Verilog Reference Manual


Preface:

This is a brief summary of the syntax and semantics of the Verilog Hardware description Language. The summary is not intended at being an exhaustive list of all the constructs and is not meant to be complete. This reference guide also lists constructs that can be synthesized. For any clarifications and to resolve ambiguities, please refer to the Verilog Reference Manual Copyright 1993 by Open Verilog International Inc. and synthesis vendors Verilog HDL Reference Manuals.

Note: this document will be constantly updated with new verilog design examples and useful tricks. A last modified date will be placed at the beginning of the file.

Last modified 14/04/97

Quick Reference for Verilog HDL


1.0 Lexical Element

2.0 Registers and Nets

3.0 Compiler Directives

4.0 System Tasks and Functions

5.0 Reserved Keywords

6.0 Structures and Hierarchy

6.1 Module Declarations

6.2 UDP Declarations

7.0 Expressions and Operators

8.0 Named Blocks, Disabled Blocks

9.0 Tasks and Functions

10.0 Continuous Assignments

11.0 Procedural Assignments

12.0 Gate Types, MOS and bidirectional switches

13.0 Specify Blocks

14.0 Verilog Synthesis Constructs

15.0 Index

16.0 Synopsys Methodology Note: Verilog RTL descriptions

17.0 Synopsys Methodology Note: ASIC Partitioning


Use and Copyright

Copyright (c) 1994, 1995 Rajeev Madhavan

Copyright (c) 1994, 1995 Automata Publishing Company

Permission to use, copy and distribute this book for any purpose is hereby granted without fee provided that

(i) the above copyright notices and this permission notice appears in all copies and

(ii) the name of Rajeev Madhavan, Automata Publishing and AMBIT Design Systems may not be used in any advertising or publicity relating to this book without the specific prior permission of Rajeev Madhavan, Automata Publishing and AMBIT Design Systems.


1.0 Lexical Elements

The language is case sensitive and all the keywords are lower case. White space, namely spaces, tabs and new-lines are ignored Verilog has two types of comments:

1. One Line comments start with // and end at the end of the line

2. Multi line comments start with /* and end with */

Variable names have to start with an alphanumeric character or underscore followed by alohanumeric or underscore characters. The only exception to this are the system tasks and functionswhich start with a dollar sign. Escaped Identifiers (identifier whose first character is a backslash (\))permit non alphanumeric characters in Verilog name. The escaped name includes all the characters following the backslash until the first whitespace character.


1.1 Integer Literals

Binary literal: 2'b1z

Octal literal: 2'o17

Decimal literal: 9 or 'd9

Hexedecimal literal: 9'h189

Integer Literals can have underscaore embedded in them for improved readability.

For example:

Decimal Literal 24_000


1.2 Data Types

The values z and Z stand for high impedance and x and X stand for uninitialised variables or nets with conflicting drivers. String symbols are enclosed with double quotes ("string") and cannot span multiple lines. Real number literals can be either in fixed notation or in scientific notation.

Real and integer Variables example:

real a, b, c ; // a, b, c to be real

integer j, k ; // integer variable

integer i[1:32] ; // array of integer variables

Time, registers and variable usage:

time newtime;

/* time and integer are similar in functionality,

time is an unsigned 64- bit used for time variables */

reg [3*14:1] string ;

/* this defines a vector with range

      [msb_expr: lsb_expr] */

initial begin

    a = 0.5 ; // same as 5.0e-1 real variable

    b = 1.2E12 ;

    c = 26.19_60_e-11 ; // the _'s are used for readability

    string = " string example " ;

    newtime = $time ;

end


2.0 Registers and Nets

A register stores its value from one assignment to the next and is used to model data storage elements.

reg {5:0] din;

/* a 6 bit vector register: individual bits din[5],... din[0] */

Nets correspond to physical wires that connect instances. The default range of a wire or reg is one bit. Nets do not store values and have to be continuously driven. If a net has multiple drivers (for example two gate outputs tied together), then the net value is resolved according to its type.

Net types:

wire                           tri
wand                           triand
or                             trior
tri0                           tri1
supply0                        supply1
trireg

For a wire, if all the drivers have the same value then the wire resolves to this value. If all the drivers except one have a value of z the the wire resolves to the non z value. If two or more non z drivers have different drive strength then the wire resolves to the stronger driver. If two drivers of equal strength have different values, then the wire resolves to x. A trireg net behaves like a wire except that when all the drivers of the net are in high impedance (z) state, then the net retains its last driven value. trireg 's are used to model capacitive networks.

wire net1;
/* wire and tri have same functionality. tri is used 
for multiple drive internal wire */
trireg (medium) capacitor ;

/* small medium and weak are used for charge strength modelling */

A wand net or triand net perates as a wired and (wand) , and a wor net or trior net operates as a wired or (wor), tri0 and tri1 nets model nets with resistive pulldown or pullup devices on them. When a tri0 net is not driven, then its value is 0. When a tri1 net is not driven then its value is 1. supply0 and supply1 model nets that are connected to ground or power supply.

wand net2; //wired and
wor net3; // wired or
triand [4:0] net4; // multiple drive wand
trior net5; // multiple drive wor
tri0 net6;
tri1 net7;
supply0 gnd; // logic 0 supply wire
supply1 vcc; // logic 1 supply wire

Memories are declared using register statements with address range specified as in the following example:

reg [15:0] mem16x512 [0:511];
// 16 bit by 512 word memory
// mem16x512[4] addresses word 4
// the order of lsb:msb or msb:lsb is not important

Note: it is not possible to access an individual element of an array. Arrays can only be accessed in terms of words.

The keyword scalared allows access to bits and parts of a bus and vectored allows the vector to be modified only collectively:

wire vectored [5:0] neta; /* a six bit vectored net */
tri1 vectored [5:0] netb; /* a six bit vectored tri1


3.0 Compiler Directives:

Verilog has compiler directives which affect the processing of the input files. The directives start with a grave accent (`) followed by some keyword. A directive takes effect from the point that it appears in the file until either the end of all files or until another directive that cancels the effect of the first one is encountered.

For example:

`define OPCODEADD  00010


This defines a macro named OPCODEADD. When the text `OPCODEADD appears in the text, then it is replaced by 00010. Verilog macros are simple text substitutions and do not permit arguments.

`ifdef SYNTH <Verilog Code> `endif


If SYNTH is a defined macro, then the verilog code until `endif is inserted for the next processing phase. If “SYNTH” is not defined macro then the code is discarded.

`include <Verilog File>


The code in <Verilog File> is inserted for the next processing phase. Other standard compiler directives are listed below.

`resetall - resets all compiler directives to default values
`define - text macro substitution
`timescale 1ns / 10ps - specifies time unit/precision
`ifdef, `else, `endif - conditional compilation
`include - file inclusion
`signed, `unsigned -operator selection (OVI 2.0 only)
`celldefine, `endcelldefine - library modules
`default_nettype_wire - default net types
`unconnected_drive pul0|pull1, `nounconnected_drive - pull up or downn unconnected ports.
`protect and `endprotect - encryption capability
`protected and `endprotected - encryption capability
`expand_vectornets and `noexpand_ectornets, 
`autoexpand_vectornets - vector expansion options
`remove_gatename, `noremove_gatenames - remove gatenames for more than one instance.
`remove netname, `noremove_netnames - remove netnames for more than one instance.





4.0 System Tasks and Functions

System tasks are tool specific tasks and functions:

$display (“Example of using function”);
$monitor($time, “a = %b, clk = %b, add = %h”,a,clk,add); // monitor signals
$setuphold(posedge clk, datain, setup, hold); // setup and hold checks

A list of standard system tasks and functions are shown below:

$display, $write - utility to display information
$fdisplay, $fwrite - write to file
$strobe, $fstrobe - display, write simulation data
$monitor, $fmonitor - monitor, display/write information to file
$time, $realtime - current simulation time
$finish - exit the simulator
$stop - stop the simulator
$setup - setup timing check
$hold, $width - hold, width timing checks
$setuphold - combines setup and hold timing checks
$readmemb, $readmemh - read stimulus patterns into memory
$sreadmemb, $sreadmemh - load data into memory
$getpattern - fast processing of stimulus patterns
$history - print command history
$save, $restart, $incsave - saving, restarting, incremental saving
$shm_open("database.shm"); -- open/create cwaves database in Verilog-XL
$shm_probe(top_level_module,"AS"); - probes all signals (AS) in the 
                                   scope top_level_module. (Verilog-XL)
$shm_close(); - will close all open shm databases in Verilog-XL
$scale - scaling timeunits from another module
$showscopes - complete list of named blocks, tasks,modules...
$showvars - show variables at scope.





5.0 Reserved Keywords

The following is a list of reserved words of Verilog HDL as of OVI LRM 2.0

and
always
assign
attribute
begin
buf
bufif0
bufif1
bufif1
case
cmos
deassign
default
defparam
disable
else
endattribute
end
endcase
endfunction
endprimitive
endmodule
endtable
endtask
event
for
force
forever
fork
function
highhz0
highhz1
if
initial
inout
input
integer
join
large
medium
module
nand
negedge
nor
not
notif0
notif1
nmos
or
output
parameter
pmos
posedge
primitive
pulldown
pullup
pull0
pull1
rcmos
reg
release
repeat
rnmos
rpmos
rtran
rtranif0
rtranif1
scalared
small
specify
specparam
strong0
strong1

supply0
supply1
table
task
tran
tranif0
tranif1
time
tri
triand
trior
trireg
tri0
tri1
vectored
wait
wand
weak0
weak1
while
wire
wor





6.0 Structures and Hierarchy
Hierarchical HDL structures are achieved by defining modules and instanciating modules. Nested module definitions (i.e. one module definition within another) are not permitted.

6.1 Module Declarations
The module name must be unique and no other module or primitive can have the same name. The port list is optional. A module without a port list or with an empty port list is typically a top level module. A macro module is a module with a flattened hiererchy and is used by some simulators for efficiency.

Module definition example:

module dff (q,qb,clk,d,rst);
input clk,d,rst; // input signals
output q,qb; output definition

// inout ofr bidirectionals

//Net type declarations
wire d1, db1;

parameter value assignment
parameter delay1 = 3,
                delay2 = delay1 + 1;  // delay2
                                      //shows parameter dependence.

/* Hierarchy primitive instantiation, port connection in this section is by ordered list */

nand #delay1 n1(cf,d1,cbf),
                      n2(cbf,clk,cf,rst);
nand #delay2 n3(d1,d,db1rst),
                       n4(db1,d1,clk,cbf),
                       n5(q,cbf,qb),
                       n6(qb,db1,q,rst);

/****** for debugging model initial begin

#500 force dff_lab.rst = 1;
#550release dff_lab.rst;

// upward path referencing

end ****/

endmodule

Overriding Parameters Example

module dff_lab;
reg data,rst;
dff d1 (.qb(outb), .q(out), .clk(clk),.d(data),.rst(rst));
//overriding module parameters
defparam
dff_lab.dff.n1.delay1 = 5;
dff_lab.dff.n2.delay2 = 6;
//full path referencing is used.
//overriding by using #(8,9) delay1 = 8...
dff d2 #(8,9) (outc, outd, clk, outb, rst);
//clock generator
always clk = #10 ~clk;
// stimulus continued...

Stimulus and Hierarchy Example:

initial begin: stimuli  //named block stimulus
   clk = 1; data = 1; rst = 0;
   #20 rst = 0;
   #20 rst = 0;
   #600 $finish;
end

initial // hierarchy, downward path referencing
   begin
      #100 force dff.n2.rst = 0;
      #200 release dff.n2.rst;
    end
endmosule




6.2 User defined primitives (UDP) declarations

The UDPs are used to augment the gate primitives and are defined by truth tables. Instances of UDPs
can be used in the smae way as gate primitives. There are two types of primitives:

1. Sequential UDPs permit initialisation of output terminals which are declared to be of reg type
and they store values. Level sensitive entries take precedence over edge sensitive declarations. An input
logic state z is interpreted as an x. Similarly only 0, 1, x or - (unchanged) logic values are permitted on
the output.

2. Combinational UDPs do not store values and cannot be initialised. The following additional
abbreviations are permitted in UDP declarations:

Logic/State Representation/transition Abbreviation
don’t care (0, 1 or X) ?
Transitions from logic x to logic y (xy)
(01), (10), (0x), (1x), (x1), (x0), (?1) ..
(xy)
Transition from (01) R or r
Transition from (10) F or f
(01), (0X), (X1): positive transition P or p
(10), (1x), (x0): negative transition N or n
Any transition * or (??)
binary don’t care (0,1) B or b


Combinational UDP’s example:

// 3 to 1 multiplexor with 2 select

primitive mux32 (Y, in1, in2, in3, s1, s2);
 input in1, in2, in3, s1, s2;
 output Y;
table
// in1 in2 in3 s1 s2    Y
0   ?    ?    0   0  :  0;
1   ?    ?    0   0  :  1;
?   0    ?    1   0  :  0;
?   1    ?    1   0  :  1;
?   ?    0    ?   1  :  0;
?   ?    1    ?   1  :  1;
0   0    ?    ?   0  :  0;
1   1    ?    ?   0  :  1;
0   ?    0    0   ?  :  0;
1   ?    1    0   ?  :  1;
?   0    0    1   ?  :  0;
?   1    1    1   ?  :  1;

endtable
endprimitive



Sequential Level Sensitive UDP’s Example:

//latch with async reset
primitive latch (q, clock, reset, data)
input clock reset, data;
output q;
reg q;

initial q = 1’b1; // initialisation

table
// clock reset data q, q+
? 1 ? : ? : 1 ;
0 0 0 : ? : 0 ;
1 0 ? : ? : - ;
0 0 1 : ? : 1 ;

endtable
endprimitive


Sequential Edge Sensitive UDP’s example:

// edge triggered D flip flop with active high
// async reset and set.
 Primitive dff (QN, D, CP, R, S);
output QN;
input D, CP, R, S;
reg QN;

table
//D CP R S : Qtn: Qtn+1
1 (01) 0  0 : ?  : 0 ;
1 (01) 0  x : ?  : 0 ;
?   ?  0  x : 0  : 0 ;
0 (01) 0  0 : ?  : 1 ;  // clocked data
0 (01) x  0 : ?  : 1 ;  // pessimism
?   ?  x  0 : 1  : 1 ;  // pessemism
1 (x1) 0  0 : 0  : 0 ;
0 (x1) 0  0 : 1  : 1 ;
1 (0x) 0  0 : 0  : 0 ;
0 (0x) 0  0 : 1  : 1 ;
?   ?  1  ? : ?  : 1 ;  // async clear
?   ?  0  1 : ?  : 0 ;  // async set
?   n  0  0 : ?  : - ;
*   ?  ?  ? : ?  : - ;
?   ? (?0)? : ?  : - ;
?   ?  ?(?0): ?  : - ;
?   ?  ?  ? : ?  : x ;
endtable
endprimitive





7.0 Expressions and Operators

Arithmetic and logical operators are used to build expressions. Expressions perform operation on one
or more operands, the operands being vectored or scalared nets, registers, bit selects, part selects,
function calls or concatenations thereof.


a = !b
if (a < b) // if expression
{c,d} = a + b;
/ concatenate and add operator


Operator Precedence:

Operator Precedence
+, -, !, ~ (unary) Highest
*, /, %
+, - (binary)
<<, >>
<, <=, >, >=
=, ==, !=
===, !==
&, ~&
^, ^~
|, ~|
&&
||
?: Lowest

Relational Operators:

Operator

Application

<

a < b // is a less than b?
// returns 1 bit true/false

>

a > b // is a greater than b

>=

a >= b // is a greater than or equal to b

<=

a <= b // is a less than or equal to b

Arithmetic Operators

Operator

Application

*

c = a * b; // multiply a with b

/

c = a/b; // int divide a by b

+

sum = a + b; // add a and b

-

diff = a - b; // subtract b from a

%

amodb = a % b; // a mod(b)

Logical Operators:

Operator

Application

&&

a && b; // is a and b true? Returns 1-bit true/false

||

a || b; // is a or b true? Returns 1-bit true/false

!

if (!a) // if a is not true
c = b; // assign b to c

Equality and Identity Operators

Operator

Application

=

c = a; // assign a to c

==

c == a; /* is c equal to a returns 1 bit true/false
applies for 1 or 0, logic equality, using X or Z
operands returns always false
`hx == `h5 returns 0 */

!=

c != a; // is c not equal to a, returns 1-bit true/
// false logic equality

===

a === b // a is identical to b (includes 0, 1,x,z)
// `hx === `h5 will return 0

!==

a !== b; / is a not identical
// to b returns 1-bit true/false

Unary, Bitwise and Reduction Operators

Operator Application

+

Unary plus and arithmetic (binary) addition

-

Unary negation and arithmetic (binary)
subtraction

&

b = &a; // AND all bits of a

|

b = |a; // OR all bits

^

b = ^a; // EXOR all bits of a

~&, ~|, ~^

NAND, NOR, EXNOR all bits together. e.g. c = ~& b; d = ~| a; e = ^c;

~,&,|,^

bit-wise NOT, AND, OR EXOR
b = ~a; // invert a
c = b & a; // bitwise AND a,b
e = b | a; // bitwise OR
f = b ^ a; // bitwise EXOR

~&,~|,~^

bitwise NAND, NOR, EXNOR
c = a ~& b; d = a ~| b;
e = a ~^ b;

Shift Operators and other Operators

Operator

Application

<<

a << 1; shift left a by 1 bit

>>

a >> 1; shift right a by 1 bit

?:

c = sel ? a : b; /* if sel is true, c = a, else c = b,
?: ternary operator

{}

{co, sum} = a + b + ci;
/* add, a, b, ci assign the overflow to co and the
result to sum; operator is called concatenation */

{{}}

b = {3{a}} /* replicate a three times, equivalent to
{a, a, a} */



7.1 Parallel Expressions
fork ... join are used for cncurrent expression assignments

fork ... join example

initial
   begin : block
      fork
           //This waits for the first event a
           //or b to occur.
           @a disable block
           @b disable block

           // reset at absolute time 20
           #20 reset = 1;
           // data at absolute time 100
           #100 data = 0;
           // data at absolute time 120
           #120 data = 1;
       join
    end





7.2 Conditional Statements
The most commonly used conditional statement is the if, if ... else ... conditions. The statement occurs if
the expressions controlling the if statement evaluates to true.

if ... else .. conditions example

always @(rst)  // simple if ... else
   if (rst)
         // procedural assignment
         q = 0;
   else // remove the above continuous assign
          deassign q;

always @ (WRITE or READ or STATUS)
   begin
   // if - else - if
       if (!WRITE) begin
           out = oldvalue;
       end
       else if  (!STATUS) begin
           q = newstatus;
          STATUS = hold;
       end
       else if  (!READ) begin
          out = newvalue;
       end
end

case, casex, casez: case statements are used for switching between multiple sections ( if (case1) ... else
if (case2) ... else ...). If there are multiple matches, only the first is evaluated. Casez treats high
impedance values as don’t care and casex treats both unknown and high impedance as don’t care’s.

Case statement example

module d2x8 (select, out); // priority encode
   input [0:2] select
   output [0:7] out;
   reg [0:7] out;

always @ (select) begin
   out = 0;   //default statement
case (select)
   0: begin   // can have begin and end statements for
     out[0] = 1;  // multiple lines...
      end
   1: out[1] = 1;
   2: out[2] = 1;
   3: out[3] = 1;
   4: out[4] = 1;
   5: out[5] = 1;
   6: out[6] = 1;
   7: out[7] = 1;
endcase
end
endmodule


casex statement example

casex (state)
   // treats both x and z as don’t care
   // during comparison: 3’b01z, 3’b01x, 3’b011
   // ... match case 3’b01x
   3’b01x: fsm = 0;
   3’b0xx: fsm = 1;
   default: begin
   // default matches all other occurences
   fsm = 1;
   next_state = 3’b011;
   end
endcase

casez statement example

casez (state)
   //treats z as don’t care during comparison:
   // 3’b11z, 3’b1zz, ... match 3’b1??: fsm = 0;
   3’b1?? : fsm = 0; // is msb   = 1, matches  3’b1??
   3’b01? : fsm = 1;
   default : $display (“wrong state”);
endcase


7.3 Looping statements

forever, for, while and repeat loops example

forever
// should be used with disable or timing control
   @(posedge clock) {co,sum} = a + b + ci;

for (i=0; i< 7; i=i+1)
   memory[i] = 0; // initialize  to 0

for (i=0; i<= bit-width; i = i+1)
   // multiplier using shift left and add
   if (a[i]) out = out + (b << (i-1));

repeat (bit-width) begin
   if (a[0]) out = b + out;
   b = b << 1; // multiplier using
   a = a << 1 ;// shift left and add
end

while (delay) begin @(posedge clk);
   ldlang = oldldlang;
   delay = delay -1;
end





8.0 Named Blocks, Disabling Blocks

Named blocks are used to create hierarchy within modulesand can be used to group a collection of
assignments or expressions. Disable statement is used to disable or de-activate any named block, tasks
or modules. Named blocks, tasks can be accessed by full or reference hierarchy paths (example
dff_lab.stimuli). Named blocks can have local variables.

Named blocks and disbale statement example:

initial forever @(posedge reset)
   disable MAIN;   // disable named block
                   // tasks and modules can also be disabled.

always begin: MAIN  // defining named blocks
   if (!qfull) begin
      #30 recv(new,newdata); // call task
      if (new) begin
         q[head] = newdata;
         head = head + 1;   // queue
      end
   end
   else
       disable recv;
end   // MAIN


9.0 Tasks and Functions

Tasks and functions permit the grouping of common procedures and then executing these procedures
from different places. Arguments are passed in the form of input/inout values and all calls to functions
and tasks share variables.

The differences between tasks and functions are:

Tasks Functions
Permits time control Executes in one simulation time
Can have zero or more arguments requires at least one input
Does not return value, assigns value to outputs Returns a single value, no special output
declarations required
Can have output arguments, permits #, @, ->,
wait, task calls
Does not permit outputs, #, @, ->, wait, task calls


task Example:

task recv;
   output valid;
   output [9:0] data;
   begin
        valid = inreg;
        if (valid) begin
           ackin = 1;
           data = qin;
           wait(inreg);
           ackin = 0;
        end
   end

task instantiation
always begin : MAIN   // named definition
        if (!qfull) begin
        revc(new,newdata);  // call task
        if (new) begin)
           q[head] = newdata;
           head = head + 1;
        end
        end else
           disable recv;
end //MAIN


function Example:

module foo2 (cs, in1, in2, ns);
   input [1:0] cs;
   input in1, in2;
   output [1:0] ns;
 
   function [1:0] generate_next_state
   input [1:0] current_state;
   input input, input2;
   reg [1:0] next_state
   // input1 causes 0->1 transition
   // input2 causes 1->2 transition
   // 2->0 illegal and unknown states go to 0
   begin
      case( current_state)
        2’h0: next_state = input1 ? 2’h1 : 2’h0;
        2’h1: next_state = input2 ? 2’h2 : 2’h1;
        2’h2: next_state = 2’h0;
        default: next_state = 2’h0;
      endcase
   generate_next_state = next_state;
end
endfunction

assign ns = generate_next_state(cs,in1,in2);

endmodule






10.0 Continuous Assignments

Continuous assignments imply that any change on the RHS of the assignment occurs, it is evaluated and assigned to the LHS. These assignments thus drive both vector and scalar values onto nets. Continuous assignments alwyas implement combinational logic (possibly with delays). The driving strengths of a continuous assignment can be specified by the user on the net types.

Continuous assignment on declaration:

/* since only one net15 declaration exists in a give module, only one such 
declarative continuous assignment per signal is allowed */

wire #10 (strong1, pull0) net15 = enable;
/* delay of 10 for continuous assignment with strengths 
of logic 1 as strong1 and lgic 0 as pull0 */

There are 4 ways of instantiating wires/logic


Continuous assignment on already declared nets.

Assign #10 net15 = enable;
assign (weak1, strong0) {s,c} = a + b;





11.0 Procedural Assignments

Assignments to register data types may occur within always, initial, task and functions. These expressions are controlled by triggers which cause the assignments to evaluate. The variables to which the expressions are assigned must be made of bit-select or part-select or whole element of a reg, integer, real or time. These triggers can be controlled by loops, if ... else construncts. Assign and deassign are used for procedural assignments and to remove the continuous assignments.

module dff (q, qb, clk, d, rst);
   output q, qb;
   input d, rst, clk;
   reg q, qb, temp;
   always
       #1 qb = ~q;  // procedural assignment

   always @(rst)
      // procedural assignment with triggers
      if (rst) assign q = temp;
      else deassign q;

always @ (posedge clk)
    temp = d;

endmodule




force and release are also procedural assignments. However they can force or release values on net data
types and registers.

11.1 Blocking Assignment

module adder (a,b,ci,co,sum,clk);
   input a,b,ci,clk;
   output co,sum;
   reg co,sum;

always @(posedge clk) // edge control
   // assign co, sum with previous value of a,b,ci
   {co,sum} = #10 a + b + ci;
endmodule

How Synopsys handles blocking and non-blocking assignments...


11.2 Non-Blocking Assignment

Allows scheduling of assignments with out blocking the procedural flow. Blocking assignmets allow timing control which are delays whereas , non-blocking assignments permit timing control which can be delays or event contol. Th non-blocking assignment is used to avoid race conditions and can model RTL assignments.

/* assume a = 10, b = 20, c = 30, d = 40 at the start of block */

always @ (posedge clk)
   begin: block
        a <= #10 b;
        b <= #10 c;
        c <= #10 d;

/* at the end of the block + 10 time units, a = 20, b = 30, c = 40 */




12.0 Gate Types, MOS and Bidirectional Switches

Gate delcarations permit the user to instantiate different gate types and assign drive strengths to the
logic values and also any delays.

  <gate-declaration> ::= <component>
     <drive_strength> ? <delay> ? <gate_instance>
         <,<gate_instance..>>;



Gate Types Component
Gates
Allows 
Strengths
and, nand, or, nor, xor, xnor, buf, not
Tristate
Drivers
Allows 
Strengths
bufif0, bufif1, notif0, notif1
MOS Switches
No Strengths
nmos, pmos, cmos, rnmos, rpmos, rcmos
Bidirectional
Switch
No strengths, 
non resistive
tran, tranif0, tranif1
Bidirectional
Switch
No strengths, 
resistive
rtran, rtranif0, rtranif1
Bidirectional
Switch
Allows strengths
pullup, pulldown


Gates, switch types and their instantiations

cmos i1 (out, datain, ncontrol, pcontrol)
nmos i2 (out, datain, ncontrol)
pmos i3 (out, datain, pcontrol)
pullup (neta) (netb);
pulldown (netc);
nor i4 (out, in1, in2, ...);
and i5 (out, in1, in2, ...);
nand i6 (out, in1, in2, ...);
buf i7 (out1, out2, in);
bufif1 i8 (out, in, control);
tranif1 i9 (inout1, inout2, control);


Gate level instantiation exmaple

// Gate level instantiation
nor (highz1, strong0) #(2:3:5) (out, in1, in2);
// instantiates a nor gate with out strength of highz1 (for 1)
// and strong0 for 0 #(2:3:5) is the min:typ:max delay

pullup1 (strong1) net1;
//instantiates a logic high pullup
cmos (out, data, ncontol, pcontrol);
// MOS devices

The following strength definitions exist:

The drive strengths for each of the output signals are:

Drive Strengths:

Logic 0

Logic 1

Strength
supply0 Su0 supply1 Su1 7
strong0 St0 strong1 St1 6
pull0 Pu0 pull1 Pu1 5
large La0 large La1 4
weak0 We0 weak1 We1 3
medium Me0 medium Me1 2
small Sm0 small Sm1 1
highz0 HiZ0 highz1 HiZ1 0





12.1 Gate Delays

The delays allow for modelling of rise time, fall time and turn off delays for the gates. Each of these delay types may be in the min:typ:max format. The order of the delays are #(trise, tfall, tturn-off).

For example:

nand #(6:7:8, 5:6:7, 12:16:19) (out, a, b); 


Delay

Model

#(delay) min:typ:max delay
#(delay,delay) rise time delay, fall time delay each delay can be with min:typ:max
#(delay,delay,delay) rise time delay, fall time delay and turn off delay each delay can be with
min:typ:max




For trireg, the delay of the capacitive network is modelled using the rise time delay, fall time delay and
charge decay. For example,

trireg (large) #(0,1,9) capacitor
//charge strength is large
// decay with tr = 0, tf = 1, tdecay = 9





13.0 Specify Blocks

A specify block is used to specify timing information for the module in which the specify block is used. Specparams are used to declare delay constants, much like regular parameters inside a module, but unlike module parameters, they cannot be overridden. Paths are used to declare time delays between inputs and outputs.

Timing Information using specify blocks

specify // similar to defparam, used for timing 
   specparam delay1 = 25.0, delay2 = 24.0;
 
//edge sensitive delays – some simulators do not support this
(posedge clock) => (out1 +: in1) = delay1, delay2);

//conditional delays
if (OPCDE = 3’h4) (in1, in2 *> out1) = (delay1, delay2);

// +: implies edge sensitive positive polarity
// -: implies edge sensitive -ve polarity
// *> implies multiple paths

// level sensitive delays
if (clock) (in1, in2 *> out1, out2) = 30
// setuphols
   $setuphols(posedge clock &&& reset, in1 &&& reset,
    3:5:6, 2:3:6);
   (reset *> out1, out2) = (2:3:5,3:4:5);

endspecify





Verilog Synthesis Constructs

The following is a set of Verilog Constructs that are supported by most synthesis tools at the time of
this writing. To prevent variations in supported synthesis constructs from tool to tool, this is the least
common denominator of supported constructs. Tool reference guides cover specific constructs.

14.0 Verilog Synthesis Constructs:

Since it is very difficult fot the synthesis tool to find hardware with exact delays, all absolute and
relative timing declarations are ignored by the tools. Also, all signals are assumed to be of maximum
strength (strength 7). Boolean operations on x and z are not permitted. The constructs are classified as




14.1 Fully supported constructs:

<module instantiation, with named and positional notations>

<integer data types, with all bases>

<identifiers>

<subranges and slices on right hand side of assignment>

<continuous assignment>

>>,<<,?:,{}
assign (procedural and declarative), begin, end, case, casex, casez, endcase
default
disable
function, endfunction
if, else, else if
input, output, inout
wire, wand, wor, tri
integer, reg
macromodule, module
parameter
supply0, supply1
task, endtask




14.2 Partially Supported Constructs

Construct

Constraints

*,/,%

when both operands constants or second operand
is a power of 2

always

only edge triggered events

for

bounded by static variables: only ise + or - to index

posedge, negedge

only with always @

primitive, endprimitive, table, endtable

Combinational and edge sensitive user defined
primitives are often supported.

<=

limitations on usage with blocking statement

and,nand,or,nor,xor,xnor,buf,not,,bufif0,bufif1,not
if0,notif1

Gate types supported without X or Z constructs

!, &&, ||, ~, &, |, ^, ^~, ~^, ~&, ~|, +, -, <, >, <=,
>=, ++, !=

Operators supported without X or Z constructs





14.3 Ignored Constructs

<intra assignment timing controls>
<delay specifications>
scalared, vectored
small medium large
specify
time (some tools treat these as integers)
weak1, weak0, highz0, highz1, pull0, pull1
$keyword (some tools use these to set synthesis constraints)
wait (some tools support wait with a bounded condition).


14.4 Unsupposrted constructs

<assignment with variable used as bit select on LHS of assignment>
<global variables>
===, !==
cmos,nmos,rcmos,rnmos,pmos,rpmos
deassign
defparam
event
force
fork,join
forever,while
initial
pullup,pulldown
release
repeat
rtran,tran,tranif0
tranif1
rtranif0,rtranif1
table,endtable,primitive,endprimitive




Index

Symbols


$display, $write
$fdisplay, $fwrite
$finish
$getpattern
$history
$hold, $width
$monitor, $fmonitor
$readmemb, $readmemh
$save, $restart, $incsave
$scale
$scope, $showscope
$setup
$setuphold
$showvars
$sreadmemb, $sreadmemh
$stop
$strobe, $fstrobe
$time, $realtime
/* */
//
autoexpand_vectornets
‘celldefine, ‘endcelldefine
‘defualt_nettype
‘define’expand_vectornets
‘noexpand_vectornets
‘ifdef, ‘else, ‘endif
‘include
nounconnected_drive
‘protect, ‘endprotect
‘protected, ‘endprotected
‘remove_gatename
‘noremove_gatenames
‘remove_netname
‘noremove_netnames
‘resetall
‘signed
‘unsigned
‘timescale
‘unconnected_drive

A
Arithemtic Operators

Arrays

B
Binary Expressions
blocking assignment

C
case
casex
casez
compiler directives

Combinational UDP example
continuous assignment

D
delays
disable

E
equality operators
escaped identifiers
expressions

F
for
forever
fork ... join
Fully supported synthesis constructs
function and function example


G
gate declaration
gate-types

I
if, if ... else
integer literals
identity operators

Ignored Synthesis Constructs

L
logical opeators

M
memories
module

N
named blocks
nets
non-blocking assigments

numbers


O
operator precedence
operators

P
partially supported synthesis constructs
procedural assignments
pulldown
pullup

R
reg, register
relational operators
repeat
reserved words

S
scalared
sequential edge sensitive UDP
sequential level sensitive UDP
Shift, other operators

shm_open

shm_probe
specify block
specparam
string symbols
supply0
supply1
switch types
synthesis constructs
synthesis ignored constructs
synthesis unsupported constructs

T
task and task example
tri0
tri1
triand
trior
trireg

U
UDP
unary expression
unary, bitwise and reduction operators

Unsupported Synthesis Constructs

V
vectored

W

wait
wand
while
wire
wor

X
x, X

Z
z, Z