\newpage \section{Running \NEWOL\ - The menu system} \NEWOL\ originated on the HP 9836C ``Chipmunk'' series of computer with the HP 9111A graphics tablet attached. A version that runs on the ``Bobcat'' machines is also available, and can use either their graphics tablets or the mouse. The \putindex{command-line syntax} of the \NEWOL\ command is: {\tt wol [-u$<$username$>$] [-t$<$tech$>$] [-q] [file1 [file2 ...]]} where \begin{description} \item[{\tt -u$<$username$>$}] \NEWOL's first action is to load a configuration file specifying technology, size of ``lambda'', and several internal mode settings. The ``-u'' flag loads the initial configuration file belonging to user $<$username$>$. \item[{\tt -t$<$tech$>$}] Load the technology file {\tt $<$tech$>$.WOL}. Only layers and colors are defined in this file; the size of ``lambda'' may have to be set explicitly. \item[{\tt -q}] Initiate quiet start-up. The message-of-the-day is not displayed, and neither is the status of CIF I/O operations. \end{description} \NEWOL\ uses menus extensively to display and select all functions, and understanding the menu system is crucial. When invoked with no filename arguements, \NEWOL\ comes to life displaying the OPTS menu (which contains the commands to load and save project files). If a project filename was specified on the command-line, \NEWOL\ starts off in the MAIN menu. The \NEWOL\ display is really quite simple. There are three areas on the screen. The first of these is the \putindex{menu area}, consisting of the group of cyan colored boxes at the top of the screen. Moving the cursor (which follows the pen on the pad) to one of these menu squares, and pushing the pen down causes \NEWOL\ to initiate whichever function the menu box represents. Some of these functions are {\em \putindex{actions}} (the function executes immediately, then returns); others are {\em \putindex{modes}} (which affect the operation of a subsequent action or pen operation). When you select a mode menu square, it is turned to purple---to show you what you have selected. The color of the printing in the menu boxes also has meaning. Selecting a box with red writing will move you to another menu, and yellow boxes represent file operations. The second area of the screen is the large area underneath the menu boxes. This is the so-called \putindex{$\lambda$-grid}, which is the working area on which \NEWOL\ draws its pictures. Notice that as you move the pen around the grid that it seems to jump around. This is because the grid is in fact a real grid. There are distinct points on the grid, and the cursor (and the corners of all \NEWOL\ objects) can only be located on a grid point. \NEWOL\ allows you to zoom in and out, and therefore when you move closer, the grid points are further apart. The zoom factor can be set by various ``keyboard commands''. It is worth mentioning here that \NEWOL\ uses a ``window'' system for the $\lambda$-grid. What you see in the grid area at any one time is only a small portion of the whole area that can be filled with cells. The ZOOM and SCROLL functions allow you to move the window, to view different parts of your work. The last area on the screen is the \putindex{status area}. This is the little box in the upper right hand section of the screen. \NEWOL\ uses this area to communicate the state of the system. Right now, there should be a colored square (the current drawing layer), and the name of the current menu (Main or Opts). There will be more on the status area later. Notice that the cursor can not be moved into the status area. This is a special reserved area, and {\em cannot} be used for drawing. There are several menu squares which have no writing in them. These are extras and will be filled with functions later. For now, selecting them will have no effect. Moving around to the various menus in \NEWOL\ is really very simple. Menus are hierarchically organized by function of their contained operations, and all of the sub-menus have a menu square called `MAIN' which will send you back to the main menu. Thus, you are always within a few pen pushes of any menu in \NEWOL. \subsection{Text pull-down menus} Specification of cell and file names is done textually, with the optional assistance of pull-down menus listing possible choices\index{text pull-down menus}. The text prompt for the name contains several fields which can be highlighted by using the pen. In particular, a default response is supplied (generated by the last cell/file name provided by the user) and is selected either with the pen or by pressing ENTER. An ABORT option is also provided. The last option is to press or type ``?'', which will pull down the full menu. Here, names are selected with the pen, or, if you wish to supply a name which is not in the menu, simply by typing the name at the keyboard (followed by ENTER). An ABORT option is also located in the menu. Rather than type ``?'' at the first text prompt, the user may wish to directly enable the pull-down menu as a default action. This is specified in the INFO.WOL configuration file, by setting the \putindex{short file/cell name prompt} variables appropriately. \subsection{Keyboard commands} Several commands, in particular those that affect the graphics state of \NEWOL\, are available at all times by pressing keys on the keyboard. This provides a shorthand mechanism for sophisticated users to quickly perform common operations. Typing ``?'' will print a short \putindex{help} screen describing most of the keyboard commands. \subsubsection{Selection commands} The \NEWOL\ algorithm for selecting objects\index{selection algorithm} on the screen is based on a hierarchical data representation in the geometry buffer. Boxes are considered to own boxes which they entirely contain. Many operations (move, delete, etc.) can be performed on the entire hierarchy at once. The selection algorithm is concerned with identifying the subtree of boxes that correspond to the pen being pressed at a particular location. The algorithm is straight-forward: \NEWOL\ tries to grab the {\em smallest} object which completely surrounds the cursor point (and of course, any children of that object), {\em unless} the cursor falls on an {\em edge}, in which case \NEWOL\ tries to grab the {\em largest} box with an edge on the cursor. A few experiments will quickly teach the user how this mechanism can be used to operate on various structures. An ambiguity exists in the case of two partially overlapping boxes; neither one contains the other (they are siblings in the hierarchy). In this case, an arbitrary (but deterministic) choice is made by the selection algorithm. In these cases, when it is difficult or inconvenient to unambiguously select a given object, two mechanisms are provided to help steer the selection algorithm: \begin{enumerate} \item The algorithms can be made principally selective to edges (as opposed to interiors of boxes) by pressing the ``E'' key. If the search for an edge fails, the selection algorithm reverts to considering interiors of boxes, as before. The ``E'' key toggles this \putindex{edge-priority selection} mode. \item The select algorithm can by cycled repeatedly, indicating in sequence ALL of the selectable objects at a given point. When the desired object is highlighted, the pen is pressed (as in a normal selection operation), to actually perform the selection. Lifting the pen at any point aborts the select process altogether. Running down the list of selectable objects is performed by pressing the \putindex{EXECUTE} key; the list itself is cyclic. The EXECUTE key sends a CTRL-C character. \end{enumerate} Because it may not always be desirable to operate on all members of the sub-hierarchy at once, various means are provided to disable its effect. If the select algorithm is run in \putindex{ONE mode}, only a single box is selected; all of its children are left untouched. This mode may be made the default by selecting the \putindex{disable heirarchy} option in either the OPTS menu of the INFO.WOL configuration file. The EXECUTE key and edge-priority select mode are still operational in ONE mode. The converse is also true: sometimes the heirarchy is inadequate, as one may want to operate simultaneously on several unrelated objects. To this end, a \putindex{SELect box} can be drawn; all boxes that are entirely contained within the box are selected. Boxes are never broken; boxes that partially overlap with the SEL box are not selected. \subsubsection{Viewing commands}\index{viewing commands} \NEWOL\ has two main buffers (see {\em geometry buffer} and {\em composition buffer}); each buffer has its own window representing the current viewing transformion. Moving between the buffers in no way changes the windows. \NEWOL\ maintains a {\em stack} of ``graphics state'' records, including such information as windowing transformations and zoom factors. Separate stacks are maintained for the geometry and composition buffers. \begin{description} \item Z -- \putindex{zoom to window}. \NEWOL\ will ask the user to draw a box, which will then be used as the next viewing window (it will be fit as best as possible on the screen, without distorting the aspect ratio). As usual, lifting the pen while drawing the box aborts the operation, and leaves the window unchanged. The old transform is pushed onto the stack. \item ``+'',``--'' -- double and halve the \putindex{zoom factor}, respectively, leaving the coordinate of the center of the screen unchanged. \item 5 -- zoom back to top level. Fits the entire contents of the buffer on the screen at once, setting zoom and center appropriately. \item @ -- center viewing window. Prompts for lambda coordinates on which to center the window. \item \putindex{arrow keys} (and \putindex{wheel}) -- scroll the screen in increments of 1/20 screen. \item 8,2,6,4 -- scroll\index{SCROLL} the screen by one page in N,S,E,W directions, respectively. \item ``)'' -- explicitly PUSH a viewing transform onto the stack. The user is asked to draw a zoom box (as for the Z command) for the new transform, but the current transform is left unchanged. \item ``\^\ '' -- POP the viewing tranform stack, redrawing the screen as required. \item ``('' -- exchange the top two tranforms. This is especially useful for jumping back and forth between two widely separated sections of the chip. Remember that the current transform is usually not yet on the stack. \item ``.'' -- PUSH the current transform onto the stack. Usually used prior to the `)' command. \end{description} \subsubsection{Display commands}\index{display commands} \begin{description} \item R, {\em space} -- force a \putindex{screen refresh}. \item F -- toggle display of boxes as filled regions. This is very convenient for examining connectivity, but outline boxes are often more useful for editing (where you need to see the edges). \item A -- toggle auto-refresh mode. \NEWOL\ tries to minimize the number of screen refreshes that are made, but even so this may slow down experienced users (especially when doing several edits in rapid succession). In these cases, it is convenient to turn off the auto-refresh, and live with a somewhat inaccurate or messy display. When required, refreshes can be forced by pressing the space bar. \item Y -- draw a single ruler\index{RULER} (Yardstick). The user is asked to select two points; a rubber-band ruler is drawn between them, and the distance in lambda is printed (if a decimal point is present in the distance value, a non-integral distance was measured). One ruler is drawn for each time ``Y'' is pressed; the rulers remain on the screen until the next refresh. \item TAB, BACKTAB -- cycle through the different display styles. \item = -- display coordinates. The (x,y) coordinates of the pen in lambda units are displayed in the lower right corner of the screen. This feature (and the ``@'' command described above) is particularly useful in locating features found by some other program (e.g. an extractor, or DRC). \item D -- toggle \putindex{display ports mode}. When enabled, ports are drawn as dotted sides on the appropriate box. \item W -- toggle \putindex{wide routing} mode. When enabled, routing wires are drawn actual width; when disabled, only centerlines are drawn (more convenient for editing). \item M -- toggle \putindex{Manhattan routing} mode. When disabled, arbitrary angle routing is permitted. \item C -- clear alpha screen. \item G -- regenerate graphics screen. This is more extensive than a simple refresh, in that the menus are regenerated as well. \end{description} \subsubsection{Filing operations}\index{filing operations} \begin{description} \item L -- load a CIF file. The user is prompted for the name of the CIF file (the pull-down filer menu is NOT available here). \item S -- save a CIF file. The user is prompted for the name of the cell to save, and the file to save it in. \item P -- enable the pull-down menus as defaults for LOAD, SAVE, NEW, MAKE operations. \end{description} \subsubsection{Other operations} \begin{description} \item ! -- \putindex{shell escape}. Puts the user in a sub-shell. When the sub-shell is exited (by typing {\em exit} or CTRL-D), the user is back in \NEWOL. \item \# -- execute a program from inside \NEWOL. The user is prompted for the program name. \item CTRL-G -- enter the \putindex{gripe system}, to submit/examine bug reports to the \NEWOL\ maintainer. \item CTRL-W -- make suggestions for high-level improvements to design tools (i.e. the next generation of \NEWOL). \end{description} \subsection{The Main menu} The \putindex{main menu} is the simplest menu in \NEWOL. There are only two actions the user can take from this menu. The first is to press the far right menu square and leave the program. The other is to press one of the other filled menu squares. This will cause \NEWOL\ to send you to the appropriate sub-menu. These sub-menus, and their basic functions, are: \begin{tabular}{ll} \putindex{LAYER} & Specify a mask layer to draw geometry \\ \putindex{BUILD} & Edit geometry cells \\ \putindex{ELEC} & Specify electrical node names and ports of geometry cells \\ \putindex{CELL} & Maintain the list of defined geometry cells \\ \putindex{COMP} & Edit and manage composition cells \\ \putindex{OPTS} & File loading and saving, and setting \NEWOL's internal state \\ \putindex{DBUG} & Provide support for program maintainers \\ \putindex{EXIT} & Leave \NEWOL. If changes have been made, you will be warned \end{tabular} \subsection{The Layer menu} This menu\index{layer menu}\index{changing layers} allows the user to select the layer of the next box(es) to be drawn. Each of the layers available is shown, and the text for each is in the color of the layer. Note: These layer text names and colors can be set by the user (see {\em technology files}). The number of available layers is also determined by the technology being used. Pressing one of the layer boxes has two effects. First, it changes the current drawing layer to the one selected (this is denoted by the color box in the status area changing color). Also it moves the user from the Layer menu to the Build menu and puts \NEWOL\ into ADD mode, ready to draw a box. The grid area shows the current geometry buffer while in the layer menu. \subsection{The Build menu} The Build menu\index{build menu}\index{drawing geometry} is the menu which allows you to draw and edit {\em geometry cells}. The menu squares perform the following functions: \begin{description} \item[\frameindex{ADD}] This menu square puts you in ADD mode, allowing you to add boxes to the geometry. While this menu square is selected (purple in color), pressing the pen down anywhere on the pad will select the first corner of the new box. After you pick the pen up, moving the pen around will make \NEWOL\ create a ``\putindex{rubber-band box}'' (RBB). This rubber-band box is not permanent, but is there to help you decide where it is that you want the other corner of the box to be. When you have put the pen where you would like the other corner of the box, simply press the pen down again. The box will then become a permanent part of the geometry of the cell. If, while drawing the RBB for the box, you decide that you would like to forget about drawing this box, simply lift the pen away from the tablet. This operation {\em aborts}\index{aborting commands} most commands, not just ADD. Alternately, pressing the pen down in the menu area will also abort the operation. \item[\frameindex{MOVE}] This menu square allows you to move boxes around the screen. While in MOVE mode, pushing the pen down on the grid area will select an object to move. When you lift the pen, the object will follow the pen around until the pen is pushed down again. If you do not wish the object to follow the pen around while MOVing (e.g. if the screen refresh is too slow), use ``\putindex{point-to-point move}'' by pressing the ``V'' key on the keyboard. In this mode, select the object to be moved by pressing the pen as before, then go to the new location of the object and press the pen again. \begin{enumerate} \item Note that while in MOVE mode, a word appears in the right of the status area. This shows the MOVE sub-mode. There are three sub-modes: ALL, ONE, SEL. Repeatedly pressing the MOVE menu square will cause \NEWOL\ to change its sub-mode. Do it several times and see what happens. The sub-modes have the following meanings: ALL means that the selected shape {\em and all shapes which are entirely inside of it} will be moved around the screen. ONE mode means that just the one shape selected will be moved. SEL (Select) is a bit more tricky. If you are in SEL mode, \NEWOL\ will wait for you to go to the pad area and press the pen down twice - drawing, in the process, a rubber-band box. Everything which is completely inside of this select box will be moved with the cursor. \item If the {\em corner} of a box is selected in MOVE ALL or ONE mode, no move is performed. Rather, a rubber-band edit of that corner of the box is done (this feature reduces the number of times you have to enter EDIT mode (below) for little corrections). \item MOVE mode has one further subtlety: when in MOVE ALL or ONE modes, if the pen is dragged while depressed in selecting an object, a SEL rubber-band box will be drawn, and the system is (temporarily) in SELect mode. \end{enumerate} \item[\frameindex{DEL}] This mode allows you to delete boxes from the geometry buffer. Like MOVE, DEL has three modes (ALL,ONE,SEL) and they allow you to delete a shape and all of its contained shapes, just the shape itself, or a selected region of the buffer. \item[\frameindex{COPY}] Copy allows you to make a copy of a box or boxes and then move the copy around the screen. Like the previous two options, you can COPY a shape and all of its included shapes, just the one shape, or all of the shapes in a select box. Once you have picked up the object to copy, you drag it around by moving the pen, and put down copies by pressing the pen. Multiple copies can be placed with multiple presses; lift the pen from the tablet when you are finished copying that object. \item[\frameindex{EDIT}] This option allows you to select one side of a box, and move that {\em whole} side in or out. While the menu square is selected, to use EDIT you move to the {\em edge} of one of the shapes on the grid. Pressing the pen down selects that edge. Now the side will follow the pen. Pushing the pen down again selects the new position of the edge. See the MOVE command for a rubber-band edit of the corner of a box. \item[\frameindex{UNDEL}] This option allows you to bring back objects that had been deleted with the delete option. Moving the pen up to the UNDEL menu square and pressing it causes the last deleted object(s) to reappear (in their last position). It does not matter how many boxes were deleted (or with which sub-mode), they all come back. Pressing the UNDEL box again will cause the boxes deleted before that to come back. In all, five objects can be brought back with the UNDEL. So if you delete six objects in a row, only the last five to be deleted can be brought back. \item[\frameindex{ZOOM}] This option allows you to select how fine the $\lambda$-grid is. When ZOOM is selected, \NEWOL\ will tell you what the current zoom factor is (i.e. how many screen pixels correspond to $1\lambda$), and will then ask for the new zoom factor. Zoom factors are most useful when integers $\geq$ 1, but non-integral zooms are permitted (primarily for viewing purposes; it is virtually impossible to select objects when the pen grid and the $\lambda$-grid do not line up). \NEWOL\ will then redraw the buffer for you, zooming around the center of the screen. If, instead of typing a number at the text prompt, the user draws a box on the screen, \NEWOL\ will try to zoom to that window (this operation is the same as the ``Z'' keyboard command\index{zoom to window}). \item[\frameindex{MAKE}] This is how you define the geometry buffer into a cell name. After you MAKE a cell, it exists in memory, independent of the GEOM buffer and can be used both in geometry and composition operations. {\em Until MAKE is done, your edits in the GEOM buffer are transient, as they were done on a {\em copy} of the cell; the MAKE operation overwrites the old entry with the GEOM buffer contents}. When you press the MAKE menu square, \NEWOL\ asks for the name of the cell. NOTE: Only one cell can have a given name. Cells names can be up to 12 characters long, so use names that mean something!! After \NEWOL\ asks for the name of the cell, it offers you two options for making the cell. If you want this cell to have the \putindex{minimum bounding box} (MBB) defined by its geometry, press the pen down again in the MAKE menu square. If you wish to define your own bounding box (which must be {\em larger} than the MBB), then move the pen to the grid, and push the pen down twice to denote the corners of the bounding box. Like before, if you decide not to make the cell (or you don't like the bounding box), and you are drawing the RBB of the bounding box, just lift the pen or move the pen up to the menu area and press it down. Like before, the box will go away, and the cell will not be made. \item[\framebox{LAYER}] Go to the layer menu. Once you select a layer, you will be returned to the BUILD menu. \item[\framebox{MAIN}] Go to the main menu. \end{description} \subsection{The Cell menu} This menu allows you to define geometry cells, and to manipulate them. Since it deals with geometry cells, it shows the current geometry buffer in the grid area. The menu squares have the following effects: \begin{description} \item[\frameindex{MAKE}] Same as MAKE in the Build menu. \item[\frameindex{LIST}] Pressing this menu square will cause \NEWOL\ to list the geometry cells (including the GEOM buffer) that it knows about. For each cell, it will list the name of the cell, the cell type (GEOM), the height, and the width. If there are more cells defined than will fit on the screen, don't worry, just hit return and the rest of the cells will be shown. \item[\frameindex{APPEND}] Pressing this menu square will cause \NEWOL\ to ask you for the name of a geometry cell. It will then ask you to press the pen down on the grid on the location of the lower left hand corner of the cell. \NEWOL\ will then take a copy off that cell and add it to the buffer. NOTE: This is {\em not} composition. Appending a cell to the buffer takes a copy of the geometry in a cell and integrates it into the buffer's data tree. Once this operation is over, there is no way to tell that any of the boxes in the buffer came from another cell. It is exactly as if the user had ADDed each box by hand in the BUILD menu. \item[\frameindex{NEW}] This operation is done immediately prior to editing a cell, and causes the cell to be copied into the GEOM buffer. \NEWOL\ will ask for the name of a geometry cell, using the cell lister menu. \NEWOL\ will then clear out the GEOM buffer, and move a copy of the named cell into the buffer. It also will change the viewing window to scale the new cell to the proper size. Remember that after the new copy of the named cell has been moved to the GEOM buffer, it has no relation to the cell in memory. Editing the buffer will only change the buffer, not the cell from which the copy was made. \item[\frameindex{ERASE}] This allows you to erase the cell definition of a cell in memory. The cell's space in memory will be collected for future use. NOTE: A cell cannot be erased if it is used in the definition of any composition cells (i.e. if it is called by any other cell). \item[\frameindex{CLEAR}] Clears the geometry buffer. The user is warned if the buffer contains a cell which has been modified but not saved. \item[\frameindex{PLOT}] Plots the contents of the GEOM buffer to a HPGL plotter. \item[\framebox{ZOOM}] Same as ZOOM in Build menu. \item[\framebox{MAIN}] Return to main menu. \end{description} \subsection{The Elec menu} This menu provides means to geometrically transform boxes, and to name nodes and ports. The latter two features are primarily used by subsequent tools (extractors, automatic routers, DRCs), as means to identify boxes. \begin{description} \item[\frameindex{ELEC}] Names a box, prompting for a string. The name is saved in the .CIF file using the ``{\tt 86 $<$name$>$ ;}'' extension. \item[\frameindex{XFORM}] Similar in concept to XFORM in the composition buffer. See the description there. There are no XFORM ONE or SEL modes, transformations apply to ALL the selected object(s). \item[\frameindex{RULER}] Enter yardstick mode; press a menu square to exit. \item[\frameindex{PORT}] Define one edge of a box as a {\em port}. When \putindex{display ports} mode is set, ports are visible as dashed edges in the geometry buffer, and as solid lines in the composition buffer (in the latter, they are visible even when the geometry cell is poked shut). The PORT command will wait for a box to be selected, then prompt for which edge (N,S,E,W) and a port name. \item[\framebox{DRC}] Enter the design-rule checker. \item[\framebox{BUILD}] Return to Build menu. \item[\framebox{MAIN}] Return to Main menu. \end{description} \subsection{The Composition Menu} This menu allows the user to compose (by abutment) as many cells as he/she would like. They can have any orientation, and can be repeated many, many times. The composition menu has its own buffer (*COMP*), and its own window to view it with. These are independent of the GEOM buffer and window. It is very important to understand how \NEWOL\ does its composition. Each composition cell (including the COMP buffer) is made up of $0-n$ instances of other cells. Each instance is an exact map of that other cell. Therefore, if you later go and change the cell from which the instances get their definition, you will also change these instances, and therefore change the composition cell. The menu squares perform the following actions: \begin{description} \item[\frameindex{ADD}] This option allows you to add an instance of either a composition or geometry cell to the buffer. \NEWOL\ will ask you for the name of the cell, and then will ask you to push down the pen on the grid where you would like the lower left corner to be. Don't worry if it's not quite right, you can always move the instance later. \NEWOL\ tries to fit as much of the name of the cell as possible into the box on the screen. \item[\frameindex{MOVE}] This allows you to move instances around the screen. As long as the MOVE menu square is lit, pressing the pen down on the grid will select an instance to move. Once the pen is lifted, the instance will follow the pen around until it is pressed down again. Unlike the build menu, you can only move one shape at a time. However, the \putindex{EXECUTE} select mechanism still works in the composition buffer. \item[\frameindex{DEL}] This allows you to delete instances that are on the screen. While the DEL menu square is lit, selecting a shape on the grid deletes it from the COMP buffer. NOTE: There is {\em no} undelete buffer for composition. You can simply ADD the cell back. \item[\frameindex{COPY}] Allows you to copy an instance, and move the copy around. When you press the pen down again, the copy stops moving, and is added to the buffer. Again, successive presses put down multiple copies; lift the pen to abort. \item[\frameindex{POKE}] Pressing the POKE menu square takes you to the POKE sub-menu, which is used to view the contents of subcells in the composition buffer. If the cell is a geometry cell, you will see the geometry in the cell. If it is a composition cell, you will see the cells which make up the composition cell. This operation is recursive, and can be repeated on the newly revealed cells. NOTE: This is just for looking at how the compositions line up, {\em NO} editing can be done on the contents revealed with a POKE. \item[\frameindex{MAKE}] When pressed, MAKE asks you for the name of the cell you wish to make. A copy of the composition buffer is then put into memory with that name. NOTE: The bounding boxes of compositions are always the minimum size. Unlike geometry cells, they cannot be set. \item[\frameindex{DRAW}] Enter the point-to-point routing menu (see below). \item[\frameindex{NEW}] This does just the same as the NEW function on the Cell menu. It clears the composition buffer and moves a copy of a new cell into the buffer. Also, it scales the window to the proper size. This is the first step to editing an existing cell. \item[\framebox{OTHER}] Since there are too many things to do for composition, there are two comp menus. Hitting OTHER causes \NEWOL\ to alternate between them. \item[\framebox{MAIN}] Returns back to main menu. \end{description} \paragraph{OTHER menu} \begin{description} \item[\frameindex{XFORM}] This menu square allows you to rotate and mirror the instances in the COMP buffer. Note that when this menu square is lit that a word appears in the right of the status area. This shows the current transform, and can be changed by repeatedly hitting the menu square. The transforms are: Rotate by 0, 90, 180 and 270 degrees, and mirror the X coordinates, mirror the Y coordinates, mirror in the line X=Y, and mirror in the line X=-Y. The transforms are all done around the center point of the instances, and the transform is applied by pressing the pen in the grid area inside of one of the instances. Since all of the transforms look like one of two boxes, \NEWOL\ uses different colors to denote the different transforms. Here are the transforms (and the strings displayed): \begin{verbatim} Rotations: null = 0 = white; R90 = red; R180 = green; R270 = amber; Reflections: m(x=0) = MX = blue; m(x=-y) = M-XY = purple; m(y=0) = MY = cyan; m(x=y) = MXY = brown; \end{verbatim} \item[\frameindex{ERASE}] When called, ERASE asks for the name of a cell to erase. It will remove the cell entirely from memory. NOTE: Like before, if the named cell is used elsewhere, \NEWOL\ will not let you erase it. \item[\frameindex{ZOOM}] Just like before, but it zooms the composition window. \item[\frameindex{ARRAY}] A fast means to lay down multiple instances of a cell. ARRAY asks for a ``basis'': two cells must be selected (by pressing the pen on them), and a replication factor, N, is requested. Then, N-2 instances of the {\em second} basis cell will be placed, separated by the distance between the two basis cells. \item[\frameindex{NAMES}] Analogous to the NODES command in the geometry buffer, NAMES allows unique names to be given to different instances of the same cell. The instance name is written out in the .CIF file using the {\tt 85 $<$name$>$ ;} extension. \item[\frameindex{LIST}] This option lists all of the composition cells that \NEWOL\ has stored in memory. The listing is different than that for geometry cells, though. For each cell, the name, cell type, height, and width is listed. Also is listed the instances that are in the cell. The instances are listed in cyan, and are inset. Also, for each instance, its transform is also listed. \item[\frameindex{PLOT}] Plot the contents of the composition buffer. \item[\framebox{OTHER}] Switches to the other comp menu. \item[\framebox{MAIN}] Back to the main menu. \end{description} \subsubsection{The Poke sub-menu} The Poke menu provides a means to view the contents of subcells within the composition buffer. It is important to remember that cell {\em editing} is only performed at the {\em top} level of the hierarchy. Thus, you cannot modify the contents of cells that you ``open up'' via the poke command; rather, you must ``NEW'' those cells into the appropriate buffer, and modify them there. \begin{description} \item[\frameindex{POKE}] This is the default action; the first time a cell is POKEd, it is opened up by one level in the hierarchy, revealing any cells that it contains. If these sub-cells are POKEd, {\em they} open up to reveal their contents. Finally, if a geometry cell is POKEd open, and is then POKEd again, it closes itself; this repeats all the way up the hierarchical tree (on successive POKEs), until the original cell is POKEd shut. \item[\frameindex{TOP}] Immediately closes all poked open cells. \item[\frameindex{BOTT}] When pressed the first time, all cells are opened {\em down to the geometry level}. The geometry cells are not opened until BOTT is pressed again; at this point, all geometry is displayed. Subsequent presses alternate between these two states. \item[\framebox{RULER}] Enters yardstick mode; press any menu button to exit. \item[\frameindex{PLOT}] Plot the contents of the composition buffer on an HPGL plotter. The ``poked'' status of the cells is the same as they appear on the screen. Also, the current directory is searched for a \putindex{LABELS.TEXT file}, which contains the (x,y) position (in lambda) and a label string to plot. \item[\frameindex{EDIT}] This command provides a shortcut to editing geometry cells. After pressing the menu button, select a geometry cell to be edited. The cell will automatically be ``NEW''ed in the geometry buffer, which will be made the current buffer. When the editing is complete, press ``POP'' in the Build menu (in the menu square which used to be occupied by ``MAKE''); the cell will automatically be Made, and the user is returned to the composition buffer. Should the user wish to abort this editing sequence, going from Build to Main causes the \putindex{POP} button to revert to MAKE. \item[\frameindex{LAYER}] Causes a sub-menu to be entered in which only selected layers are displayed. This is often convenient for design rule checking. Pressing the pen in the main display window causes a return to the Poke menu. NOTE: any cells saved to a .CIF file while in this menu will only contain the mask layers that were being displayed.\index{saving selected layers} \item[\frameindex{XTRACT}] Enters the network extraction sub-menu. \item[\framebox{COMP}] Return to the composition menu. \item[\framebox{MAIN}] Return to the main menu. \end{description} \subsubsection{The \NEWOL\ extractor}\index{topology extraction} To provide connectivity and topology verification, \NEWOL\ contains an extractor that locates all transistors and generates a netlist of their interconnection. This netlist can then be compared, via the \putindex{NETCMP} program, to that generated via a schematic capture system. In order to maximize its versatility, \NEWOL's extractor operates on a {\em flattened} representation of the geometry. Memory limitations consequently limit extraction to chips of under 20K transistors. For chips within this limit, the extractor is exceptionally fast, making use of box-level and cell-level information together with a bin-sorting algorithm to yield a speedup factor of 100+ over pixel-based extractors. Once the chip has been extracted, the connectivity information that has been generated can be displayed interactively. Entire electrical nodes can be highlighted, and estimates of their area and electrical capacitance can be made. \begin{description} \item[\frameindex{XTRACT}] Extract the cell currently loaded into the composition buffer. It is {\em vital} that this cell be freshly loaded (via a ``NEW'' command) into the buffer. You will be prompted for a file name for the .NTK output file. \item[\frameindex{FLATTEN}] Write a .CIF file with a flattened representation of the cell loaded in the composition buffer. \item[\frameindex{NODES}] Toggle the display style {\em after} the extraction. When enabled, the actual node number of each box is displayed. When disabled, a conventional display style is selected. \item[\frameindex{FLASH}] A convenient way to find a particular node, FLASH prompts for a node number, then flashes that electrical node. Nodes can also be flashed by pressing the pen down on the geometry itself. Whenever a node is flashed, some information about the node (fanout to n- and p-transistors, and optionally, area and capacitance) is also displayed. \item[\frameindex{AREA}] Enables the computation of area and capacitances of FLASHed nodes. \item[\frameindex{SHORT}] Treats all P+/N+ diodes as short circuits. These structures arise (legally) in CMOS ohmic conacts. Extracting the chip twice (once with this switch set), and comparing it against itself, ensures that all such contact structures have both sides connected to a power source. \item[\framebox{DRC}] Enter the composition buffer DRC facility. \item[\framebox{COMP}] Return to composition menu. All extraction data structures are freed when the Xtract menu is exited. \item[\framebox{MAIN}] Return to main menu. \end{description} \subsubsection{The point-to-point router}\index{routing wires} \NEWOL\ has a manual pont-to-point router, intended mainly for high-level routing such as pad-wrapping, where density considerations are offset by a desire for fast and easy wiring. The router is available only in the DRAW sub-menu of the COMP menu. The router uses {\em wires} as a primitive data structure. They require separate generation and editing paradigms than the boxes used everywhere else. The display and specification properties of routing wires are set primarily by keyboard commands, including ``W'' (toggle fleshed-out routing display), ``M'' (toggle Manhattan routing). \begin{description} \item[\frameindex{DRAW}] Draw a wire by specifying end points of line segments that make up the centerline of the wire. The wire is completed by lifting the pen (the end of the wire is the last point pressed). By definition, a wire is the set of points within 1/2 of the {\em width} from the centerline. In Manhattan mode, only right-angle bends are permitted; otherwise, arbitrary angles are permitted (the extractor, however, will ignore any non-Manhattan geometry). \item[\frameindex{LAYER}] Select the layer for DRAW. \item[\frameindex{DEL}] Delete an entire wire\index{deleting routing wires}, by pressing the pen on one of the line segment end points. \item[\frameindex{EDIT}] Edit a routing wire\index{editing routing wires}. In Manhattan mode, points or centerlines can be selected and moved; in non-Manhattan mode, only vertices can be edited. The connectivity of the wire is maintained. \item[\frameindex{WIDTH}] Set the width of the wire in lambda. \end{description} \subsubsection{The \NEWOL\ design rule checker}\index{design rule checking} \NEWOL\ provides a simple geometric design rule checker. Like the extractor, it does not operate continuously in the background, but rather, exploits the high-level box internal representation to operate quickly in a ``batch'' mode (typical performance is approximately 2500 square lambda per second). Unlike the extractor, the DRC is {\em NOT} guaranteed to find all the errors the user may have made (design-rule checking is a difficult, and presently open, research topic). The philosophy for the DRC is the following: (i) finding some errors is more useful than finding none, and (ii) only true errors should be reported. Consequently, the DRC fails by ``errors of omission'' (i.e. not finding or reporting particular errors), rather than by errors of commission (reporting a valid construct as an error). The DRC is presently being improved, so this documentation is possibly incomplete. For a more complete specification of the DRC configuration file, see the System Maintainer's Guide. \def\drcrule#1{\newline\null\hspace{1in}{\tt #1}\newline} There are four categories of rules that are required. These are, and \NEWOL's current status regarding them: \begin{description} \item[\putindex{Separation Rules}.] These rules specify some minimum distance between boxes on two particular layers. A typical example is that disconnected pieces of polysilicon must be separated by $2\lambda$. \NEWOL\ reads a configuration file of the design rules. The syntax of this rule is: \drcrule{separation } A variant on this rule provides a logical-AND between two layers to generate ``$<$layer1$>$'' above. For example: \drcrule{separation2 diff select well 5} guarantees that n-diffusions (green (diff) surrounded by select) are at least 5 lambda from a well edge. \item[\putindex{Overlap rules}.] These rules specify that some layer must be entirely contained within some other layer. For example, contact cuts must be surrounded by boxes corresponding to the two layers which are to be electrically connected: \drcrule{overlap poly polycut 2} guarantees that a poly surround of $2\lambda$ around each cut to poly. Again, binary interactions are specified by: \drcrule{overlap2 diff well well 5} which specifies that all active within the well must lie within the well by at least $5\lambda$. {\em Note that this applies to ohmic contacts to the well, in addition to any p-diffusions.} \item[\putindex{Width rules}.] These are single-layer rules, and specify the minimum width of a particular layer. For example: \drcrule{width diffcut 2} specifies that $2\lambda \times 2\lambda$ is the minimum size for a cut to active. \item[\putindex{Extension rules}.] These rules specify composite rules such as the distance that poly must extend beyond a transistor channel in order to guarantee the integrity of the channel in the case of mask alignment errors. This rule is currently not tested in \NEWOL. \end{description} The DRC has two user interfaces: it is accessible through both the geometry buffer and the composition buffer. In general, it is executed by drawing a box around the region to be checked. Any errors found are reported as white boxes drawn at the location of the error. These boxes can then be examined for information regarding the nature of the error. Errors are also reported in a file created in the current directory called {\tt drc.out}. The menu squares common to the two DRC interfaces are: \begin{description} \item[\frameindex{RUN DRC}] Execute the DRC. The user is prompted to draw a box enclosing the area to be checked. \item[\frameindex{CLEAN}] Clean up the data structures generated by the DRC. In the geometry buffer, DRC errors are incorporated into a cell's data structure as boxes on a special layer. The node names of these boxes (visible through the ELEC command, below) describe the error. In the composition buffer, special boxes are maintained as part of the buffer, and are automatically cleared when the DRC menu is exited. \item[\frameindex{WELLS}] Perform the additional tests required by the {\em separation2} and {\em overlap2} rules above. These tests are required for testing well-related rules, but are rather slow to execute. \item[\frameindex{FLASH}] Highlight the errors found by the DRC by flashing the white error boxes. As in the extractor, FLASH is terminated by lifting the pen from the tablet. \end{description} In the geometry buffer, there is an additional command to inquire about the error in question: \begin{description} \item[\frameindex{ELEC}] A short summary of the error is saved as the electrical node name of the error box. Thus, ELEC can be used to display the name. \end{description} In the composition buffer, a special command must be used to display the results of a previous DRC: \begin{description} \item[\frameindex{SHOW DRC}] When the DRC is first run, this function is executed automatically. However, as the screen is refreshed, the DRC boxes also disappear (since, technically, they are not part of the data structure); SHOW DRC is used to re-display them. \end{description} \subsection{The Options menu}\index{options menu} The last menu accessible from the main menu is the \putindex{OPTS}\index{options menu} menu. This menu contains a collection of file management functions, as well as provisions to monitor and modify \NEWOL's operation. This is the menu that allows you to save and recover your work, from session to session. \NEWOL\ saves its data in Caltech Intermediate Format (CIF -- see Chapter 4 of Mead \& Conway)\index{.CIF files}, which is directly compatible with the suite of design tools ({\bf cifp}, {\bf cifplot}, {\bf extract}) used at Caltech, as well as the verification/PG-tape generation software used by MOSIS. \begin{description} \item[\frameindex{LOAD}] This allows you to read in a file containing one or more cells. These cells may be either geometry cells, in which case they are appended to the list in the *GEOM* buffer, or composition cells, in which case the composition information will be added to the *COMP* buffer list. Any sub-cells in a composition cell will also be read in and added to the appropriate buffer lists (remember that you have to enter these buffers and NEW the cells before they will appear on the screen for editing). Thus, if your finished project is in the form of a single cell (which is required by MOSIS), it can be read in simply by pressing LOAD, and typing in the name of the file in which it is found. Filenames of up to 12 characters are permitted in the CHIPmunk operating system. \NEWOL\ will automatically append a {\tt .CIF} extension to the name you supply, if no extension is given. By default, only the current directory will be searched; full pathnames must be supplied if the file is not in this directory. \item[\frameindex{SAVE}] This command saves \NEWOL's internal data representation in a file as CIF. Cell revision numbers and last modification dates are also updated as required and written to the file. \NEWOL\ will prompt for a filename to write into (again, 12 characters maximum), and a cell name (either composition or geometry) to save. If the file already exists, \NEWOL\ will rename the original to file.CFO, as a backup feature. \index{.CFO files} Please note: only cells which you save explicitly will in fact survive when you exit \NEWOL\ (however, the program {\em will} keep track of your work, and warn you if you try to exit with unsaved changes). To save the state of the entire system, instead of a single cell, use the DUMP command (below). \item[\frameindex{PORTS}] This command creates a file PORTS.WOL that contains the (heirarchical) portnames of all instanced cells, as well as their absolute (x,y) coordinates (in lambda, with (0,0) in the lower left corner of the chip). \typeout{Check OPTS-PORTS with Johnw} This is primarily useful for interfacing with an external router. \item[\frameindex{MEM}] Prints the current state of \NEWOL's memory manager, including the (user-set) memory threshold. \typeout{Program maintainers are referred to sysmaint for further details.} \item[\frameindex{DIR}] Enters the file-lister environment (see above), to display the {\tt .CIF} files found in a particular directory. This feature has been largely replaced by the file-lister in LOAD/SAVE, and by the ability to perform shell-escapes. \item[\frameindex{OPTS}] Enters a CAGED-style textual options specification sub-menu. These are the options that are written to the INFO.WOL configuration file. The options, some typical values, and their meaning, are: {\footnotesize \begin{tabular} {lll} Lambda:& 150 & sets $\lambda$ in hundreths of microns \\ Display style:& 0 & see {\em \putindex{display styles}} \\ Short cell name prompt:& Off& if On, enter cell-lister environment immediately \\ Short file name prompt:& On & if On, enter file-lister environment immediately \\ Auto refresh:&On& see {\em \putindex{auto-refresh}} \\ Edge-priority selection:&Off& see {\em \putindex{edge-priority selection}} \\ Point-to-point move:&Off& set \putindex{point-to-point move} as default \\ Disable Hierarchy:&Off& if On, \putindex{disable hierarchy} for select mechanism \\ Display Bounding Boxes:& On& \\ Display Ports:& Off& set \putindex{display ports} mode \\ Manhattan routing:& On& if On, limits DRAW to rectangular geometry \\ Fleshed-out routing:& Off& If Off, draws centerlines only \\ Background color:& 0& sets gray level of screen background \\ Memory threshold:& 20000& \NEWOL\ will warn when this is reached \\ CIF symbol start:& 1 & Symbol number to start numbering at \\ Technology file:& SCG& \end{tabular} } \item[\frameindex{CNFG}] Writes an INFO.WOL file into the current directory. When \NEWOL\ begins execution it searches for this configuration file in (i) the current directory, (ii) the user's home directory, and (iii) the WOL library directory (/LIB/WOLLIB). \item[\frameindex{TECH}] Prompts for the name of a technology file to read in. The default file is specified in the INFO.WOL configuration file. Since different technologies often have different numbers of layers, it is often not possible to change technologies from within \NEWOL. The preferred mode is to use the MOSIS program to perform layer conversion, and/or use the {\tt -t$<$tech$>$} command-line option to \NEWOL. \item[\frameindex{DUMP}] Causes all cells, geometry and composition, to be written to a single file, with no top-level call. This is a convenient way to save all of your work at intermediate points in your editing session. For submission to MOSIS, however, CIF must be generated via the SAVE command. \item[\framebox{MAIN}] Return to MAIN menu. \end{description} \subsection{The DBUG Menu} This menu, accessible from the main menu, is primarily intended for emergency debugging and problem identification. There are very {\em *DANGEROUS*} operations in this menu, and they should not be fooled around with, unless you think you know what you are doing. The exception to this caveat is the WCOMP command; it is located in this menu because of its fairly specialized nature. \begin{description} \item[\framebox{GEOM}] Dump the data structures of the geometry buffer. \item[\framebox{GCELL}] Dump the data structures of the geometry cell dictionary list. \item[\framebox{COMP}] Dump the data structures of the composition cell dictionary list. \item[\framebox{CCELL}] Dump the data structures of the composition buffer. \item[\frameindex{EMERG}] Take emergency buffer action. The entire contents of the geometry or composition buffers can be destroyed. No attempt is made to reclaim any memory. This command is only intended for use in truly dire straits, and to buy enough time to save what you can, then exit \NEWOL. \item[\frameindex{OPTS}] Another path into the OPTS text sub-menu, should the primary path (through the Options menu) become corrupted. \item[\frameindex{MAKE}] Another path to MAKE either the geometry or composition buffer. \item[\frameindex{WCOMP}] A means to individually save all the cells loaded in \NEWOL\index{saving all cells}. This is useful for the WOLCOMP composition system, which requires files to contain single geometry cells. The user is prompted if only newly modified cells are to be dumped, and if only geometry cells should be considered. \item[\frameindex{FLUSH}] Re-initializes the memory manager. Effectively, this wipes out the \NEWOL\ data structures, clearing the cell dictionaries and buffers, and basically resetting \NEWOL\ to a power-up state. \end{description}