1、Copyright Government Electronics the only change would be in the appearance of the name itself. Thus, no hidden meaning should be assumed or expected with respect to the actual choice of names. This sharply contrasts with native EDIF keywords (and symbolic constants), whose semantics are fully defin
2、ed by the EDIF Reference Manual. The motive behind the use of a restricted alphabet for identifiers is to avoid lexical ambiguity in EDIF itself (as would be the case if ( and 9 were acceptable identifiers), and further to avoid much of the ambiguity and confusion that could occur in a receiving sys
3、tem. Such systems typically have an underlying syntax which prevents many characters such as + or ; from being accepted. Throughout this document and the EDIF Reference Manual, rather than specifying items as identifier, more descriptive terms are used. For example, entries which end in . . .NameDef
4、 or . . .NameRef, such as cellNameDef or portNameRef, are of type identifier or name. Where a name is being defined, it ends with NameDef; if the name is being referenced, it ends with NameRef. A particular identifier may only be defined once in the file, but different identifiers may have the same
5、name in different contexts. Example: (library ABC . (cell ABC .) The array construct is used to define an array which may be multi-dimensional. Example: (array Arr 2 3) Note that 2 and 3 specify the total number of elements in the first and second dimension respectively. Array reference is achieved
6、through an explicit member construct. Any name created through the array construct may be used within member to indicate individual members. Each index value must be an integer which is at least zero, and less than the size of the corresponding array dimension. The first integer in member correspond
7、s to the first integer in array, and so on. The array definition above creates six names for six objects, which could be referenced as follows: 2-3 Copyright Government Electronics it is simply a coincidence of spelling. Very few design systems employ flat naming schemes; names are typically valid w
8、ithin a well-defined scope. For example, the name chosen for a port in one cell cannot be confused with a port of another cell that happens to have the same name spelling. Such local names require some form of augmentation or qualification before the local object can be referenced from another conte
9、xt, such as an indication of a component port in a netlist. This requires both the instance name of the component and the name given to the corresponding port within the definition of the referenced cell. The define-before-use rule in EDIF ensures the integrity of the name mechanism: each name must
10、be defined before it may be referenced. 2.4.2 Name Class The concept of name class is an essential aspect of object referencing in EDIF. The type of object being named determines the class of its name, such as cell name or logic value name. Any identifier used as a reference is associated with an ob
11、ject class by its immediate context. The various name classes are all subject to the same scoping rules, and can be thought of as parallel, independent naming schemes. There is no connection or interference between names of different classes, even when such names happen to involve the same identifie
12、r spelling and have overlapping ranges. In addition, none of these classes has any conflict with identifiers used as keywords or symbolic constants: a cell could be named and or even cell without implying any connection with the corresponding keywords. Some name classes may be appropriate for more t
13、han one keyword. For example, external and library both introduce library names. The following table shows the EDIF name classes. EDIF Name Classes cell design figureGroup formal keyword layer logic net property rule value view edifFile instance library PO* simulate 2-5 Copyright Government Electron
14、ics they are not allowed between digits. Example: 100 O -123454321 +64 - are all legal integer values. 2.5.2 Number Value A number is either an integer or scaled-integer quantity. Either one may be used in any EDF description that expects a number. A scale factor can be applied to an integer to repr
15、esent fractional numbers or numbers greater than can be expressed using an integer alone. These scaled integers are represented in the form: (e integeriroken integeriroken) The first value represents the mantissa. The second value is the exponent of the scale factor. The resulting number has a value
16、 equal to the mantissa times ten to the power of the exponent. The scaled integer must lie within the range:* . Example: (e 5125 -3) represents 5.125 Numbers are used to represent a variety of types of data. For example, numbers are used for the width of paths (a distance) and the delay of a logic g
17、ate (a time). Numbers are dimensionless in an EDF file and may be translated upon output to dimensioned, SI quantities via the scale statement; that is, while a path width may be expressed as 3, it will be converted to three mm or three microns depending on the value of the distance scale factor for
18、 the current library. A scale factor of, (e 254 -10) will convert the distance to 3 mils. 2.5.3 String Value Strings can contain any characters, including blanks, and are delimited by quotation marks ”. If characters such as backspace and newline are required to appear in the string (e. g. when it i
19、s displayed) then they must be explicitly represented. This means that whitespace characters, other than space and tab, have no effect in the string. The characters % and ” must also be explicitly represented if required in the string. Copyright Government Electronics the range has no lower limit an
20、d must have an upper limit of 60. 2.5.7 Displaying Values As in the case of names, values can also be assigned a location for display purposes when they are used in a context that has an associated coordinate system. For that purpose, the following display constructs are provided: booleanDisplay, in
21、tegerDisplay, keyworduisplay, miNoMaxDisplay, numberDisplay, parameterDisplay, propertyDisplay, and stringDisplay. Example: (numberDisplay numbervalue display ) Parameters and properties (and also constants and variables for higher EDIF levels) assign values to names. These constructs also have appr
22、opriately named display constructs in order to allow the value to be displayed. Example: (parameterDisplay W (display (origin -3 O) displays the value of parameter W at the point (-3,O). 2-1 1 - Copyright Government Electronics there may be information (such as detailed cell libraries) added to the
23、data later by the user of the file. At the top level, the file may contain only five types of information, other than comments and userData: o one and only one edifFileNameDef, ediflersion, edifkvel, and keywordMap o. at most one status block o zero or more external library specifications o zero or
24、more cell definition libraries o zero or more design blocks The ediflrersion must be present after the edipileNumeDef and must be followed immediately by edifhvel and keywordMap. The remaining constructs may then follow in any order, provided all named objects are defined before use. (edif edifFileN
25、ameDef ed iflersion edifkvel key wordMap external I library I design I comment I userData I edif name of EDIF file (for external use) version of EDIF used level of EDF file user-defined keyword macros and aliases status block for file used for specifying external libraries libraries containing cell definitions the starting points for this file user comments user-defined constructs edif 3-2 Copyright Government Electronics & Information Technology Association Reproduced by IHS under license with GEIA Not for ResaleNo reproduction or networking permitted without license from IHS-,-,-
