Home > Community > Forums > Digital Implementation > About timing and slack

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: *

 About timing and slack 

Last post Sat, Jul 6 2013 11:35 AM by kasyab. 2 replies.
Started by kasyab 30 Jun 2013 09:56 AM. Topic has 2 replies and 632 views
Page 1 of 1 (3 items)
Sort Posts:
  • Sun, Jun 30 2013 9:56 AM

    • kasyab
    • Top 500 Contributor
    • Joined on Sat, Mar 13 2010
    • Posts 22
    • Points 305
    About timing and slack Reply

     Hi,

      I am using EDI 10.12.002 to implement a processor design. I have my own library of logic cells which I mix with certain sequential cells, buffers and inverters of a vendor library. The geometric aspects and characterization conditions for the custom library cells match that of vendor librar but there are some differences in that the vendor timing libs are produced using synopsys tools whereas I obtain my libs from ELC.

      During the routing phase I get this warning:

     #WARNING (NRDR-128) Can not do via swapping using command 'globalDetailRoute', use 'detailRoute' command instead . Will do normal detail routing if possible.
    #Using multithreading with 4 threads.
    #Worst slack with path group effect 340282346638528859811704183484516925440.000000

     What does this indicate?

    Additionally, the optimization phase of routing goes on for a really long time and runs for a couple of hours(It has been over two hours now and still going strong). The timing constraint is a very relaxed 30ns and up until the routing phase I get a slack of 18+ ns. I intentionally have a relaxed timing constraint as the custom library cells have small and limited drive strength. I have a few essentaildelay constraints on the input and out put portsof the design but nothing else.

    Could the long run times indicate missing false path constraints and/or multi-cycle path constraints? Is there a way to ascertain this and get more information? I suspect that the long runtimes are due to missing/erroneous constraints because I also have a generous floorplan and the design starts out with about 20% density but is buffered up to about 75% density (the slack also falls to only about 0.18ns). What could be the possible reasons for such large discrepancies in the reported slack and is there any way to ensure that false/multi-cycle paths are accurately accounted for?

     

    Any help is appreciated.

    Thank you,

    Kasyab

     

    • Post Points: 20
  • Fri, Jul 5 2013 7:29 AM

    • VKhlyupin
    • Not Ranked
    • Joined on Fri, Jul 5 2013
    • Posts 9
    • Points 120
    Re: About timing and slack Reply

     Routing in EDI consist fom to big parts - global routing and detail routing. Looks like this warning informs you that one part could not do via swapping and recomend you to use another part for this. Rasons of this behaviour coul be in you setup try to find somethig related in descriptions of folloeing encounter commands:

    "setNanoRouteMode" "setOptMode"

     

    Long run issue could have different roots. Short library is in the list.

    To check your concerns try to use "checkDesign" and "checkTimingLibrary" comands.

    • Post Points: 20
  • Sat, Jul 6 2013 11:35 AM

    • kasyab
    • Top 500 Contributor
    • Joined on Sat, Mar 13 2010
    • Posts 22
    • Points 305
    Re: About timing and slack Reply

     Hi VKhlyupin,

     Thanks for your response. The root cause of this issue lay in incorrectly set timing constraints during synthesis. Once I corrected those, the resulting SDC from RTL compiler passed the correct path constraints to Encounter. The message about the  worst slack also disappeared.

    Your points are well taken however.

     

    Thanks,

    Kasyab

    • Post Points: 5
Page 1 of 1 (3 items)
Sort Posts:
Started by kasyab at 30 Jun 2013 09:56 AM. Topic has 2 replies.