Functional verification using System verilog introduction

Content
• Functional Verification
• Verification Process
• Introduction to SV HVL
• Data Types

Functional Verification:
• Test Bench produces the stimulus and stimulus given to the
DUV (Design Under Verification) then DUV produces the
output, and output collected by test bench
• We always choose the direct test cases to verify the
functionality
• Most important question would be, how we going to find
bugs, more bugs and faster
• That’s where we look a new methodologies like “constrained
Random Driven Verification
• As direct test cases time consuming, we look into new
languages called system Verilog

Functional Verification:
Test Bench
RTL
(DUV)

Verification Process:
Test-n
Test-2
Test-1
DUT
Specifications +
Verification Plan
Coverage
closure
Verification
signoff
DUT for
Synthesis
PM
DUT
Spec
V Plan
Test case
writers
Test Bench
Developers
DUT
TB
Environ
ment
TB
Implementation
Regression
Testing

Introduction to SV HVL
HDL Limitations:
• HDLs are basically static in nature
• As it is static, it consumes more memory, one can’t change the
data structure dynamically
• Randomization –Constraints: There are very limited features
are available in HDL, you won’t able to randomize the
complete packets
• Code coverage-RTL Measure w.r.to design Intent missing
• Functional coverage missing
• Reusability-Reuse without breaking the already working
code

System Verilog HDVL:
• A set of extensions to Verilog -2001
• An unified language for design and verification
• A language which supports all the verification
• Constrained Random Coverage Driven Verification
• Assertion based verification
• Functional Coverage
• A language which supports object oriented programming

System Verilog Sources:
• System Verilog is an IEEE standard language based on Verilog
2001
• System Verilog extensions come from a number of sources
including :
• C & C++ : Data types and Object Oriented Programming
• PSL: Assertion based Verification
• VERA: Randomization
• VHDL: enum, Packages
• DPI(Direct Programming Interface): from Synopsys
• System Verilog makes interface easily

Data Types: Verilog 2001
• Two main groups of data types-nets and registers
• Four values 0,1,X and Z
• Register data types-reg, integers, time and real
• Stores values-static storage
• Are assigned in a procedural block(initial /always)
Example:
reg [15:0] data_bus; //16-bit unsigned variable
int counter; //32-bit signed variable
real float_num; //64-bit unsigned floating point numbers
time sim_time; //64-bit unsigned Variable

Data Types: System Verilog
• logic: 4-state data type (similar to reg)
• Two state data types
• enum: Declares a variable
• typedef: user defined Data Type
• struct and class

Data Types: logic
• logic- functionality is similar to reg
• Logic data type can be used anywhere that the reg and net
data types are used, except with inout ports
• The logic datatype is synthesizable, and it an used to define
both combinational and sequential logic
• Confusion in Verilog HDL, reg datatype can be used for both
combinational and sequential but in hardware reg [Register]
refers to sequential

Data Types: logic
Example:
module tb_dff( );
parameters CYCLE=20;
logic q,q_b,d,clk,rst_n;
initial
begin
clk=0; rst_n;
forever
#(CYCLE/2) clk=~clk;
@(negedge clk)
rst_n=1;
@(negedge clk)
d=1;
end
assign q_b=~q
diff DUT(q, d, clk, rst_n);//DUT instantiation
endmodule:tb_dff

Data Types: 2-State Data Type
• bit : 2-state scalar type
• byte : 8-bit signed integer 2-sate type
• shortint : 16-bit signed integer 2-state type
• int : 32bit signed integer 2-state type
• longint: 64-bit signed integer 2-state type
When a value is assigned to an object of any of these types, X or
Z values are converted to 0

Data Types: 2-State Data Type
Example:
bit clk; //2-state , bit value : 0 or 1
bit [31:0]vbus; //2 state 32-bits unsigned integer
int cnt; //2 state 32-bits signed integer
int unsigned cnt; // 2 state 32-bits unsigned integer
byte crc_byte; //2-state 8-bits signed integer
shortint sint; //2-state 16-bits signed integer
longint lint; //2-state 64-bits signed integer

Data Types: Struct Type
Struct Types:
• Struct –collection of data static
• Array of different data types
• Synthesizable
• Subset of class-use class for TB
Example:
struct{bit[7:0] address;
logic [7:0] payload[0:63];
enum [Good, Error, ] pkt_type;
logic [7:0] parity}packet;
packet.adress=8’d25;
packet.pkt_type=Error;

Data Types: User Defined Type
 Typedef used to define a user defined data type
typedef bit[31:0]word_t;
word_t word1,word2;
struct{ bit[7:0] r,g,b;}pixel;
pixel.r=8’d25;
pixel.g=8’d255;
pixel.b=8’d11;
typedef struct{bit [7:0] r,g,b;}pixel_st;
pixel_st samsung_pixel;
pixel_st sony_pixel;
samsung_pixel.r=8’d25;
sony_pixel=‘{8’d38,8’d98,8’d69};

Data Types: Enumerated
• Enum-enumerated data type(list of values)
enum{init, read, write, brd, bwr}fsm_state;
fsm_state=init;
fsm_state=read;
Example:
typedef enum {init,read,write,brd,bwr}fsm_state_et;
fsm_state_et Pre_state,Next_state;
fsm_state_et state=state.first;
initial
begin
$display(“%s:%d”,state.name, state);
if(state==state.last(1))
break;
state=state.next();
end

Data Types: Enumerated
• Default values can be given to constants
Ex: typedef enum{init, read=2, write, brd=7, bwr} fsm_state;
• Built in methods:
1. first() : //returns first element
2. last() : //returns last element
3. next() : //returns next element
4. prev() : //returns previous element
5. next(N) : //returns Nth
next element
6. prev(N) : //returns Nth
previous element

Data Types: Type Conversion
• Conversion between datatypes can be done in 2 ways
• Static casting – No checking of values
int ic;
real rc;
ic= int’(20.0-0.1);
rc= real’ (20);
• Dynamic casting : using $cast
• It allows to check for out of bound values
• It can be used as a function or task

Data Types: Type Conversion
typedef enum{init, read, write, brd, bwr}fsm_state_et
fsm_state_et state;
Int c;
Initial
begin
state=read;
c=state;
state=1; //can’t be done
state=fsm_state_et(5) // static casting : no type checking and won’t
give error
$cast(state,5);//Dynamic casting as a task, gives us
as a 5 is out if bound values
if($cast(state,5))
$display(“CASTING SUCCESSFUL”);
else
$display(“CASTING FAILED”)
end

Data Types: Strings
• Variable length – grows automatially
• Memory is dynamically allocated
string str;
initial
begin
str=“Rgukt_Sklm”;
str=str.tolower();//rgukt_sklm
$display(“Character in 5th
position is ”,str.getc(5)):
str.putc(5, “-”); // “_” is replaced with “-”
$display(“%s”, str.substr(6,str.len-1)); //sklm is printed
str={str, “.com”};//rgukt-sklm.com
str={{3{w}}, “.”,str};
disp($sformat(“https://%s”,str);
end
task disp(string s);
$display(“at time t=%d, %s”, $time,s);
//at time t=0, https://www.rgukt-sklm.com
endtask

Data Types: Packages
• User defined packages
• Define type definitions, functions and global variables for reuse
throughout the design or verification environment
• Good for functions like printing error messages, reading/writing a
bos etc.
• Good for reusable environment –test cases import the packages

Example:
package pkg;
int no_of_trans;
function void display(string s);
$display($time,”%s,n=%d”,s,no_of_trans);
endfunction
endpackage
module A
import pkg::*;
initial
begin
#1;
no_of_trans=10;
#1;
display(“From Module A:”);
end
endmodule

module B
import pkg::*;
initial
begin
#4;
no_of_trans=20;
display(“From Module B:”);
end
endmodule
Result:
2, From Module A, n=10
4, From Module B, n=20

SV Memories :Packed Arrays
• Some times you may want to access the entire array value and also
divide it into smaller elements
• For example, A 32-bit register can be considered as a group of four
8-bit registers or a group of 32 bits
• SV Packed array is treated as both an array and a single value
//$-bytes packed into 32-bits
bit [3:0][7:0]byte_packed;
byte_packed=32’h5454_5454;
$displayh(byte_packed,byte_packed[0],byte_packed[0][0]);
Show all 32 bits
Least significant byte Least significant bit

SV Memories :Multidimensional Arrays
//unpacked Arrays of 3 packed elements
bit[3:0][7:0] bytes[0:2]
bytes [0]=32’d255;
bytes [0][3]=8’d255;
bytes [0][1][6]=1’b1;

SV Memories :Accessing the elements of an Arrays
logic [31:0] msrc[5], mdst[5];
initial
begin
for (int i=0; i< $size (msrc);i++);
mdst[j]=msrc[j]*4;
end
int mda [3][3]=‘{‘{0,1,2},’{3,4,5},’{6,7,8}};
initial
begin
foreach(mda[I,j])
$display(“mda[%0d][%0d]=%0d”,I,j,mda[i][j])
end

SV Memories : Array –Aggregate copy and compare
bit [31:0] msrc[5]={0,1,2,3,4},
mdst[5]={5,4,3,2,1};
//copy and compare all values without any loop
initial
begin
if(msrc==mdst) //compare all msrc values with mdst
$display(“msrc is equal to mdst”);
else
$display(“msrc is not equal to mdst”);
mdst=msrc //copy all values to mdst
//Compare 4 values of msrc with mdst
if(msrc[1:4]==mdst[1:4])
$display(“msrc is equal to mdst”);
else
$display(“msrc is not equal to mdst”);
end

SV Memories : Dynamic Arrays
• Packed arrays –size to be defined before compilation
• How can we define the array size to store random number if
packets
• Random number of packets could be less than the maximum value
expected leads to memory wastage
• SV provides dynamic memory which can be allocated and resized
dynamically during simulation

SV Memories : Dynamic Arrays
• Variable size single dimension memory-size changes dynamically
during run time
int da1[],da2[];
initial
begin
da1=new[10];//Allocate 10 elements
foreach (da1[i]) //initializing
da1[i]=i;
da2=da1; //copying
da1=new[30]da(da1); //Allocate 30 new & copy existing int
da1=new[100]; //Allocate 100-new ints-old values are lost
da2.delete( ); //Delete all elements of da2
end

SV Memories : Queues
• Size of the Queue grows and shrinks automatically-No need to
define the size; before running simulation
• Dynamic array size should be allocated with a fixed value-needs
size before the usage
• Queue-New elements can be added/removed easily
• Queue-Elements can be accessed from any location deirectly using
index
• Flexible Memory-Variable sizing
• Single dimension array with automatic sizing
• Many searching, sorting and insertion methods
Syntax: type name[$]

SV Memories : Queues
int qm1[$]={1,3,4,5,6};
int qm2[$]={2,3};
int k=2;
qm1.insert(1,k); //K-insert before ‘3’
qm1.delete(1); //Delete the element #1
qm1.push_front(7); //insert at front
k=qm1.pop_back(); //k=6
qm2.push_back(4); // insert at back
K=qm2.pop_front(); //k=2
foreach(qm1[i])
$display(qm1[i]); // display the element of the entire queue
qm2.delete(); //delete entire queue

SV Memories : Associative Arrays
• How do you model the memories of huge size-10GB RAM
• You may write data into only very few locations of 10GM RAM
• Allocating 10GB might lead to either wastage of memory or running
out of memory
• Associative arrays: memory will be allocated only writing the data
• Associative arrays : Memory read and write can happen in non-
contiguously
Example:
• Great for creating memory models –sparse memories
• Dynamically allocated, non-contiguous elements
• Accessed with integer, or string index
• Great for spare arrays with wide ranging index
• Builtin functions: exists, first, last, next, prev
Syntax: type name[*];

SV Memories : Associative Arrays
Example:
• Great for creating memory models –sparse memories
• Dynamically allocated, non-contiguous elements
• Accessed with integer, or string index
• Great for spare arrays with wide ranging index
• Builtin functions: exists, first, last, next, prev
Syntax: type name[*];
All Memory
Allocated, even
unused
elements
Un used
Elements don’t
use memory
Standard Array Associative Array

SV Memories : Array Methods
• Array methods can be used only with unpacked arrays
• Fixed, Dynamic, Queue, and associative Arrays
int cnt, sa;
int da[ ]={10,1,8,3,5,5};
cnt=da.sum with(item>7); //2:[10,8]
Sa=da.sum with(item==5); //2:[5,5]
//Sorting Methods
int da[ ]={10,1,7,3,4,4};
da.revrse( ); //{4,4,3,7,1,10}
da.sort( ); //{1,3,4,4,7,10}
da.rsort( ); //{10,7,4,4,3,1}
da.shuffle( ); //{10,4,3,7,1,4}

SV Memories : Array Methods
//Array locator methods: min, max unique
int da[6]=[1,5,2,6,8,6];
int mq[$];
mq=da.min( );//[1]
mq=da.max( );//[8]
mq=da.unique( );//[1,5,2,8,6]
int da[6]={1,6,2,6,8,6};
mq=f.find_first with (item>5); //[6]
mq=f.find_first_index with (item>5); //[1] da[1]=6
mq=f.find_last with (item>5); //[6]
mq=f.find_last_index with (item>5); //[5] da[5]=6

Functional verification using System verilog introduction

More Related Content

Similar to Functional verification using System verilog introduction (20)

Recently uploaded (20)

Functional verification using System verilog introduction