by using either commands?

     

    • Post Points: 35
  • Mon, Mar 3 2014 10:10 PM

    • ajay01
    • Top 500 Contributor
    • Joined on Tue, Feb 12 2013
    • ahemdabad, Gujarat
    • Posts 32
    • Points 415
    Re: postRoute in EDI13.2 Reply

     Hi,

    Its look like you are running(may be your internal script) refinePlace command.

    Before running refineplace command we need to delete filler.

    After refineplace you can add filler.

    Its normal error take care above thing it will get resolve.

    Let me know if any other error.

    And please update which result is better with siaware true or false?

    Thanks,

    Ajay.

    • Post Points: 5
  • Wed, Mar 5 2014 9:18 AM

    • wally1
    • Top 50 Contributor
    • Joined on Mon, Aug 4 2008
    • Bellevue, WA
    • Posts 151
    • Points 2,940
    Re: postRoute in EDI13.2 Reply

    push-out is added delay due to crosstalk (i.e. victim and aggressor switching in opposite directions). pull-in is decreased delay due to crosstalk (i.e. victim and aggressor switching in same direction). OCV is required so the proper libraries are used to calculate these delays.

    Brian

    • Post Points: 20
  • Wed, Mar 5 2014 9:19 AM

    • wally1
    • Top 50 Contributor
    • Joined on Mon, Aug 4 2008
    • Bellevue, WA
    • Posts 151
    • Points 2,940
    Re: postRoute in EDI13.2 Reply

    Make sure the filler cells are defined properly with setFillerMode so optDesign can remove them and add them back in during optimization.

    Brian

    • Post Points: 20
  • Wed, Mar 5 2014 3:14 PM

    • BackerShu
    • Not Ranked
    • Joined on Wed, Dec 23 2009
    • Posts 7
    • Points 140
    Re: postRoute in EDI13.2 Reply

    Thank you Guys,

    I just want to check if I understand the ERRORs correctly.

    the refinePlace command is internally called by this command:
    optDesign -postRoute -drv

     

    To Ajay:

    Based on your description, I guess you mean we should do addFiller command both before and after the refinePlace command (in this case should be the command:optDesign -postRoute -drv ). Am I right?

    To Brian:

     For proper setting for setFillerMode, do you mean this command:

    setFillerMode -honorPrerouteAsObs true

    If set properly, optDesign -postRoute -drv  can take care of the Filler and we don't need to do addFiller again. Am I right?

     

    Some further questions:

    1) Actually, I checked the log file in previous Encounter vesion. The command  optDesign -postRoute -drv doesn't call refinePlace command and there is no ERROR, is this also new in EDI13.2?

    2) Can I just move the addFiller -cell {FILL128A10TL FILL64A10TL FILL32A10TL FILL16A10TL FILLCAP8A10TL FILL4A10TL FILL2A10TL FILL1A10TL} -prefix FILL -merge true

    after the command  optDesign -postRoute -drv to avoid this ERROR?

    3) I tried and the ERRORs did go away. But, is there any potential problem related to this?

     

    Thank you.

     --Guanghua

    -------------------------------part of script------------------------------------------------

    # Run in-place optimization
    optDesign -postCTS

    # Fix all remaining violations
    optDesign -postCTS -drv

    reset_path_group -name CLOCK
    #clearClockDomains
    #setClockDomains -fromType register -toType register
    optDesign -postCTS -hold
    #clearClockDomains

    # Add filler cells
    setFillerMode -honorPrerouteAsObs true
    #addFiller -cell {FILL1 FILL2 FILL4 FILL8 FILL16 FILL32 FILL64 FILLCAP3 FILLCAP4 FILLCAP8 FILLCAP16 FILLCAP32 FILLCAP64} -prefix FILL -merge true
    addFiller -cell {FILL128A10TL FILL64A10TL FILL32A10TL FILL16A10TL FILLCAP8A10TL FILL4A10TL FILL2A10TL FILL1A10TL} -prefix FILL -merge true

    # Run global routing
    routeDesign

    setDelayCalMode -engine aae -SIAware true
    setAnalysisMode -analysisType onChipVariation -cppr both
    # Final opt
    #optDesign -postRoute -si
    optDesign -postRoute

    #clearClockDomains
    #   setClockDomains -fromType register -toType register
    #   setOptMode -considerNonActivePathGroup true
    #   optDesign -postroute -hold
    #   clearClockDomains
    #   setOptMode -considerNonActivePathGroup false

    #optDesign -postRoute -hold

    optDesign -postRoute -drv

    # Export DEF
    defOut -routing -floorplan final.def

    ## Output GDSII
    streamOut final.gds2 -mapFile tsmc090.map -libName DesignLib -merge $env(GDS_LIB) -stripes 1 -mode ALL

     

    • Post Points: 35
  • Wed, Mar 5 2014 4:18 PM

    • wally1
    • Top 50 Contributor
    • Joined on Mon, Aug 4 2008
    • Bellevue, WA
    • Posts 151
    • Points 2,940
    Re: postRoute in EDI13.2 Reply

    Use the following flow for filler insertion. Otherwise, post route opt cannot remove and add back in filler cells:

    setFillerMode -core {FILL128A10TL FILL64A10TL FILL32A10TL FILL16A10TL FILLCAP8A10TL FILL4A10TL FILL2A10TL FILL1A10TL} -merge true -corePrefix FILL
    addFiller

    Brian

    • Post Points: 20
  • Wed, Mar 5 2014 4:24 PM

    • BackerShu
    • Not Ranked
    • Joined on Wed, Dec 23 2009
    • Posts 7
    • Points 140
    Re: postRoute in EDI13.2 Reply

    Thank you Brian,

    I will try your suggestion.  Do you mean specify no option at addFiller?

     

    In between, I actually tried  to move the addFiller -cell {FILL128A10TL FILL64A10TL FILL32A10TL FILL16A10TL FILLCAP8A10TL FILL4A10TL FILL2A10TL FILL1A10TL} -prefix FILL -merge true

    after the command  optDesign -postRoute -drv to avoid this ERROR.

    and  the ERRORs did go away. But, is there any potential problem related to this?

     

    --Guanghua

    • Post Points: 5
  • Wed, Mar 5 2014 9:18 PM

    • ajay01
    • Top 500 Contributor
    • Joined on Tue, Feb 12 2013
    • ahemdabad, Gujarat
    • Posts 32
    • Points 415
    Re: postRoute in EDI13.2 Reply

     Hi Brian,

    Thanks for clear explanation  about push in and push-out delay.

    Thanks,

    Ajay

    • Post Points: 5
  • Wed, Mar 5 2014 9:59 PM

    • ajay01
    • Top 500 Contributor
    • Joined on Tue, Feb 12 2013
    • ahemdabad, Gujarat
    • Posts 32
    • Points 415
    Re: postRoute in EDI13.2 Reply

     Hi Guanghua,

    1.If you are getting error while running optdesign command then you need to do below setting.

    setFillerMode -core $filler_list -corePrefix FILLER -fitGap true  -deleteFixed true -ecoMode true 

    addFiller -cell $filler_list -doDRC true

    checkFiller 

    2.If you are getting error after optdesign while running refineplace command.Then you need to take care below thing.

    First do checkplace if no overlapping cell or unplaced cell then you can ignore error while running refineplace.

    But if overlapping or unplaced cell there then you need to execute below comman:

    deleteFiller  -prefix FILLER

    refinePlace 

      addFiller -cell $filler_list -doDRC true

     checkFiller 

     checkPlace

    Thanks,

    Ajay

     

     

     

     

    • Post Points: 5
Page 1 of 1 (13 items)
Sort Posts:
Started by BackerShu at 01 Mar 2014 11:21 AM. Topic has 12 replies.