GDSview User’s Manual

Version 4.5
Integrand Software, Inc.
Copyright c 2008–2014

1 Introduction
GDSview is a layout viewer for GDSII files. It is intended for use with the EMX∗
electromagnetic simulator. GDSview reads an EMX process file in addition to
the GDSII layout, and it processes the layout in exactly the same way that
EMX does. Thus, it can be used to visualize the effect of the various layout
manipulation and simplification operators in the process file.
You invoke GDSview like this:
gdsview layout.gds mystructure foundry.proc
This command tells GDSview to display the structure mystructure within the
GDSII file layout.gds. The layout is processed according to the process file
foundry.proc. If you omit the process file, GDSview will display the raw layout.
There are two versions of GDSview. One (the gdsview executable) has a
basic X Window System interface, and the other (the gdsview-ogl executable)
has a more fancy OpenGL† interface. Two executables are provided in case your
computer lacks the necessary shared libraries for the OpenGL version. Normally
when you run gdsview, it will automatically try to pass control to the OpenGL
executable; if this fails, then a warning is printed and the basic X Window
System interface is used instead.
GDSview also accepts some command-line options. The following options
have the same meaning as they do for EMX:
--label-depth=number By default, GDSview only displays labels in the top
level GDSII structure. If you have labels further down in the hierarchy,
you can specify how deep GDSview should look using this option. The
default, 0, means only look at the top level. Specifying --label-depth=1
would mean to look in sub-structures of the top level structure too. You
can abbreviate this option as -l.
--unique-labels Adds a numeric suffix to all labels so as to ensure that all
labels are distinct. For example, multiple occurrences of label a become
∗ EMX is a registered trademark of Integrand Software, Inc.
† OpenGL is a registered trademark of SGI.

1
 a_1, a_2, etc. This is mostly useful when you have multiple instances of a
single cell with some labels in it and you want all of the labels to remain
distinct. Using this option with GDSview lets you tell where the unique
labels are.
--cadence-pins=number When a GDSII file is streamed out from Cadence,
there is an option to write special property records in the file that encode
pin information. This option tells GDSview to interpret the pin infor-
mation and add labels to the layout at the pin locations. See the EMX
manual for more information.
--port=port specifier This is the same as the --port option to EMX, and it
tells GDSview to display only the labels relevant to the given ports. You
may abbreviate --port as -p.
--signal-ground-separator=character This sets the separator for the signal
and ground parts of a port to the specified character, as in EMX.
--internal=specifier This option is like the same option in EMX; it specifies
the size of the internal connection point for a label. You can abbreviate
this as -I.
--via-separation=number To be consistent with EMX, by default GDSview
merges vias that are closer than a certain distance to each other. The
default is distance 0.5 (microns). If you don’t want to merge vias, specify
0. This option may be abbreviated by -v.
--local-reflex-scaling=number This option affects the merging of diagonal
vias. For more information, see the documentation on this option in the
EMX manual.
--width=number When the process file contains width- and spacing-depen-
dent quantities, the calculation of their values can be somewhat time-
consuming. This option makes GDSview assume that the local geometry
width is equal to the specified number of microns (instead of calculating
it). If you specify both --width and --spacing, then the time-consuming
calculations are avoided.
--spacing=number Assume that the local geometry spacing is equal to the
specified number of microns.
--no-geometry-dependence Turn off local width and spacing computation.
--scaling=number Scale the GDSII layout by the specified amount, i.e., do
a lateral shrink or expansion of the layout. Scaling is applied when the
GDSII file is read. Operations in the process file (such as grow, merge,
etc.) are done after the scaling is applied. You can use -s as a short form
for this option.
--define=name,number This is equivalent to adding a definition of the form:

2
 define name=number

in the process file.

--definitions-file=name This specifies an additional process file which will
be read before GDSview reads the main process file.
--key=string When using an encrypted process file, this option provides the
decryption key.
--case-sensitive By default identifiers in the process file are internally con-
verted to lower case. This option makes GDSview preserve the case of
identifiers instead.
--excluded-cells-file=name Omit any cells in the GDSII file whose name
matches one of the regular expressions in the specified file. The file should
have one regular expression per line.
--included-cells-file=name This is like the --excluded-cells-file op-
tion, but GDSview will only include cells that match (or subcells of those
cells), instead of excluding them.
--device-cells-file=name Any cells in the GDSII file whose name matches
one of the regular expressions in the specified file will be treated as “black
boxes.” GDSview will show internal labels of the form ~x0000, ~x0001,
. . . , for the terminals of black-box devices.
--corner=name Selects a particular process corner. Any random variable in
the process file with a corresponding corner value will be fixed at that
value. Multiple --corner options may be given, but only one may apply
to any given random variable.
--colinear-fuzz=number In EMX, input polygons usually are simplified by
eliminating “nearly colinear” vertices. This option has the same effect as
the corresponding EMX option in controlling this behavior. See the EMX
manual for details.
--internal-units-per-micron=number The geometry in GDSview is repre-
sented by polygons whose vertices lie on a grid. The internal grid spacing
defaults to one nanometer (the same as in most GDSII files). This option
changes the number of grid units per micron. Increase it to make the grid
finer; for example, setting this option to 10000 makes the internal grid
spacing 0.1 nm.
--label-snap-distance=number GDSview by default draws a line between a
label and the edge to which the label will “snap” in EMX. This option
controls the snap distance, as in EMX.
--device-label-snap-distance=number This option controls the snap dis-
tance for labels that come from device terminals.

3
--device-pins-internal Treats all Cadence shape pins within black-boxed
device as if they have access direction TBLR, so the pins will become
internal ports.
--version Print GDSviews’s version number and exit.
--x11 Force GDSview to use only the basic X Window System interface. Use
this if the OpenGL version will not run on your system due to, e.g., missing
shared libraries.
These options are unique to GDSview:
--layers=layer list Tell GDSview to show only the certain layers, or to omit
certain layers. Example layer lists:
me1,me5,me6 Show only me1, me5, and me6.
-me1,me2 Do not show me1 or me2.
--no-id This turns off the display of whatever the mouse is pointing to. This
can be significantly faster for complicated layouts. The abbreviation for
this option is -n.
--no-labels Tells GDSview to turn off the label display. You can abbreviate
this as -N.
--no-nets Turn off the display of nets (OpenGL version only).
--font=name Use the specified font for display of labels and rules. The --font
option can be abbreviated as -f. Available font names for the X Window
System version can be found using the standard xlsfonts executable.
Fonts for the OpenGL version have the form name-size or name, where
name is a font name and size is the height of the font in pixels (defaulting
to 14 if the size is omitted). The valid names can be found using the
--list-fonts option.
--list-fonts Show all the valid font names and exit (OpenGL version only).
--raw-polygons When used without a process file, this option keeps GDSview
from merging polygons. This is mostly useful for debugging, to see exactly
what is in the GDSII file.

2 Controlling GDSview (X version)

The basic X Window System version of GDSview displays a simple window
with the layout, as in figure 1. The line styles in the layout have the following
meaning:
Thin solid line a regular edge

4
Thin dotted line connects an edge label and the edge that it will snap to.
If a label is too far from any edges to snap to them, it will be displayed
between question marks.
Thin short dashed line the boundary of named resistor and capacitor re-
gions in the process file
Thick dashed line the region for an internal label
Thin long dashed line the boundary for a black-boxed device

Figure 1: GDSview (X Window System version) layout window

You use the keyboard and mouse to control what is displayed. If you hold
down the first mouse button (typically the left one), sweep out a rectangular
area, and then release the button, GDSview will zoom in on the selected region
(figure 2). If you hold down the second mouse button (usually the right one)
and sweep, GDSview displays a rectangle with length and width information,
but does not zoom in when you release the button.

5
 ⇒

Figure 2: Zooming in with the left mouse button

6
Typing z makes GDSview zoom in by a factor of 2 (figure 3).

Figure 3: Zooming in with the z key

Typing x makes GDSview zoom out by a factor of 2 (figure 4).

Figure 4: Zooming out with the x key

7
 The c key makes GDSview recenter and rescale the layout so that everything
is visible (figure 5).

Figure 5: Recentering with the c key

The arrow keys (up, down, left, and right) can be used to scroll around the
layout. The viewport moves in the direction of the arrow key (figure 6).

Figure 6: Scrolling with the left arrow key

8
 If you move the mouse pointer near an edge, GDSview displays the name of
the layer. When a layer name is displayed, typing h hides that layer (figure 7).

Figure 7: Hiding a layer with the h key

Type v to make all layers visible again (figure 8).

Figure 8: Making layers visible with the v key

9
 When a layer name is displayed, click the third mouse button (usually the
middle button on a three-button mouse, or a simultaneous click on a two-button
mouse) to print an --add-label option to standard output (figure 9). If you
copy the option and use it on the EMX command line, EMX will operate as if
the GDSII file has a label on the specified layer. The label defaults to p1 for the
first click, p2 for the second click, etc., but you can edit the label name. This
method is useful for adding edge ports to a layout without having to modify
the GDSII file. If you type the i key when a layer name is displayed, and then
move the mouse and click the third mouse button, GDSview will print an EMX
option to add an internal port at that position.

⇒ --add-label=m3,5.7746,-27.5508,p1

Figure 9: Adding labels with the mouse

Typing p produces an encapsulated PostScript picture of the current win-

dow (this is not available in the OpenGL version). The output file is called
layout.eps.
Type q or the escape key to quit GDSview.

10
3 Summary of GDSview controls (X version)
First (left) mouse button Sweep an area to zoom in
Second (right) mouse button Measure an area
z key Zoom in by a factor of 2
x key Zoom out by a factor of 2
c key Recenter the layout
Arrow keys Scroll around the layout
h key Hide a layer
v key Make all layers visible
Third (middle) mouse button Print an EMX option to add a label
p key Write encapsulated PostScript file
q key Quit GDSview

4 Controlling GDSview (OpenGL version)

The window for the OpenGL version of GDSview looks like figure 10, with a
main layout display and a sidebar. The sidebar has control buttons and shows
layer names. If you’re using GDSview with a process file, then the sidebar also
shows net names.
The basic key and mouse controls within the layout display are the same as
those mentioned earlier (and summarized in the previous section), plus there
are a few other available keys:
w key Go back to the previous view (useful if you want to zoom in closely for
a moment to check something)
r key Raise the pointed-to layer to the top in the display order
l key Lower the pointed-to layer to the bottom in the display order
0, 1, . . . , 9 keys Set the opacity of the pointed-to layer to 0%, 10%, etc.
f key Set the opacity of the pointed-to layer to 100%.
s key Makes the opacity of the pointed-to layer be controlled by the fill slider.
d key Reset the display order to the default (layers are drawn in the order that
they appear in the process file, typically from substrate upward)
t key Toggle layer visibility for all layers

11
Figure 10: GDSview (OpenGL version) layout window

12
m key Mark the pointed-to net (unmarked nets are shown dimly). You can
mark multiple nets.
u key Unmark the pointed-to net. If the mouse is not pointing within any
layers, then all nets are unmarked.
n key Hide the pointed-to net.
In the sidebar there are buttons that can be used for the navigation func-
tions discussed earlier. Buttons labeled with a letter in brackets have the same
effect as pressing the corresponding key in the layout display. For example,
the Re[c]enter button has the same effect as pressing the c key in the layout
display. The following buttons have no key equivalents:
All hidden Make all layers invisible
Reversed order Make the display order the reverse of the order in the process
file
Uniform fill Use the same opacity value for filling all the layers (the value is
set by the fill slider)
All nets visible Make all nets visible. (This is independent of layer visibility,
so if the layers for a net are all hidden, you won’t see the net.)
All nets hidden Make all nets invisible. Things that aren’t part of any net
(like device boundaries and labels that are not close to any edge) will still
be visible.
The Fill slider adjusts the opacity of the fill for layer interiors. Move the
slider all the way to the left for full transparency and all the way to the right
for full opacity. If you’ve marked some nets, then their opacity is what the
fill slider controls. The opacity for the remaining unmarked nets is set by the
Unmarked fill slider.
The buttons showing layer names can be left-clicked to toggle the display of
individual layers on and off. Right-clicking on a layer name brings up a menu
where the layer can be raised or lowered in the display order, and where the
opacity of the layer can be set (independent of the fill slider value). Keyboard
shortcuts for the menu items are available and are shown in brackets in the
item labels. The shortcut is available when the button with the layer name has
keyboard focus (typically because you’re pointing at it with the mouse).
Similarly, the buttons for net names can be used to control net visibility,
and right-clicking will show a menu with net operations. All the nets that don’t
have any ports (or device terminals) are grouped together and controlled by the
button labeled unnamed nets.
Also note that as you move the mouse pointer around the layout window, the
names for any layers that are under the pointer are highlighted in red. Actions
like hiding (h), raising (r), setting layer opacity (0, 1, . . . ), etc., normally act
on the layer whose name is displayed (because the mouse is pointing at an edge

13
of that layer). However, if no name is displayed (i.e., the pointer is not close to
any layer edge), then the action is instead applied to the layer that the pointer
is within. If the pointer is within multiple layers, then the one that is topmost
in the display order is the one that is affected. (If no name is displayed and the
pointer is not within any layer, then the action has no effect.) Nets that the
pointer is within are also shown in red. Keystrokes normally operate on the net
whose name is displayed. However, if no name is displayed but the pointer is
within a unique net, then that net is the one affected.

5 Acknowledgements
Robust geometric predicates are by Jonathan R. Shewchuk.

Routines for Arbitrary Precision Floating-point Arithmetic and Fast

Robust Geometric Predicates
(predicates.c)
May 18, 1996
Placed in the public domain by
Jonathan Richard Shewchuk
School of Computer Science
Carnegie Mellon University
5000 Forbes Avenue
Pittsburgh, Pennsylvania 15213-3891
[email protected]

Regular expression support is provided by the PCRE library package, which

is open source software, written by Philip Hazel, and copyright by the University
of Cambridge, England. Source code for the PCRE library may be obtained
from:
ftp://ftp.csx.cam.ac.uk/pub/software/programming/pcre/
The PCRE library is available under a BSD license:
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions
are met:
1. Redistributions of source code must retain the above copyright
notice, this list of conditions and the following disclaimer.
2. Redistributions in binary form must reproduce the above copy-
right notice, this list of conditions and the following disclaimer
in the documentation and/or other materials provided with the
distribution.

14
 3. Neither the name of the University of Cambridge nor the name
of Google Inc. nor the names of their contributors may be
used to endorse or promote products derived from this software
without specific prior written permission.
This software is provided by the copyright holders and contributors
“as is” and any express or implied warranties, including, but not
limited to, the implied warranties of merchantability and fitness for
a particular purpose are disclaimed. In no event shall the copy-
right owner or contributors be liable for any direct, indirect, inci-
dental, special, exemplary, or consequential damages (including, but
not limited to, procurement of substitute goods or services; loss of
use, data, or profits; or business interruption) however caused and
on any theory of liability, whether in contract, strict liability, or tort
(including negligence or otherwise) arising in any way out of the use
of this software, even if advised of the possibility of such damage.
The user interface for the OpenGL version of GDSview is based in part on
the work of the FLTK project (http://www.fltk.org).

15