Digital ASIC Design

4. Design With Verilog

Dr. Paul D. Franzon
Outline
1. Procedural Examples
2. Continuous Assignment Always Design Before Coding
3. Structural Verilog
4. Common Problems
5. More sophisticated examples

©2013, Dr. Paul D. Franzon, www.ece.ncsu.edu/erl/faculty/paulf.html 1

 Digital ASIC Design

4.2 Operators, Continuous Assignment, and

Structural Verilog

Dr. Paul D. Franzon

Outline
1. Operators
2. Continuous Assignment
3. Structural Verilog

©2013, Dr. Paul D. Franzon, www.ece.ncsu.edu/erl/faculty/paulf.html 2

 Digital ASIC Design

Objectives and Motivation

Objectives:
 Identify the functions captured by the different operators available in
Verilog.
 Understand how continuous assignment can be used to specify logic and
wire arrangements.
 Understand how module instancing is used to specify a netlist connecting
modules together.

Motivation:
 Enrichen your knowledge of synthesizable Verilog that can be described
using continuous assignment and structure
 Understand the operators that can be used in procedural code as well

©2013, Dr. Paul D. Franzon, www.ece.ncsu.edu/erl/faculty/paulf.html 3

 Digital ASIC Design

References
 Ciletti:
 Inside front cover : Summary
 Sections 6.4: Synthesis of tri-states
 Appendix C : Verilog Data Types
 Appendix D: Operators
 Smith and Franzon
 Sections 2.1, 2.2 : Datatypes
 Chapter 3: Structural code and continuous assignment
 Sutherland Reference Guide

©2013, Dr. Paul D. Franzon, www.ece.ncsu.edu/erl/faculty/paulf.html 4

 Digital ASIC Design

Arithmetic Operators

Usage Meaning Comments

+ A+B add
- A-B subtract
- -A negate
* A*B multiply Large parallel multiplier

/ A/B divide Not synthesizable

% A%B modulus Not synthesizable
** A ** B A to the power B Not synthesizable
Verilog 2001

e.g. B = -A; Forms the two’s complement of A and stores it in B

©2013, Dr. Paul D. Franzon, www.ece.ncsu.edu/erl/faculty/paulf.html 5

 Digital ASIC Design

Bitwise operators
Operate vectors bit-by-bit

Usage Description
~ ~A Invert each bit of A
& A&B AND each bit of A with B
| A|B OR each bit of A with B
^ A^B XOR each bit of A with B
~^ A ~^ B XNOR each but of A with B

reg [1:0] A, B, C, D;
assign C = A & B;
A[0] C[0]
B[0] A
C
A[1] B
C[1]
B[1]

©2013, Dr. Paul D. Franzon, www.ece.ncsu.edu/erl/faculty/paulf.html 6

 Digital ASIC Design

Unary Reduction Operators

 Operate on all the bits of the vector to produce a unary (one bit) result

Usage Description
& &A AND all bits of A together
~& ~&A NAND all bits of A together
| |A OR all bits of A together
~| ~| NOR all bits of A together
^ ^A XOR all bits of A together
~^ or ^~ ~^A or ^~A XNOR all bits of A together

reg [1:0] B;
B[0]
wire C; C
B[1]
assign C = & B;

©2013, Dr. Paul D. Franzon, www.ece.ncsu.edu/erl/faculty/paulf.html 7

 Digital ASIC Design

Logical Operators
 Take one-bit inputs and produce a 1-bit result
 Generally used in if-else and case statements but can be used for single
bit logic
 Conventional operators can also be used in if-else and case statements IF
they produce a 1-bit result

Usage Description
! !A A is not true
&& A&&B A and B are both true
|| A||B A or B are true

Examples:
reg A, B;
reg [1:0] A, B;
if (A&&B)
if (A&&B) // NOT OK
if (A&B) // also OK

©2013, Dr. Paul D. Franzon, www.ece.ncsu.edu/erl/faculty/paulf.html 8

 Digital ASIC Design

Equality and Identity Operators

 Also produce a 1-bit result
 Can be used in if-else, case statements or to create logic
Equality operators only compare logic 0 and 1
Usage Description
== A==B True if A is equal to B
!= A!=B True if A is not equal to B

Identity operators compare logic 0 and 1, x and z

Usage Description
=== A===B True if A is identical to B
!== A!==B True if A is not identical to B

Example: A=4’b00xx; B=5’b100xx

C=(A===B[4:0]); // C=1
D=(A==B[4:0]); // D=x : x returned if z or x in operators
©2013, Dr. Paul D. Franzon, www.ece.ncsu.edu/erl/faculty/paulf.html 9
 Digital ASIC Design

Relational Operators
 Also produce a 1-bit result:

Usage Description
< A<B True if A<B
> A>B True if A>B
<= A<=B True if A<=B
>= A>=B True if A>=B

©2013, Dr. Paul D. Franzon, www.ece.ncsu.edu/erl/faculty/paulf.html 10

 Digital ASIC Design

Shift
Logical Shift

Usage Description
<< A << B Shift A left by B bits, zero fill on right
>> A >> B Shift A right by B bits, zero fill on left

Arithmetic Shift (Verilog 2001)

Usage Description
<<< A <<< B Shift A left by B bits, zero fill on right
>>> A >>> B Shift A right by B bits, sign fill on left if A is
signed, zero filled if not

Note: Shifting by a fixed amount just rearranges wires

Shifting by a variable amount creates a shifter in combinational logic.

©2013, Dr. Paul D. Franzon, www.ece.ncsu.edu/erl/faculty/paulf.html 11

 Digital ASIC Design

Miscellaneous Operators

Usage Description
?: sel ? A : B If sel, choose A else choose B
(similar to if else)
{} {A,B} Concatenate A and B, creating a longer
vector
{{}} {A{B}} Replicate B A times, creating a longer
vector
-> ->A Trigger
Not synthesizable

©2013, Dr. Paul D. Franzon, www.ece.ncsu.edu/erl/faculty/paulf.html 12

 Digital ASIC Design

Comments on Operations
Operator Expansion
 Operators expanded to the largest element in the expression, before
evaluation
 Unsigned vectors are zero filled

 Signed vectors are sign bit filled

 {} and {{{} are done BEFORE expansion, and then treated as an operator at the
new width
 If assigned to a shorter variable, high order bits are lost
Ex: wire [3:0] A, B, C;
assign C = ({A,A} >> 2) | B[1:0]
//C = lower 4 bits of (8 bits | 0000 00B[1]B[0])
Signed Arithmetic (Verilog 2001)
 If any operand is unsigned, arithmetic is unsigned
 If ALL operators are signed, signed arithmetic is conducted
 Operator can be cast as $signed or $unsigned

©2013, Dr. Paul D. Franzon, www.ece.ncsu.edu/erl/faculty/paulf.html 13

 Digital ASIC Design

Operator Precedence ()
~ ! – (unary)
Order in which operators are evaluated
{} {{}}
**
For example, in
* / %
+-
A = A==B ? {C,D} : {2{E}} | F^G;
>> << >>> <<<
< <= > >=
Order evaluation:
== !== === !===
1. {2{E}} {C,D}
2. F^G & ~&

3. A==B ^ ~^

4. | | ~|

5. ?: &&

No () needed in this example ||

?:

©2013, Dr. Paul D. Franzon, www.ece.ncsu.edu/erl/faculty/paulf.html 14

 Digital ASIC Design

Examples
reg [1:0] A,B,C; // Q1: If A = 2’b10;B=2’b01; E = 3’b101;
reg [2:0] D,E,F; what is C, D, G, F, H after execution?

reg [3:0] H;
reg G;

always@(*) begin
C = A | E; // C=11
D = E << 2; // D=100
G = |A; // G=1
F = {A,B[0]}|E; // F=101
H = {2{B}}; // H=0101
end

©2013, Dr. Paul D. Franzon, www.ece.ncsu.edu/erl/faculty/paulf.html 15

 Digital ASIC Design

Examples
What is the logic being described (in detail)?
reg [1:0] A,B,C,D; A[1][0] B[1][0] E[2][1][0]
reg [2:0] E,F;
reg [3:0] H;
C[1]
reg G;
C[0]

D[2]
0 D[1]
0 D[0]
always@(*) begin
G
C = A | E;
D = E << 2; F[2]
G = |A;
F[1]
F = {A,B[0]}|E;
F[0]
H = {2{B}};
H[3]
end H[2]
H[1]
H[0]

©2013, Dr. Paul D. Franzon, www.ece.ncsu.edu/erl/faculty/paulf.html 16

 Digital ASIC Design

Examples
wire [2:0] E
wire [2:0] signed F; G[1] G[0]
wire [1:0] G, H;
A[0] 0 E[0]
// A, B 3-bits wide 0
A[1] E[1]
A[2] E[2]
assign E = A << 2;
assign F = B >>> 1;
B[0] F[0]
assign G = {2{A[0]}}; B[1] F[1]
assign H = {A[1],B[1]}; B[2] F[2]

H[1] H[0]

©2013, Dr. Paul D. Franzon, www.ece.ncsu.edu/erl/faculty/paulf.html 17

 Digital ASIC Design

Continuous Assignment
A[0]
C[0]
Sketch the logic being specified … B[0]

A[1]
input [3:0] A, B; C[1]
B[1]
wire [3:0] C, E;
wire D, F, G; A[2]
C[2]
assign C = A ^ B; B[2]
assign D = |A;
assign E = {{2{A[3]}}, A[2:1]}; A[3] C[3]
assign F = A[0] ? B[0] : B[1]; B[3]
assign G = (A == B);
A[0]
A[1] D
A[2]
A[3] B[0] A
F = G
B
B[1]
A[3] E[3]
A[3] E[2] A[0]
A[2] E[1]
A[1] E[0]

©2013, Dr. Paul D. Franzon, www.ece.ncsu.edu/erl/faculty/paulf.html 18

 Digital ASIC Design

Examples A[3] B[3]

A[2] B[2]
 Shifts, arithmetic, signed, unsigned A[1] B[1]
A[0] B[0]
0
Example:
A[3] C[3]
reg signed [3:0] A, B, C; A[2] C[2]
reg [3:0] D, E, F; A[1] C[1]
A[0] C[0]
always@(*)
begin
\\ assume for calculation that A=D=1001
B = A <<< 1; // B=0010
C = A >>> 1; // C=1100
E = D << 1; // E=0010
F = D >> 1; // F=0100
end

©2013, Dr. Paul D. Franzon, www.ece.ncsu.edu/erl/faculty/paulf.html 19

 Digital ASIC Design

Summary
 Verilog contains many useful operators
 Arithmetic +-*
 Logic &A ^A |A A&B A^B A|C etc.
 Relational === == > etc.
 Shift – logical and arithmetic
 Concatanation and replication – extending vectors
 ?: = if else
 Go to sub-module quiz and then to next sub-module

©2013, Dr. Paul D. Franzon, www.ece.ncsu.edu/erl/faculty/paulf.html 20

 Digital ASIC Design

Continuous Assignemt
 Directly specifies logical structure in one line of code
 Procedural code often specifies behavior of logic

wire [1:0] B, C, D;
tri [1:0] E;

assign D = B & C; All variables assigned in this way

assign E = B[0] ? C : D; must be type wire or tri

Statements executed whenever anything on RHS changes

(“continuously assigned”)

©2013, Dr. Paul D. Franzon, www.ece.ncsu.edu/erl/faculty/paulf.html 21

 Digital ASIC Design

Examples
wire [3:0] sum;
wire carry;
// A, B, C 4 bits wide
// D 1 bit wide

assign {carry,sum} = A + (D ? B : C);

// Note A+D ? B : C would be very different!

A + sum
B
C
carry
D

©2013, Dr. Paul D. Franzon, www.ece.ncsu.edu/erl/faculty/paulf.html 22

 Digital ASIC Design

Continuous Assignment
Sketch the logic being specified …

input A, B, C;

tri F;

assign F = A ? B : 1’bz;
assign F = ~A ? C : 1’bz;
A

Tri-State Buffer
•Connect to shared bus B

In general only use a shared bus as a last

C
resort (can be hard to debug).
F
(Exception : Some FGPAs have lots of tri-state buffers
- use as a mux.)
©2013, Dr. Paul D. Franzon, www.ece.ncsu.edu/erl/faculty/paulf.html 23
 Digital ASIC Design

Further Examples
wire [3:0] A, B, D; B: Rotator
C-bits
wire [1:0] C, E;
A[3]A[2]A[1]A[0]
assign B = {A,A} >> C;
A[3]A[2]A[1]A[0] A[3]A[2]A[1]A[0]
C bits
assign D = {4{A[3]},A} >> E; B is 4-bits

A[3]A[2]A[1]A[0] A[3]A[2]A[1]A[0]
D: Signed Right Shift

E Signed Right Shift C Rotator

wire [3:0] A; B[3]B[2]B[1]B[0]

D[3]D[2]D[1]D[0]
wire [1:0] B;
A[3]A[2]A[1]A[0]
wire C;
assign C = A[B]; B
C: Mux C
©2013, Dr. Paul D. Franzon, www.ece.ncsu.edu/erl/faculty/paulf.html 24
 Digital ASIC Design

Summary – Continuous Assignment

 Any one line expression is efficiently captured using continuous
assignment
 Tri-state muxes MUST be specified using continuous assignment
 All variables assigned continuously must be of type wire or tri

Go to sub-module quiz before proceeding to next module

©2013, Dr. Paul D. Franzon, www.ece.ncsu.edu/erl/faculty/paulf.html 25

 Digital ASIC Design

Structural Verilog
Complex modules can be put together by ‘building’ (instancing) a number of smaller
modules.
e.g. Given the 1-bit adder module with module definition as follows, build a 4-bit adder with
carry_in and carry_out
module OneBitAdder (CarryIn, In1, In2, Sum, CarryOut);
4-bit adder:
module FourBitAdder (Cin, A, B, Result, Cout);
input Cin;
input [3:0] A, B;
output [3:0] Result;
output Cout;
wire Cout;
wire [3:1] chain;
OneBitAdder u1 (.CarryIn(Cin), .In1(A[0]), .In2(B[0]),
.Sum(Result[0]), .CarryOut(chain[1]));
OneBitAdder u2 (.CarryIn(chain[1]), .In1(A[1]), .In2(B[1]),
.Sum(Result[1]), .CarryOut(chain[2]));
OneBitAdder u3 (.CarryIn(chain[2]), .In1(A[2]), .In2(B[2]),
.Sum(Result[2]), .CarryOut(chain[3]));
OneBitAdder u4 (Chain[3], A[3], B[3], Result[3], Cout); // in correct
order
endmodule

©2013, Dr. Paul D. Franzon, www.ece.ncsu.edu/erl/faculty/paulf.html 26

 Digital ASIC Design

Structural Example
 Sketch:

FourBitAdder
u1 u2
OneBitAdder OneBitAdder
chain[1]
Cin CarryIn Sum CarryIn Sum
A[0] In1 CarryOut In1 CarryOut
B[0] In2 In2

A[1]
B[1]
Result[0]
Result[1]

Etc.
©2013, Dr. Paul D. Franzon, www.ece.ncsu.edu/erl/faculty/paulf.html 27
 Digital ASIC Design

Instance Formats
 Module instance formats

// Simple format in which variable order matters

OneBitAdder u1 (Cin, A[0], B[0], Result[0], chain[1]);

Same order as module port order

// Advanced format in which variable order does not matter

OneBitAdder u2 (.CarryIn(chain[1]), .In1(A[1]), .In2(B[1]),
.Sum(Result[1]), .CarryOut(chain[2]);

Name of port in module

Name of variable in
OneBitAdder
module FourBitAdder

©2013, Dr. Paul D. Franzon, www.ece.ncsu.edu/erl/faculty/paulf.html 28

 Digital ASIC Design

Structural Verilog
Features:
Four copies of the same module (OneBitAdder) are built (‘instanced’) each
with a unique name (u1, u2, u3, u4).
Module instance syntax:
OneBitAdder u1 (.CarryIn(Cin),

Module Name Instance Name Port Name inside Module (optional)

Net name

All nets connecting to outputs of modules must be of wire type (wire or tri):
wire [3:1] chain;

(Note: Illustrative only, NOT a good way to build an adder)

©2013, Dr. Paul D. Franzon, www.ece.ncsu.edu/erl/faculty/paulf.html 29

 Digital ASIC Design

Applications of Structural Verilog

 To Assemble modules together in a hierarchical design.
 Final gate set written out in this format (“netlist”).
 Design has to be implemented as a module in order to integrate with the
test fixture

Hiearchy and Scope:

 Implements hierarchy
 Copies of OneBitAdder are instanced inside the module FourBitAdder
 Variable scope
 CarryIn: Scope is inside OneBitAdder Module
 chain[1]: Scope is inside FourBitAdder Module
 u2.CarryIn: Allows CarryIn in module u2 to be referenced from
FourBitAdder (useful in traversing hierarchy in simulator)

©2013, Dr. Paul D. Franzon, www.ece.ncsu.edu/erl/faculty/paulf.html 30

 Digital ASIC Design

No glue logic!
 Generally it is a good idea to only implement logic in the leaf cells of a
hierarchical design, and not at a higher level
 i.e.
module good(A,B,C); module bad(A,B,C);
good_leaf u1(A,B); assign D=C&D;
not
good_leaf u2(A,C); good_leaf u1(A,B);
endmodule good_leaf u2(A,C);
endmodule

 Why?
 Hint: Consider what module must be synthesized in a single run

good_leaf can be synthesized once and replicated in good

In bad, bad must be synthesized including all logic (longer synthesis run)

 Note: See Hieararchy notes for more on partitioning

©2013, Dr. Paul D. Franzon, www.ece.ncsu.edu/erl/faculty/paulf.html 31

 Digital ASIC Design

Sample Netlist
module counter ( clock, in, latch, dec, zero );
input [3:0] in;
input clock, latch, dec;
output zero;
wire \value[3] , \value[1] , \value53[2] , \value53[0] , \n54[0] ,
\value[2] , \value[0] , \value53[1] , \value53[3] , n103, n104, n105,
n106, n107, n108, n109, n110, n111, n112, n113, n114, n115;
NOR2 U36 ( .Y(n107), .A0(n109), .A1(\value[2] ) );
NAND2 U37 ( .Y(n109), .A0(n105), .A1(n103) );
NAND2 U38 ( .Y(n114), .A0(\value[1] ), .A1(\value[0] ) );
NOR2 U39 ( .Y(n115), .A0(\value[3] ), .A1(\value[2] ) );
XOR2 U40 ( .Y(n110), .A0(\value[2] ), .A1(n108) );
NAND2 U41 ( .Y(n113), .A0(n109), .A1(n114) );
INV U42 ( .Y(\n54[0] ), .A(n106) );
INV U43 ( .Y(n108), .A(n109) );
AOI21 U44 ( .Y(n106), .A0(n112), .A1(dec), .B0(latch) );
INV U45 ( .Y(zero), .A(n112) );
NAND2 U46 ( .Y(n112), .A0(n115), .A1(n108) );
OAI21 U47 ( .Y(n111), .A0(n107), .A1(n104), .B0(n112) );
DSEL2 U48 ( .Y(\value53[3] ), .D0(n111), .D1(in[3]), .S0(latch) );
DSEL2 U49 ( .Y(\value53[2] ), .D0(n110), .D1(in[2]), .S0(latch) );
DSEL2 U50 ( .Y(\value53[1] ), .D0(n113), .D1(in[1]), .S0(latch) );
DSEL2 U51 ( .Y(\value53[0] ), .D0(n105), .D1(in[0]), .S0(latch) );
EDFF \value_reg[3] ( .Q(\value[3] ), .QBAR(n104), .CP(clock), .D(
\value53[3] ), .E(\n54[0] ) );
EDFF \value_reg[2] ( .Q(\value[2] ), .CP(clock), .D(\value53[2] ), .E(
\n54[0] ) );
EDFF \value_reg[1] ( .Q(\value[1] ), .QBAR(n103), .CP(clock), .D(
\value53[1] ), .E(\n54[0] ) );
EDFF \value_reg[0] ( .Q(\value[0] ), .QBAR(n105), .CP(clock), .D(
\value53[0] ), .E(\n54[0] ) );
endmodule

©2013, Dr. Paul D. Franzon, www.ece.ncsu.edu/erl/faculty/paulf.html 32

 Digital ASIC Design

Variable type rules

module Test (A,B,C,D,E) Inside module:
in A;
out B, C, D; Inputs implicitly a net data type (e.g. wire)
inout E; (Not usually an issue in synthesizable design
as should not be modified in module)
wire B, D;
reg C; Outputs neither net or register type
tri E; (Must be declared based on how assigned)

always@(A) C=A; Inout can NOT be a register type

assign B=A (a tri declaration makes most sense in
assign E = A ? 1’bz : D; a synthesizable design)

Nested u1 (D); Outputs of instanced modules are of type wire

endmodule
Note: Variables inside module Nested are DIFFERENT variables, even if I give them
the same name

©2013, Dr. Paul D. Franzon, www.ece.ncsu.edu/erl/faculty/paulf.html 33

 Digital ASIC Design

Parameter
 Parameter is like “constant” in C

parameter bit_width = 8;

reg [bit_width-1 : 0] Register;

©2013, Dr. Paul D. Franzon, www.ece.ncsu.edu/erl/faculty/paulf.html 34

 Digital ASIC Design

In Verilog 2001…
 Parameter Passing
 Can pass parameters to modules, overwriting their local values
module top;
Parameter RFsize1 = 64;
Parameter AddressSize1 = 6;

RegFile #(.size(RFsize1), .Asize(AddressSize1) U1 ( … );

endmodule

module RegFile( …
input [Asize-1:0] WriteAddress, ReadAddress,
);
parameter Asize=5;
parameter size=32;
reg [15:0] Register [0:size-1];

©2013, Dr. Paul D. Franzon, www.ece.ncsu.edu/erl/faculty/paulf.html 35

 Digital ASIC Design

Exercise
module mod2 (G,H,I)
Sketch Design, including hierarchy;
Declare needed variables and ports input [1:0] G;
input H; output [1:0] I;
module Ex1 (A, B, C); wire [1:0] I;
input [1:0] A; assign I = G & {2{H}};
output B; output [1:0] C; endmodule
wire B; wire [1:0] C; u1
mod1 u1 (A,B); A[0]
B
mod2 u2 (A,B,C);
endmodule A[1]
module mod1 (E,F)
input [1:0] E;
u2
output F; A[0] C[0]
wire F;
assign F = &E; A[1]
C[1]
endmodule

©2013, Dr. Paul D. Franzon, www.ece.ncsu.edu/erl/faculty/paulf.html 36

 Digital ASIC Design

Summary – Structural Verilog

 Structural Verilog permits instanced modules to be wired together in a
netlist
 The outputs of all modules must be declared as type wire within the
module they are instanced
 Parameter passing is useful to make resizable designs whose size can be
defined from the top level of the hierarchy

 Goto sub-module quiz and next sub-module in next unit

©2013, Dr. Paul D. Franzon, www.ece.ncsu.edu/erl/faculty/paulf.html 37