Fall 2008

Dr. Meeta Yadav

Prof. Paul Franzon

©2008, Dr. Meeta Yadav, www4.ncsu.edu/~myadav/Research 1

 Topics

• Why use randomization

• Randomization options
• Randomization of objects
• Class constraints and distributions
• In-line constraints and distributions

©2008, Dr. Meeta Yadav, www4.ncsu.edu/~myadav/Research 2

 Randomization Overview

Why Randomize?

 As designs grow it becomes more difficult to test their features

through directed test cases
 Directed testcases checks specific features of a design and can only
detect anticipated bugs
 Directed testcase approach is a time consuming exercise
 Directed testscases are unable to detect devious bugs
 Directed tests detect only anticipated bugs in the design

Solution is Constrained Random Testcases (CRT) which randomize the input stimulus

Random testing detects the bugs you did not expect in

your design

©2008, Dr. Meeta Yadav, www4.ncsu.edu/~myadav/Research 3

 What to Randomize?

What to Randomize?
• Device configurations
 Try different device configurations
 For instance for a switch verification try different configurations of
input and output ports

• Environment configurations
 Randomize the entire environment that the device operates in
 The number of objects
 Object configurations
 For instance randomize the number of PCI busses connected to a
particular device

©2008, Dr. Meeta Yadav, www4.ncsu.edu/~myadav/Research 4

 What to Randomize?

• Primary Input Data

 Randomize input data
 Constrain the inputs to be within valid limits

• Encapsulated input data

 If the data is encapsulated, the different layers of encapsulation
should be randomized and the device should be tested with different
input configurations
 The encapsulated data should also be randomized

©2008, Dr. Meeta Yadav, www4.ncsu.edu/~myadav/Research 5

 What to Randomize?

• Protocol exceptions, errors and violations

 Verify how system handles errors
 Anticipate error cases and inject into the system to ensure that the
design gracefully handles them
 Inject errors randomly

• Delays
 Use random legal delays specified by the protocols
 Check for clock jitter sensitivity
 Do not try to violate set up and hold requirements

©2008, Dr. Meeta Yadav, www4.ncsu.edu/~myadav/Research 6

 Randomization in SystemVerilog

• Overview
 Randomization enables users to automatically generate random input
stimulus for functional verification
 SystemVerilog enables user to specify random constrained (legal)
values
 Random constraints should be specified using OOP

©2008, Dr. Meeta Yadav, www4.ncsu.edu/~myadav/Research 7

 Randomization in SystemVerilog

• rand keyword
 Random variables are declared with the rand keyword
 Their values are uniformly distributed over the specified range
 If unconstrained the variable shall assign any value in the specified range
with equal probability

rand bit [7:0] y; 8-bit unsigned integer with the range 0 to 255

The probability of the same value repeating on successive calls to randomize is 1/256
•If unconstrained y is assigned any value in the range 0 to 255 with equal probability

•The probability of a value occurring on simultaneous calls to randomize is 1/256

©2008, Dr. Meeta Yadav, www4.ncsu.edu/~myadav/Research 8

 Randomization in SystemVerilog

• randc keyword
 Random cyclic variables are declared with the randc keyword
 They cycle through all the values in a random permutation of their
declared range
 Can only be of the type bit or enum

randc bit [1:0] y; 2-bit unsigned integer with the range 0 to 3

 randc randomly iterates over all the values in the range and no value is
repeated within an iteration
– When the iteration finishes, a new iteration automatically starts
y can take values 0,1,2,3

initial permutation 0 -> 3 -> 2 -> 1

next permutation 2 -> 1 -> 3 -> 0

next permutation 2 -> 0 -> 1 -> 3

©2008, Dr. Meeta Yadav, www4.ncsu.edu/~myadav/Research 9

 Randomization in SystemVerilog

• Simple class with random variables

 The Bus class models a simplified bus with two random variables:
addr and data
 The range1 constraint specifies the limits for the values of addr
 Ensure non conflicting and legal constraints

class Bus
rand bit[15:0] addr; random variable
randc bit[31:0] data; random cyclic: the random solver
constraint range1 {addr>1024; addr<16384;} will not repeat a permutation of
endclass values
Example: Simple class with random variables constraining the random variables
to values between 1024 and
16384

©2008, Dr. Meeta Yadav, www4.ncsu.edu/~myadav/Research 10

 Randomization in SystemVerilog
• randomize()function
 Calling randomize() causes new values to be selected for all of the
random variables in an object
 The random values obey the constraints
 randomize function returns a 1 on success and 0 on failure
 Unconstrained variables are assigned any values in their declared
range
a bus object is created and
randomized 50 times result of each randomization
is checked

Bus bus=new;
repeat (50) begin
if(bus.randomize()==1)
$display(bus.addr, bus.data);
else
$display (“Randomization failed”);
end
Example: randomize() function example

©2008, Dr. Meeta Yadav, www4.ncsu.edu/~myadav/Research 11

 Randomization in SystemVerilog

• The constraint solver

 Solves constraint expressions
 The same seed results in the same random values
 Use a different seed to generate different set of random values
 The solver is specific to the simulation vendor

©2008, Dr. Meeta Yadav, www4.ncsu.edu/~myadav/Research 12

 Constraint Details
• Constraint Blocks
 Values of random variables are determined using constraint expressions that
are declared using constraint blocks
 Smart stimulus tests relationships between variables
 They are class members like tasks, functions and variables
 Constraint declaration
 Constraint identifier: is the name of the constraint block.
 Constraint block: is a list of expression statements that restrict the range of variable
or define relations between variables

In the example below randomize() fails sometimes and succeeds sometimes, why?
constraint_identifier constraint_block
Error!!!
rand bit[15:0] addr;
class Bus
bit[15:0] addr;
randc bit[31:0] data;
constraint range1 {addr>1024; addr<16384;}
endclass
Example: Simple constraint example

©2008, Dr. Meeta Yadav, www4.ncsu.edu/~myadav/Research 13

 Constraint Details
• Simple expressions
 Constraint variables have to be in a fixed order
 There can be only one relational operator (<, <=, ==, >= or >)
 For multiple variables use multiple expressions

Declare a constraint to solve: 0 < a < b < c

Error!!! wrong way right way

class Bus class Bus

rand bit[15:0] a,b,c; rand bit[15:0] a,b,c;
constraint c1 { 0 < a < b < c; } constraint c1 { 0 < a;
endclass a < b;
b < c;}
endclass

©2008, Dr. Meeta Yadav, www4.ncsu.edu/~myadav/Research 14

 Constraint Details: Example
• Set membership Operator: inside
 If other constraints are absent, all values have an equal probability of
being chosen by the inside operator
 The negated (!) form of inside operator denotes that expression lies
outside the set

addr is between 0 to 100 multiple mutually exclusive

or between 1024 to 16384 constraints can be defined
within one constraint

class Bus
rand bit[15:0] addr;
randc bit[31:0] data;
constraint range1
{
addr inside {[0:100],[1024:16384]};
data > 1000;
data < 10000;
}
endclass
Example: Constraint example with inside operator
©2008, Dr. Meeta Yadav, www4.ncsu.edu/~myadav/Research 15
 Constraint Details: Quiz
Declare constraint blocks c1,c2,c3 and c4 so that variable x, a and v
get the values shown:
x takes any values described in the constraint_block :
• x is: 3,5,9,10,11,12,13,14,15,24,25,26,27,28,29,30,31,32
• x is: 3,5,9,10,11,12,13,14,15,24,25,26,27,28,29,30,31,32,
between y and 2y, z
rand integer x,y,z;
constraint c1 {x inside {3,5,[9:15],[24:32];}
constraint c1 {x inside {3,5,[9:15],[24:32],[y:2*y],z};}

• b<=a and a<=c or b<=a<=c

rand integer a,b,c;
constraint c2 {a inside {b,c};}

• c3: v is either 5,10,15,20

• c4: v is never 5,10,15,20
integer fives[0:3] = {5,10,15,20};
rand integer v;
constraint c3 {v inside fives;}
constraint c4 !{v inside fives;}

©2008, Dr. Meeta Yadav, www4.ncsu.edu/~myadav/Research 16

 Weighted Distribution
• Distributions: dist operator
 Used to weigh some values more than the others
 Property 1: They are a relational test for set membership
 Property 2: They specify a statistical distribution function for the result
 Can use := or :/ operator
 := specifies that the weight has to be the same for every specified value in
the range
 :/ operator specifies that the weight is to be equally divided between all
values. If there are n values in a range the weight is range_weight/n
 Values can be a single value or range such as [lo:hi]
 The weights are not percentage and do not have to add up to a 100
 Cannot be used with a randc

c1 should be 0 once and 1 five times

Error!!! wrong way right way
rand integer a; rand integer a;
constraint c1 {a inside {0,1,1,1,1,1}); constraint c1 {a dist {0:=1, 1:=5}};

©2008, Dr. Meeta Yadav, www4.ncsu.edu/~myadav/Research 17

 Weighted Distribution

• Another example
rand int src, dst;
costraint c1 {
src dist {0:=40, [1:3]:=60};
dst dist {0:=40, [1:3]:/60};
}
Example: Weighted random distribution with dist

src weight dst weight

0 40/220 0 40/100
1 60/220 1 20/100
2 60/220 2 20/100
3 60/220 3 20/100

©2008, Dr. Meeta Yadav, www4.ncsu.edu/~myadav/Research 18

 Weighted Distribution: Quiz
• Distributions: dist operator := and :/
• x is 100, 200 and 300 with a weight of 1,2 and 5 respectively

x dist {100:=1, 200:=2, 300:=5}

• x is 100 and 300 with a weight of 1 and 5 respectively

• x is never 200

x!=200;
x dist {100:=1, 300:=5}

• x is 100, 101 , 102 with a weight of 1 each

• x is 200 and 300 with a weight of 2 and 5 respectively

x dist {[100:102]:=1, 200:=2, 300:=5}

• x is 100, 101 , 102 with a weight of 1/3 each

• x is 200 and 300 with a weight of 2 and 5 respectively

x dist {[100:102]:/1, 200:=2, 300:=5}

©2008, Dr. Meeta Yadav, www4.ncsu.edu/~myadav/Research 19

 Bidirectional Constraints

• Bidirectional Constraints
 Constraint blocks are not procedural but declarative
 All active at one time

Solve the constraint

30>d>b==c>=25

Solution b c d
rand logic [15:0] b,c,d;
costraint c1 { A 26 26 27
b<d;
c==b; B 26 26 28
d<30; C 26 26 29
c>=25;
} D 27 27 28
Example: Bidirectional constraints E 27 27 29
F 28 28 29
Solutions for Bidirectional Constraints

©2008, Dr. Meeta Yadav, www4.ncsu.edu/~myadav/Research 20

 Conditional Constraints
• Conditional constraint operators
 Constraint provide two constructs for declaring conditional relations
 Implication operator ->
 if…else

if(mode==small)
len<10; mode == small -> len<10;
else if (mode==large) mode == large -> len>100
len>100;
Example: if…else example Example: Equivalent implication example

bit [3:0] a,b; bit [3:0] a,b;

if(a==0) constraint c {(a==0)->(b==1);}
b==1;
Example: if…else example Example: Equivalent implication example

©2008, Dr. Meeta Yadav, www4.ncsu.edu/~myadav/Research 21

 Solution Probabilities
• Unconstrained
 Probability of distribution

Solution x y Probability
A 0 0 1/8
B 0 1 1/8
class Unconstrained;
rand bit x; C 0 2 1/8
rand bit [1:0] y;
D 0 3 1/8
endclass
E 1 0 1/8
Example: Class with unconstrained random variables
F 1 1 1/8
There are 8 possible solutions and G 1 2 1/8
because there are no constraints each
has the same probability H 1 3 1/8
Solutions for Unconstrained class

©2008, Dr. Meeta Yadav, www4.ncsu.edu/~myadav/Research 22

 Solution Probabilities
• Implication
 Probability of distribution changes due to the implication operator
 Implication is bidirectional

class Imp1; Solution x y Probability

rand bit x;
A 0 0 1/2
rand bit [1:0] y;
constraint c_xy { B 0 1 0
(x==0)->y==0;
}
C 0 2 0
endclass D 0 3 0
Example: Class with implication E 1 0 1/8
F 1 1 1/8
G 1 2 1/8
Value of y depends on x. When x is 0, y
is 0. Hence for x=0, y cannot have any H 1 3 1/8
other value and hence the probability of Solutions for Imp1 class
x=0 and y!=0 is 0

©2008, Dr. Meeta Yadav, www4.ncsu.edu/~myadav/Research 23

 Solution Probabilities
• Implication and bidirectional constraints
 Probability of distribution changes due to an addition constraint

class Imp_Bid; Solution x y Probability

rand bit x;
rand bit [1:0] y; A 0 0 0
constraint c_xy { B 0 1 0
y>0;
(x==0)->y==0; C 0 2 0
} D 0 3 0
endclass
E 1 0 0
Example: Class with implication and constraint
F 1 1 1/3
G 1 2 1/3
H 1 3 1/3
When x is 0, y is 0 but when y is 0 Solutions for Imp_Bid class
there is no constraint on x. Since
implication is bidirectional, if y was
forced to a non zero value, x would
have to be 1. Hence x can never be
0.

©2008, Dr. Meeta Yadav, www4.ncsu.edu/~myadav/Research 24

 Solution Probabilities
• Solve…before
 Solve…before does not change the solution space but changes the probability
of the results

class SolveBefore;
rand bit x;
rand bit [1:0] y;
constraint c_xy {
(x==0)->y==0;
solve y before x;
}
endclass
Example: Class with implication and solve…before

©2008, Dr. Meeta Yadav, www4.ncsu.edu/~myadav/Research 25

 Solution Probabilities
class SolveBefore;
rand bit x;
rand bit [1:0] y;
constraint c_xy {
(x==0)->y==0;
solve y before x;
}
endclass
Example: Class with implication and solve…before
Solution y x Unconstrained Solution Probability
Probability Possibility
A 0 0 1/8 1/8
B 0 1 1/8 1/8
C 1 0 1/8 0
D 1 1 1/8 1/4
E 2 0 1/8 0
F 2 1 1/8 1/4
G 3 0 1/8 0
H 3 1 1/8 1/4
Solutions for solve y before x constraint
©2008, Dr. Meeta Yadav, www4.ncsu.edu/~myadav/Research 26
 Solution Probabilities
• Solve…before
 Solve…before does not change the solution space but changes the probability
of the results

Solution y x Probability
class SolveBefore;
A 0 0 1/8
rand bit x;
rand bit [1:0] y; B 0 1 1/8
constraint c_xy {
C 1 0 0
(x==0)->y==0;
solve y before x; D 1 1 1/4
}
E 2 0 0
endclass
F 2 1 1/4
Example: Class with implication and solve…before
G 3 0 0
H 3 1 1/4
Solutions for solve y before x constraint

©2008, Dr. Meeta Yadav, www4.ncsu.edu/~myadav/Research 27

 Solution Probabilities
• Solve…before

Solution x y Probability
class Imp4; A 0 0 1/2
rand bit x;
rand bit [1:0] y; B 0 1 0
constraint c_xy { C 0 2 0
(x==0)->y==0;
solve x before y; D 0 3 0
} E 1 0 1/8
endclass
F 1 1 1/8
Example: Class with implication and solve…before
G 1 2 1/8
H 1 3 1/8
Solutions for solve x before y constraint

©2008, Dr. Meeta Yadav, www4.ncsu.edu/~myadav/Research 28

 Constraints
• Iterative constraints
 Allows arrayed variables to be constrained in a parameterized manner using
loop variables

C1 constraints each element of an array to be in the set [2,4,8,16]

C2 constraints each element of an array to be greater than twice its index

class C;
rand byte A[4];
constraint C1{ foreach (A[i]) A[i]inside {2,4,8,16};}
constraint C2{ foreach (A[j]) A[j]> 2*j;}
endclass

©2008, Dr. Meeta Yadav, www4.ncsu.edu/~myadav/Research 29

 Constraints
• Iterative constraints
bit [3:0][2:1] B [5:1][4]
int A [2][3][4];
2 1

3 3 2 1 0

4 5:1

foreach (A[i,j,k])…
4
i iterates from 0 to 1
j iterates from 0 to 2 foreach (B[q,r,,s])…
k iterates from 0 to 3
q iterates from 5 to 1
r iterates from 0 to 3
s iterates from 2 to 1

©2008, Dr. Meeta Yadav, www4.ncsu.edu/~myadav/Research 30

 Constraints
• Functions in constraints
 Some properties are unwieldy or impossible to express in a single expression
 For instance computing the sum of one’s in a packed array uses a loop
 Without the loop the loop will have to be unrolled
 SystemVerilog allows constraint expressions to include function calls
 Functions cannot contain output or ref arguments
 Functions should be automatic
 Functions that appear in constraints cannot modify the constraints
 Functions shall be called before constraints are solved, and their return values shall
be treated as state variables
 Random variables used as function arguments shall establish an implicit variable
ordering or priority

class B;
rand int x,y;
constraint C{x<=F(y);} Forces y to be solved before x
constraint D{ y inside {2,4,8};}
endclass

©2008, Dr. Meeta Yadav, www4.ncsu.edu/~myadav/Research 31

 Constraints
• Constraint guards
 Constraint guards are predicate expressions that function as guards against
creation of constraints
 They are not logical relations that have to be satisfied by the constraint solver
 Prevents the solver from generating evaluation errors
 Constraint guards are solved before the constraints and involve
 Constants
 State variables
 Object handle comparisons
The constraint will fail on the last element due to a non existent handle in the linked list
class SList;
rand int n;
constraint sort{n<next.n;}
endclass

if statement acts as a constraint guard

class SList;
rand int n;
constraint sort{if(next!=null)n<next.n;}
endclass

©2008, Dr. Meeta Yadav, www4.ncsu.edu/~myadav/Research 32

 Constraints
• Constraint guards
 Guard expressions can themselves include sub-expressions that result in
evaluation errors
 A 4-state representation is used where
– 0 : false -> Subexpression evaluates to FALSE
– 1 : true -> subexpression evaluates to TRUE
– E: Error -> Subexpression causes an evaluation error
– R: Random -> Expression includes random variables and cannot be evaluated

class D
int x;
endclass
class C;
rand int x,y;
D a,b;
constraint C1{(x<y || a.x > b.x || a.x==5) -> x+y==10;}
endclass

Case a b a.x b.x Constraint

1 !0 0 5 error x+y=10
2 0 - error - always error
3 !0 !0 10 20 ( x<y) -> (x+y==10)
©2008, Dr. Meeta Yadav, www4.ncsu.edu/~myadav/Research 33
 Constraints
• Constraint guards

class D
int x;
endclass
class C;
rand int x,y;
D a,b;
constraint C1{(x<y && a.x > b.x && a.x==5) -> x+y==10;}
endclass

Case a b a.x b.x Constraint

1 !0 0 5 error error
2 0 - error - always error
3 !0 !0 5 2 ( x<y) -> (x+y==10)

©2008, Dr. Meeta Yadav, www4.ncsu.edu/~myadav/Research 34

 Disabling Random Variables
• Disabling random variables with rand_mode()
 Can be used to control whether a random variable is active or inactive
 When a random variable is inactive it implies that the variable was never declared
rand or randc
 rand_mode() method is inbuilt and cannot be overridden

Value Meaning Description

0 OFF Sets the specified variable to inactive so that they are not randomized on
subsequent calls to randomize() method
1 ON Sets the specified variable to active so that they are randomized on
subsequent calls to randomize() method

class Packet;
rand integer src, dst;
endclass

int r;
Packet packet_a=new;
packet_a.rand_mode(0);
packet_a.src.rand_mode(1);
r=packet_a.dst.rand_mode();

©2008, Dr. Meeta Yadav, www4.ncsu.edu/~myadav/Research 35

 Controlling Constraints
• Controlling constraints with constraint_mode()
 Can be used to control whether a constraint is active or inactive
 When a constraint is inactive is not considered by randomize()
 All constraints are initially active
 constraint_mode method is built it and cannot be overridden

Value Meaning Description

0 OFF Sets the specified constraint block to inactive so that it is not enforced by
subsequent calls to randomize() method
1 ON Sets the specified constraint block to active so that it is considered by
subsequent calls to randomize() method

class Packet;
rand integer src, dst;
constraint filter{src>2*dst;}
endclass

function integer toggle_rand (Packet p);

if (p.filter.constraint_mode()==1)
p.filter.constraint_mode(0);
else
p.filter.constraint_mode(1);
toggle_rand=p.randomize();
endfunction

©2008, Dr. Meeta Yadav, www4.ncsu.edu/~myadav/Research 36

 Thank You

©2008, Dr. Meeta Yadav, www4.ncsu.edu/~myadav/Research 37