Remove Cells (Structures)

A user may want to remove one or more cells from a GDSII file. The command line option, -remove filename, deletes cell definitions listed in the file. If the number of cells to remove is large and they all share some common string, one can also use a regular expression in place of the list.

Command Line Syntax

gdsfilt input_file output_file structure -remove filename


 input_file          gdsii input file
 output_file         gdsii file to create
 structure           input file structure to process
 -remove filename    remove structures with names matching
                     those in the text file.

Remove File Contents

The rename file has two entries per line. The first entry must match (exactly - case sensitive) the name of a structure in the input file. The second entry will be the new name in the output file.


If for some reason (since it is really not supposed to happen) you have structure names with spaces in them then you must use quotes to delimit input and output names.


Unique Names

There must be unique relationship between input and output names. Two input names must not be mapped to a single output name (even inadvertently). Doing so will completely garble the resulting output file.


As an example we will use our very small file demo1.gds. Below you can see a screen shot of demo 1 along with the structure list (using Qckvu).

sample chip demo1.gds

Now we will produce three different output files using the following command line and rename files:

This first example shows how to handle spaces in the structure names.

gdsfilt demo1.gds demo1_ex1.gds = -rename example1.lst
  where example1.lst contains:

  "TOPMOSTST"top most cell"
  "PAD4MILS"pad four mils"
  "TQE"tee queue eee"
  "PADCONT"pad contact"
  "SCNT"second contact"
  "PLUG2"plug two"
  "QE"queue eee"
new structure names as displayed by Qckvu

The second example changes the names from upper case to lower case.

gdsfilt demo1.gds demo1_ex2.gds = -rename example2.lst
  where example2.lst contains:

  TOPMOSTST topmostst
  PAD4MILS pad4mils
  TQE tqe
  PADCONT padcont
  SCNT scnt
  PLUG2 plug2
  QE qe

new structure names as displayed by Qckvu

The third example only changes some of the names.

gdsfilt demo1.gds demo1_ex3.gds = -rename example3.lst
  where example3.lst contains:

  PAD4MILS pad4mils
  PADCONT padcont
  PLUG2 plug2

new structure names as displayed by Qckvu

Using a Regular Expression to Remove Structures

We recently wanted to remove dummy metal from a large GDSII file as it slowed down the processing of the file (for net tracing) and we knew that that the dummy metal was not part of any net we wanted to trace.

Unlike some designs, where a separate data type is used to delineate dummy metal fill, in this design there were numerous structures that had been inserted at many different levels of the hierarchy that contained dummy metal fill on the same layer as metal we wanted to analyze. So we could not use layer filtering.

metal layer with dummy fill

Dummy metal file we want to remove

We examined the GDSII file and noticed that all of the structures that held the dummy fill had the string: _DMx in them.

1) DM7_L_DMx
3) RF1SHD_64X28CK2_DMx
5) RF1SHD_64X28M24BOT_DMx
6) RF1SHD_64X28_DMx
7) RF2SH_128X20CTRL_SA_R_DMx
8) RF2SH_128X20_DMx
10) RF2SH_32X22_DMx
12) RF2SH_256X19CTRL_SA_R_DMx
13) RF2SH_256X19HXP24_DMx
14) RF2SH_256X19HXP24R_DMx
15) RF2SH_256X19HYP24_DMx
16) DM2_S_DMx
17) DM1_S_DMx
18) RF2SH_256X19_DMx
20) RF2SH_512X18CTRL_SA_R_DMx
21) RF2SH_512X18_DMx
22) RA1SHDbit_1024X32HBFX_DMx
2366) RA1SHD_800X28VIANOMX32B
2369) RA1SHD_800X28CDXX
2370) RA1SHD_800X28CDX
2371) RA1SHD_800X28MUX_CD_ODD
2372) RA1SHD_800X28SA8
2373) RA1SHD_800X28VIAIOSG
2374) RA1SHD_800X28IO8

This means that we can use a "regular expression" to match all those structure names and remove them. This would be much easier than actually creating the list.

Regular Expression

The regular expression that would match our dummy fill cells is:

Short Explanation of this Regular Expression

"^"     indicates the beginning the string to be matched

".*"    matches zero or more occurrences of any character
        basically we don't care what happens in the front of
        the structure name string.

"DMx$"  the last three characters of the string must match DMx.
        this does our filtering work for us.

Running GDSFILT with the regular expression

We now run the file, caw_test1.gds through GDSFILT using the -remove command line argument which will remove any cells in the resulting output file that match our regular expression. The output file should be a lot smaller and easier to handle.

$ gdsfilte caw_test1.gds caw_test1b.gds = -remove remove_list.txt


 gdsfilte                       name of the gdsfilt executable engine
  caw_test1.gds                 our input file
   caw_test1b.gds               the output file to create
    =                           use the top structure of the input file
     -remove remove_list.txt    the remove argument referencing the file remove_list.txt
The file remove_list.txt contains a single line:


Here's a snapshot of the same layer/location after filtering:

metal layer with dummy fill cells removed

Example 2 - Removing Vias using Regular Expression

Let's look at another example of removing a large number of cells based on name. Our large file had a lot of vias; most of them arrayed. Assuming we want to remove these without affect the rest of the layout how can we do it?

An inspection of the structure names quickly showed that these via arrays all started with

. So a regular expression can be used to identify and remove them.

The expression is:



   ^     beginning of string
   \$    match the $ character (escaped with \ because $ is a control character)
   \$    match the $ character (escaped with \ because $ is a control character)
   via   match the character string via
   .*    any number of characters thereafter
   $     end of the sting

The same command line can be used:

$ gdsfilte caw_test1.gds caw_test1b.gds = -remove remove_list.txt

and remove_list.txt contains one line:


Can I use more than one regular expression?

Yes, if you wanted to remove all the dummy metal based on the ending string DMx and all the via arrays based on the starting string $$via you could have two lines in your remove_list.txt file: