GDSFILT Revision History

v2.49 04/23/07

corrected some problems handling large files.

corrected a flow problem (Unix/Linux) that occured when exploding files.

v2.30 05/25/05

GDSFILT had a rounding error of 1nm when exploding arrays. This version fixes the problem.

v2.23 09/02/04

This version is fixes a crash problem when the PATH (environmental variable) was longer than 1024 characters.

This version has a new -remove command line option which allows the user to remove specific cell names from the file.

v2.21a 09/07/04

Revision to GDSExplode to support data type

This version is equivalent to previous releases of GDSFILT with the exception that the GDSXPLODE module now preserves DATATYPES during explosion. Previously all DATATYPES were mapped to "0" in the output.

v2.20 08/02/04

Major revisions to the GDSFILT code.

Major revisions to GDSFILT were made in response to customer feedback. In particular, users are processing very large GDSII files through gdsfilt requiring in some cases 64 bit hardware. Speed has become a very significant issue. Also, many users are building their final GDSII tape out file using the blend mode -- this mode ran very slow as it had never been tested with very large files.

The GSCAN (GDSII scanning module) has been speeded up and a memory allocation problem seen in earlier versions has been corrected. (Reported by Applied Materials)

In response to customer requests the blend mode has been greatly optimized and speeded up. For example, blending of a 4GB (approx. 1500 structures) and a 2GB (approx. 1000 structures) file on a 1.7GHz CPU now takes just 10 minutes where previously this required several hours. (reported by ATI and by 3DLabs)

The shared libraries issue has been resolved. (reported by Intel)

v2.05 04/02/04

This version includes the latest gscan which is parallel safe - you can run it on different files and at the same time on the same workstation.

This has ben fixed.

gdsfilt would crash when running on Linux with Flexlm. This has been fixed. Current version was tested successfully on Red Hat 7.2, 9.0, and Enterprise WS 3.

v1.99 11/06/03

This version is now now synched on all platforms with Flexlm. Several Motif issues were fixed in the shell. Added HPUX 64 bit support.

v1.79 Linux7.2 and Solaris 09-26-2002

This version supports the Globetrotter FLEXlm license manager. Requires that you run lmgrd 8.1 or newer.

v1.71 07-3-2002

v1.63 01-28-2002

This version has a floating point diagnostic tool. We've added a -fpchk:N
command line argument. See what 

gdsfilt -h

returns. The new option emits information regarding FP data encountered
in HEADER, SREF, AREF and TEXT entities. These entities contain FP data
per the following table:

    GDSII ENTITY      FP VALUE(S)               OPTIONAL      DEFAULT
    
    HEADER            user unit
                      database unit
                      
    SREF, AREF        insertion MAGnitude       YES            1.0
                      ANGLE                     YES            0.0
    
    TEXT              MAGnitude                 YES            1.0
                      ANGLE                     YES            0.0
                      
                      
The :N in -fpchk is optional. If specified gdsfilt checks for FP data
associated only with the desired entities by ORing the following values

   1   HEADER information
   2   SREF, AREF data
   4   TEXT data
 
   Option Value    Info Output
   
   -fpchk:1        HEADER 
   -fpchk:2        REF
   -fpchk:3        HEADER and REF 
   -fpchk:4        TEXT
   -fpchk:5        HEADER and TEXT 
   -fpchk:6        REF and TEXT 
   -fpchk:7        HEADER and REF and TEXT

NOTE: -fpchk and -fpchk:7 are equivalent

-fpchk option will tabulate all unique FP values, and output counts and
the exact binary image for each, as well as flag an un-normalized numbers.
The information collected regarding FP numbers is NOT reset when a new
file is input, i.e. as the result of a -merge or -add. 

A sample output follows:


FP values for HEADER and REF and TEXT data:

   Count Value

       1 0.01
       1 2.54e-07
       4 0.99999995886464
       1 1.5950381620042e-85
       1 0.99999995886464
       1 1.5950381632027e-85

MAG values

   Value Image

    0.01 00111111.00101000.11110101.11000010.10001111.01011100.00101000.11110110
2.54e-07 00111011.01000100.00101110.10111111.01000001.01011010.00100011.10100011
       1 01000000.11111111.11111111.11111111.01001111.01010011.00110000.00110001
1.54e-85 00000000.00000000.00000000.00000000.01001111.01010011.00110000.00110001
                  ^^^^
                  |
                  |___ un-normalized floating point number

       1 01000000.11111111.11111111.11111111.01001111.01010011.00110000.00110010
1.54e-85 00000000.00000000.00000000.00000000.01001111.01010011.00110000.00110010
                  ^^^^
                  |
                  |___ un-normalized floating point number


Finally, the -fpchk command line option can be used in conjunction with
ANY gdsfilt operation, but for purposes of gleaning FP information you
might want to run gdsfilt in the most simplistic way and specify that
the output file be "/dev/null", i.e.

gdsfilt TEST.gds /dev/null = -fpchk:7

You will want to refer to TEST.log.

v1.61 08-08-2001

Drop Boundary Segments Shorter Than
New feature to drop boundary segments shorter than a user specified value.

Output Text Only
New feature to explode all text only into the output file.

Polygon Filtering No Text
In previous versions, all text are dropped if polygon filtering is turned on. The feature to keep text under polygon filtering is added in this version.

New HostID Locking
This version applies the new HostID locking scheme without the need for special system network setup.

Translation Progress
Better progress reporting. The polygon filtering process has 3 passes:
1. Scan through all the layers. (You will see a percentage progress.)
2. Extents and polygons computation. (You will just see a moving progress bar, no percentage. Under Windows 98, using hostid locking, you will not see any progress at all.)
3. Generating GDS Output. (You will see a percentage progress.)

Default Output File Name
Default GDS output file name is "inputfile_out.gds" in the working directory.

Maximum Number of Structures
New limit is raised from 8192 to 32768. Older versions would just crash when the input file has more than 8192 structures.

Combine Function
Added combine function for merge and add for users to specify a new top structure in the output which refers to the top structures of the files to add or merge.

Trimming Speed Up
Trimming is speeded up up to 10x +.

Specifying a New Top Structure
Added -combine option to the engine, gdsfilt, which allows users to combine (merge and add) structures with a new top structure referring to the top structures in the input files. Refer to gdsfilt -h for more information.

Licensing
Updated the licensing to support the new 7x7 ACS.KEY file codes.

Major GUI Revision
Major facelift with all other motif standard dialogs added to xgdsfilt.

XGDSVU Viewer Linkage
This version now ships with Artwork's GDSII viewer, Xgdsvu. Xgdsvu and Xgdsfilt are keyed separately during the installation, so xgdsvu requires it's own license.

Added Layer and Datatype Mapping
Layer numbers and datatype numbers can be mapped to different values in the output file using the interface provided. Requested by multiple customers who wanted to do filtering by datatype.

Merge/Add/Blend Two GDSII Files
Xgdsfilt provides a new feature to merge/add/blend 2 or more GDS files together. A user interface is available for selecting the way the two files are combined.

Text Translation
Added text control to omit text greater than a certain size.

Windows 95/NT Release

Bug Fix: Similar Structure Names