Digital Implementation - All Comments

Thanks Bob, with ...

Anonymous — Thu, 18 Mar 2010 04:10:48 GMT

Thanks Bob,

with Best Regards

Shivaraj

The default behavior ...

BobD — Thu, 18 Mar 2010 03:02:20 GMT

The default behavior is to snap modules to rows vertically. This preference can be changed via File->Preferences->Floorplan. Change the "Snap Guides/Regions/Fences to:" from Std Row/M2 pitch to Manufacture Grid, User-defined Grid, or Placement Grid.

Bob, Is it possible ...

Anonymous — Wed, 17 Mar 2010 17:35:39 GMT

Bob,

Is it possible to run cmds setObjFPlanBox/Ploygon or definePartition -corespacing XY coordinates with noninteger multiples of row height..When i ran these cmds with nonint multiples,They are rounded off to int multiples.

Bob Thanks alot for ...

Anonymous — Wed, 17 Mar 2010 05:10:23 GMT

Bob Thanks alot for the comments...

Yes, it is possible ...

BobD — Tue, 16 Mar 2010 04:20:10 GMT

Yes, it is possible to specify a module that has no hard macros within as a partition. In fact, you can specify an empty module as a partition and I see designers effectively leveraging this approach quite frequently.

A tip in this area: By default the tool doesn't display modules with less than 100 instances. Try setting this threshold to 0 by going to Tools->Preferences->Display->Min Floorplan Module Size to make empty modules visible.

Hi,Is this possible ...

Anonymous — Sun, 14 Mar 2010 21:20:19 GMT

Hi,Is this possible to specify a module as partition,which contains only standard cells inside it(i,e there are no hard macros)

Awesome tips, Jason. ...

BobD — Fri, 26 Feb 2010 05:22:10 GMT

Awesome tips, Jason. Very useful.

Your use model of redirect will be useful in sharing with R&D how the more advanced options are used in practice. I'll pass this along to them as it will help improve the documentation when redirect becomes officially supported.

Bob, I've found r ...

jgentry — Fri, 26 Feb 2010 04:38:00 GMT

Bob,

I've found redirect to be very useful when you don't want your logfile to be dominated by, what's in my opinion, useless output. For instance:

deleteInst *INSTPATTERN*

...will give you a "Deleting instance [INSTNAME]" messages for each and every instance that is being deleted. In some cases, we have thousands of these messages in our logfiles. To get rid of this:

redirect -quiet -nolog {

deleteInst *INSTPATTERN*

} >& /dev/null

The several thousand messages gets reduced to none. Be careful though, if anything when wrong during the commands being executed in the redirect body, you won't know about it.

Regarding reading/writing files, it is pretty simple to add compression to the mix. For instance:

### To read a file that might be gzipped:

if {[file extension $fileName] == ".gz"} {

set inFile [list | gzip -dc $fileName]

} else {

set inFile $fileName

}

if {[catch {open $inFile r} IN]} {

... do any cleanup necessary, then error gracefully.

error $IN

}

### To gzip an output file (make sure $fileName has .gz at the end):

set outFile [list | gzip > $fileName]

if {[catch {open $outFile w} OUT]} {

... do any cleanup necessary, then error gracefully.

error $OUT

}

Don't forget to close your file handles using the 'close' command.

Enjoy,

I was using subversi ...

Anonymous — Thu, 25 Feb 2010 19:08:22 GMT

I was using subversion for version control of the flow scripts, and the input data, like LEF and Verilog.

I was not using Subversion for binary databases. I believe I have seen Subversion used with CAD databases, but the user was not so satisfied and was looking for a different solution.

hi John, Could yo ...

keithd — Thu, 25 Feb 2010 04:24:46 GMT

hi John,

Could you elaborate on the use of subversion? Were you using it for revision control of your design data?

thanks

-keith

User Review of The Encounter Foundation Flow

Digital Implementation — Tue, 23 Feb 2010 04:56:11 GMT

This is a guest post from John McGehee. John is an independent consultant in Silicon Valley, specializing

Thanks for the comme ...

BobD — Fri, 12 Feb 2010 02:40:19 GMT

Thanks for the comment Stéphane!

I think I know what you're talking about and agree it can be cumbersome to deal with. If I follow correctly, you're referring to when you "arrow-up" to retrieve a previously executed command and if that command is longer than the width of the Encounter console the wrapping of that line is malformed. You can't get to the portion of the command after the line break to alter it.

A workaround I've used in this area is to use the "ctrl-a"/"ctrl-e" to navigate the cursor to the beginning/end of the line to alter portions of the command that are inaccessable due to this issue.

I'll bring this to R&D's attention to see if there's something we can do to improve the situation. Great suggestion! It's exactly the kind of thing I'm looking for.

Anyone else have suggestions or requests?

-Bob

Dear Bob, Thanks. ...

Anonymous — Wed, 10 Feb 2010 01:35:08 GMT

Dear Bob,

Thanks.

Multi-line commands in SOC: When you type a long report_timing for instance, and you recover a long one from the history and that you want to edit it, the command is sometimes displayed only on one line whereas it should be displayed on two lines. A part of the command is hidden. This is not user friendly.

Best regards.

Stéphane.

Most excellent point ...

Anonymous — Tue, 02 Feb 2010 09:47:29 GMT

Most excellent points. At advanced nodes, implementation and signoff should not be two separate parts of the flow but one leading into the other seamlessly. EDI system goes a step beyond with the ability to complete signoff analysis from within the platform thereby reducing iterative loops to near zero.

Suprio, Good analog ...

PeteMc — Fri, 29 Jan 2010 04:39:45 GMT

Suprio,

Good analogy, I think we are in total agreement here. If checks have been made throughout the flow, then the final check could be to simply make sure that all of the previous checks showed that each stage was successfully signed off.

Obviously such a solution would also need to effectively manage signoff for designs with multiple modes of operation and multiple process corners, where typically today validation within the design flow does not cover all of the necessary combinations that would be used for final signoff. This is where I see statistical signoff analysis providing significant value.

Pete