SystemVerilog

Ming-Hwa Wang, Ph.D.

COEN 207 SoC (System-on-Chip) Verification
Department of Computer Engineering
Santa Clara University
Introduction
SystemVerilog is a standard (IEEE std 1800-2005) unified hardware design,
specification, and verification language, which provides a set of extensions to
the IEEE 1364 Verilog HDL:

design specification method for both abstract and detailed specifications

embedded assertions language and application programming interface

(API) for coverage and assertions

testbench language based on manual and automatic methodologies

direct programming interface (DPI)

Purpose: provide a standard which improves productivity, readability, and
reusability of Verilog-based code, extends for higher level of abstraction for
system modeling and verification, provides extensive support for directed
and
constrained-random
testbench
development,
coverage-driven
verification, and formal assertion-based verification

Extensions to Verilog

extended data types

C data types: int, typedef, struct, union, enum

other data types: bounded queues, logic (0, 1, X, Z) and bit (0, 1),
tagged unions

dynamic data types: string, class, dynamic queues, dynamic arrays,

associated arrays including automatic memory management

dynamic casting and bit-stream casting

automatic/static specification on per-variable-instance basis

extended operators

wild equality and inequality

built-in methods to extend the language

operator overloading

streaming operators

set membership

extended procedural statements

pattern matching on selection statements

loop statements

C-like jump statements: return, break, continue

final blocks that execute at the end of simulation (inverse of initial)

extended event control and sequence events

extended process control

extensions to always blocks to include synthesis consistent

simulation semantics

extensions to fork join to model pipelines

fine-gram process control

extended tasks and functions

C-like void functions

pass by reference

default arguments

argument binding by name

optional arguments

import/export function for DPI

classes: object-oriented mechanism (abstraction, encapsulation, safe
pointers)
automatic testbench support with random constraints
interprocess communication synchronization

semaphores

mailboxes

event extension, event variables, and event sequencing

clarification and extension of the scheduling semantics
cycle-based functionality: clocking blocks and cycle-based attributes

cycle-based signal drives and samples

synchronous samples

race-free program context

assertion mechanism

property and sequence declarations

assertions and coverage statements with action blocks

extended hierarchy support

packages for declaration encapsulation with import for controlled

access

compilation-unit scope nested modules and extern modules for

separation compilation support

extension of port declarations to support interfaces, events, and

variables

$root to provide unambiguous access using hierarchical references

interfaces to encapsulate communication and facilitate communicationoriented design
functional coverage
DPI for clean, efficient interoperation with other languages (C provided)
assertion API
coverage API
data read API
Verilog procedure interface (VPI) extension for SystemVerilog constructs
concurrent assertion formal semantics

Extended Literal Values

integer literals and logic literals

unsized literal with a preceding apostrophe (), e.g.,

0, 1, X, x, Z, z // sets all bits to this value

real literals: fixed-point format or exponent format

time literals: s, ms, us, ns, ps, fs, step

string literals enclosed in quotes: \v for vertical tab, \f for form feed,
\a for bell, \x02 for hex number

array literals: replicate operator ( {{}} ), index/type keys and default

values, e.g.,
int n[1:2][1:3] = {{0,1,2}, {3{4}}};
int m[1,2][1,6] = '{2{'{3{4,5}}}};
// same as '{'{4,5,4,5,4,5},'{4,5,4,5,4,5}}
typedef int triple [1:3]; $mydisplay(triple'{0,1,2});
triple b = '{1:1, default:0}; // indexes 2 and 3 assigned 0
structure literals, e.g.,
typedef struct {int a; shortreal b;} ab;
ab c = '{0, 0.0}; /* same as c = '{a:0, b:0.0}; c =
'{default:0}; or c = ab'{int:0, shortreal:0.0}; */
ab abarr[1:0] = '{'{1, 1.0}, '{2, 2.0}};
typedef struct {int a,b[4];} ab_t;
int a,b,c;
ab_t v1[1:0] [2:0] = '{2{'{3{a,'{2{b,c}}}}}};
// expands to '{'{3{'{a,{2{b,c}}}}}, '{3{{a,'{2{b,c}}}}}}
/* expands to
'{'{'{a,'{2{b,c}}},'{a,'{2{b,c}}},'{a,'{2{b,c}}}},
'{'{a,'{2{b,c}}},'{a,'{2{b,c}}},'{a,'{2{b,c}}} } } */
/* expands to
'{'{'{a,'{b,c,b,c}},'{a,'{b,c,b,c}},'{a,'{b,c,b,c}}},
'{'{a,'{b,c,b,c}},'{a,'{b,c,b,c}},'{a,'{b,c,b,c}}}} */

Data Types
A data type is a set of values and a set of operations that can be performed
on those values. Data types can be used to declare data objects or to define
user-defined data types that are constructed from other data types.

integer types

2-state - can simulate faster and take less memory: shortint (16-bit
signed), int (32-bit signed), longint (64-bit signed), byte (8-bit
signed or ASCII character), bit (unsigned with user-defined vector
size)

4-state - can have unknown (x) and high-impedance (z) values:

logic (unsigned with user-defined vector size), reg (unsigned with
user-defined vector size), integer (32-bit signed), time (64-bit
unsigned)

integral types - the data types that can represent a single basic integer
data type: packed array, packed struct, packed union, enum, time.
A simple bit vector type is the data types that can directly represent a
one-dimensional packed array of bits.

real types: real (64-bit signed), shortreal (32-bit signed)

void data type - for function returns nothing or represent nonexistent

data

chandle data type - for storing pointers passed using DPI (default null)

only allow the following operation with another chandle variable or

null or Boolean values: equality (==), inequality (!=), case equality
(===), case inequality (!==)

only allow assignment from another chandle or null

chandles can be inserted into associative arrays, can be used within

a class, can be passed as arguments to functions or tasks, and can
be returned from functions

chandles shall not be assigned to variables of any other type, shall

not be used as follows: as ports, in sensitivity lists or event
expressions, in continuous assignments, in untagged unions, in
packed types
string: variable length array of bytes indexed from 0

string operators: ==, !=, <, <=, >, >=, {str1, str2, , strN),
{multiplier{str}}, str[index], string.method()

str.method(): len( ), putc(int index, byte c), getc(int index),

toupper( ), tolower( ), compare(string s), icompare(string s),
substr(int i, int j), atoi( ), atohex( ), atooct( ), atobin( ), atoreal( ),
itoa(integer i), hextoa(integer i), atoreal( ), octtoa(integer i),
bintoa(integer i), realtoa(real r)
event data type: event variables can be explicitly triggered and

waited for

syntax: event <var_name> [= (<initial_value> | null) ];

user-defined types: typedef (forward definition and actual definition)
enumeration data types with strong type checking

methods: first( ), last( ), next(int unsigned i=1), prev(int unsigned

i=1), num( ), name(int unsigned i)
structures and unions

packed and unpacked, signed and unsigned, 2-state and 4-state

a tagged union saves a value and a tag (or a member name) for
strong type access checking
class is declared using the class endclass keywords

class properties

methods
casting: a data type can be changed by using a cast () operation

static casting

<type> ( <expression> ) or <type> { { <literal>, ,

<literal> } }

<size> ( <expression> )

signed (<expression> )

$shortrealtobits, $bitstoshortreal, $bits, $itor, $rtoi, $bitstoreal,

$realtobits, $signed, $unsigned

dynamic casting: $cast

bit-stream casting

for converting between different aggregate types

example
uses bit-stream casting to model a control packet transfer over a
data stream:
typedef struct {
shortint address;
reg [3:0] code;
byte command [2];
} Control;
typedef bit Bits [36:1];

Control p;
Bits stream[$];
p = ...
// initialize control packet
// append packet to unpacked queue of bits
stream = {stream, Bits'(p)}
Control q;
// convert stream back to a Control packet
q = Control'(stream[0]);
stream = stream[1:$]; // remove packet from stream
uses bit-stream casting to model a data packet transfer over a byte
stream:
typedef struct {
byte length;
shortint address;
byte payload[];
byte chksum;
} Packet;
function Packet genPkt();
Packet p;
void'( randomize( p.address, p.length, p.payload )
with { p.length > 1 && p.payload.size == p.length; }
);
p.chksum = p.payload.xor();
return p;
endfunction
The byte stream is modeled using a queue, and a bit-stream cast is
used to send the packet over the stream.
typedef byte channel_type[$];
channel_type channel;
channel = {channel, channel_type'(genPkt())};
And the code to receive the packet:
Packet p;
int size;
size = channel[0] + 4;
// convert stream to Packet
p = Packet'( channel[0 : size - 1] );
// remove packet data from stream
channel = channel[ size, $ ];
Arrays

in Verilog, all data types can be declared as arrays

a dimension declared before the object name is referred to as the vector

width dimension, and the dimensions declared after the object name are
referred to as the array dimensions

SystemVerilog uses the term packed array to refer to the dimensions

declared before the object name, and the term unpacked array is used
to refer to the dimensions declared after the object name; a packed
array is guaranteed to be represented as a contiguous set of bits, and an
unpacked array may or may not be so represented

multi-dimensional arrays

uses part-select to refer to a selection of one or more contiguous bits of

a single dimension packed array, use slice to refer to a selection of one
or more contiguous elements of an array
array querying functions: $left, $right, $low, $high, $increment, $size,
$dimensions, and $unpacked_dimensions
dynamic arrays: any dimension of an unpacked array whose size can be
set or changed at run time

new [ expression ] [ ( expression ) ]

size( )

delete( )
array assignment between fixed-size arrays and dynamic arrays
arrays as arguments: pass by value
associative arrays

indexing operator: wildcard index type *, string index, class index,

integer or int index, signed packed array index, unsigned packed
array index, packed struct index, user-defined type index

methods: num( ), delete( [input index] ), exists(input index),

first(ref index), last(ref index), next(ref index), prev(ref index)

associative array assignment

associative arrays are passed as arguments

associative array literals use the {index:value} syntax, index can

be default
queues with position 0 represents the first element, and $ represent the
last

queues are declared using the same syntax as unpacked arrays, but
specifying $ as the array size

empty queue { }

right bound [$:N], where N+1 is the maximum size of the queue

operators: indexing, concatenation, slicing, equality

methods: size( ), insert(input int index, input type item),

delete(int index), pop_front( ), pop_back( ), push_front(input type
item), push_back(input type item),
array manipulation methods

syntax: expr.array_method { attribute_instance } [ ( arguments ) ]

[ with ( expr ) ]

array locator methods: find( ), find_index( ), find_first( ),

find_first_index( ), find_last( ), find_last_index( ), min( ), max( ),
unique( 0, unique_index( )

array ordering methods: reverse( ), sort( ), rsort( ), shuffle( )

array reduction methods: sum( ), product( ), and( ), or( ), xor( )

iterator index querying: index( )

Data Declarations

data have to be declared before they are used, except implicit nets

forms of data: literals, parameters, constants (genvars parameters,

localparams, specparams), variables (static or dynamic), nets (reg, wire,
logic), attributes

constants

3 constructs for defining elaboration-time constants: the

parameter, localparam and specparam; the default value of a
parameter of an instantiated module can be overridden in each
instance of the module using one of the following:

implicit in-line parameter redefinition (e.g., foo #(value, value)

u1 (...); )

explicit in-line parameter redefinition (e.g., foo #(.name(value),

.name(value)) u1 (...); )

defparam statements, using hierarchical path names to redefine

each parameter

value parameters: a module, interface, program, or class can have

parameters, which are set during elaboration and are constant
during simulation

$ as a parameter value to represent unbounded range specification

$isunbounded(const_expr)

type parameters: parameter type

parameter port lists: the parameter keyword can be omitted, and a

parameter can be depend on earlier parameter

const constants are set during simulation

variables declared with var have default type of logic
type
default
4-state integral
X
2-state integral
0
real, shortreal
0.0
enumeration
base type default initial value
string
(empty string)
event
new event
class
null
chandle (opaque handle) null
nets: trireg, wire, reg

a net can be written by continuous assignments, by primitive output,

or through module port

assign, force, release

scope and lifetime: automatic and static

global and static, local and static, local and automatic

signal aliasing: the members of an alias list are signals whose bits share
the same physical nets, aliasing is performed at elaboration time and
cant be undone
type compatibility: 5 levels

matching type: typedef, anonymous enum, struct, or union

equivalent type

assignment compatible: have implicit casting rules

cast compatible: have explicit casting rules

nonequivalent or type incompatible

type operator

Classes

class properties and methods

constructor new( )
static class properties shared by all instances of the class using static
static class method with automatic variable lifetime: static task foo( );
end task
nonstatic class method with static variable lifetime: task static foo( );
end task
shallow copy (putting an object after new) v.s. deep copy (custom code
is typically needed)
this and super
it is always legal to assign a subclass variable to a variable of a class
higher in the inheritance tree, but it is never legal to directly assign a
superclass variable to a variable of one of its subclasses; it is legal to
assign a superclass handle to a subclass variable if the superclass handle
refers to an object of the given subclass, and use $cast( ) to check
whether the assignment is legal
unqualified (public), local (private), and protected
const: read-only

global constants with initial values (optionally with static), instance

constants without
abstract class using virtual
polymorphism
class scope resolution operator ::
the extern qualifier for out-of-block declarations
parameterized classes

the combination of a generic class and the actual parameter values

is called a specialization (or variant)
typedef class: for cross-referencing
memory management: automatic garbage collection

Operators and Expressions

assignment_operator ::= = | += | -= | *= | /= | %= | &= | |= | ^=

| <<= | >>= | <<<= | >>>=

conditional_expression ::= cond_predicate ? { attribute_instance }

expression : expression

unary_operator ::= + | - | ! | ~ | & | ~& | | | ~| | ^ | ~^ | ^~

binary_operator ::= + | - | * | / | % | == | != | === | !== | ==? |

!=? | && | || | ** | < | <= | > | >= | & | | | ^ | ^~ | ~^ | >> | <<
| >>> | <<<

inc_or_dec_operator ::= ++ | -
unary_module_path_operator ::= ! | ~ | & | ~& | | | ~| | ^ | ~^ | ^~

binary_module_path_operator ::= == | != | && | || | & | | | ^ | ^~ |

~^

built-in package: std::

concatenation using braces ( { } ) or replication operator (multiple

concatenation)

assignment patterns for assigning struct fields and array elements using
( { } ) either by positions, by type:value, or by member:value or by
default:value

array assignment pattern, structure assignment pattern,

tagged union expression and member access: union tagged { }

aggregate expressions
operator overloading: bind op function type func_name ( formals )

match formal types exactly or the actual types are implicitly cast to
formal types

the operators that can be overloaded are the arithmetic operators,

the relational operators, and assignment
streaming operators (pack/unpack)

>> causes data to be streamed in left-to-right order

<< causes data to be streamed in right-to-left order

streaming dynamically sized data using with [expr[+|-]:expr]

conditional operator
set membership: expr inside { open_range_list }

observed

post-observed

reactive

re-inactive

Scheduling Semantics

evaluation event and simulation time

time wheel or time-ordered set of linked lists

a time slot is divided into a set of ordered regions as table below;

provide predictable interaction between the design and testbench
code (including PLI callbacks)

pre-postponed

from previous time slot

region
preponed

pre-active

active
inactive

pre-NBA

NBA
post-NBA

semantics
The
#1step
sampling
delay
provides the ability to sample data
immediately before entering the
current time slot
allows PLI application routines
(cbAfterDelay,
cbNextSimTime,
cbAtStartOfSimTime) to read and
write values and create events
before events in the Active region
are evaluated
holds
current
events
being
evaluated
holds the events to be evaluated
after all the active events are
processed
allows PLI application routines
(cbNBASynch, cbReadWriteSynch)
to read and write values and
create events before the events in
the NBA region are evaluated
A nonblocking assignment creates
an event in this region
allows PLI application routines
(cbReadWriteSynch) to read and

postponed

note
IEEE1364

IEEE1364

IEEE1364
iterative
IEEE1364
iterative
IEEE1364
iterative

IEEE1364
iterative
IEEE1364
iterative

write values and create events

after the events in the NBA region
are evaluated
for the evaluation of the property
expressions
when
they
are
triggered
allows PLI application routines
(currently no PLI callback yet) to
read values after properties are
evaluated
property pass/fail code shall be
scheduled here of the current time
slot
a #0 control delay specified in a
program block schedules the
process for resumption in this
region
allows PLI application routines
(cbAtEndOfSimTime) to read and
write values and create events
after processing all other regions
except the Postponed region
cbReadOnlySynch
No new value changes are allowed
to happen in the time slot
to

iterative

iterative

iterative

iterative

IEEE1364
iterative

IEEE1364

next time slot

the SystemVerilog simulation reference algorithm

execute_simulation {
T = 0;
initialize the values of all nets and variables;
schedule all initialization events into time 0 slot;
while (some time slot is nonempty) {
move to the next future nonempty time slot and set T;
execute_time_slot (T);
}
}
execute_time_slot {
execute_region (preponed);
execute_region (pre-active);
while
(any
region
in
[active...pre-postponed]
is
nonempty) {
while (any region in [active...post-observed] is
nonempty) {
execute_region (active);
R
=
first
nonempty
region
in
[active...postobserved];
if (R is nonempty) move events in R to the active
region;
}

while (any region in [reactive...re-inactive] is

nonempty) {
execute_region (reactive);
R
=
first
nonempty
region
in
[reactive...reinactive];
if (R is nonempty)
move events in R to the reactive region;
}
if (all regions in [active re-inactive] are empty)
execute_region (pre-postponed);
}
execute_region (postponed);

}
execute_region {
while (region is nonempty) {
E = any event from region;
remove E from the region;
if (E is an update event) {
update the modified object;
evaluate processes sensitive to the object and
possibly schedulefurther events for execution;
} else { /* E is an evaluation event */
evaluate the process associated with the event and
possibly schedule further events for execution;
}
}
}
the PLI callback control points

2 kinds of PLI callbacks: those are executed immediately when some

specific activity occurs and those that are explicitly registered as a
one-shot evaluation event

Callbacks and their event region (in the table above)

Procedural Statements and Control Flow

procedural statements: initial, final, always, always_comb,

always_latch, always_ff, task, function

control flow

selection, loops, jumps

task and function calls

sequential and parallel blocks

timing control

blocking ( = ) and nonblocking ( <= ) assignments

selection statements: if, else if, else, case, casez, casex, default,
endcase, unique (for mutual exclusive and can be executed in
parallel), priority (ordered evaluation), inside (for set membership),
matches (using &&& in if statements), ?:

loop statements: forever, repeat, while, for, do while, foreach

jump statements: return, break, continue, disable

named blocks and statement labels: begin end, fork join, join_any,
join_none

event control: @, #, iff, posedge, negedge

sequence, triggered
level-sensitive sequence control: wait
sequence abc;
@(posedge clk) a ##1 b ##1 c;
endsequence
sequence de;
@(negedge clk) d ##[2:5] e;
endsequence
program check;
initial begin
wait( abc.triggered || de.triggered );
if( abc.triggered )
$display( "abc succeeded" );
if( de.triggered )
$display( "de succeeded" );
end
endprogram

Processes

New always blocks design intent is understood - IEEE1800 does not

specify which constructs are synthesizable and which are not, EDA
vendors implement differently and portability will be an issue

always_comb for modeling combinational logic behavior

inferred/implicit sensitivity list (within the block or within any

function called within the block), automatically executes once at time
zero, the variables written on the LHS of assignment shall not be
written to by another process

always_latch for modeling level triggered latch logic behavior

always_ff for modeling edge triggered synthesizable sequential logic

behavior
always_ff @(posedge clock iff reset == 0 or posedge reset)
begin
r1 <= reset ? 0 : r2 + 1;
...
end

fork join for creating concurrent processes

control option
description
join
the parent process blocks until all the processes
spawned by this fork complete
join_any
the parent process blocks until any one of the
processes spawned by this fork complete
join_none
the
parent
process
continues
to
execute
concurrently with all the processes spawned by this
fork, and the spawned processes do not start
executing until the parent thread executes a
blocking statement
Assertions

assertions specify behaviors of the system, and are primarily used to

validate the behavior of a design

provide functional coverage and generate input stimulus for

validation.
immediate assertions follow simulation event semantics and are
executed like a procedure statements

the immediate assertion statement is a test of an expression

performed when the statement is executed in the procedure code;
the expression is non-temporal and is interpreted the same way as
an expression in the condition of a procedural if statement

syntax:
assert
(
expression
)
[pass_statement]
[else
fail_statement]

if an assertion fails and no else clause is specified, the tool shall, by

default, call $error, unless a tool specific option, such as a
command-line option, is enabled to suppress the failure

there are 4 severity levels: $fatal, $error, $warning, and $info; all
the severity system tasks shall print a tool-specific message
indicating the severity of the failure and specific information about
the specific failure, which shall include the following information:

the file name and line number of the assertion statement

the hierarchical name of the assertion, if it is labeled, or the

scope of the assertion if it is not labeled
a concurrent assertion is based on clock semantics and is evaluated only
at the occurrence of a clock tick; the values of variables used in the
evaluation are the sampled values, thus a predictable result can be
obtained from the evaluation, regardless of the simulators internal
mechanism of ordering events and evaluating events

syntax: assert property ( expression ) [pass_statement] [else

fail_statement]