State university of New York at New Paltz

Electrical and Computer Engineering Department

Introduction to VHDL

By
Dr. Yaser Khalifa
Electrical and Computer Engineering Department
State University of New York at New Paltz
 What is VHDL?

• VHDL stands for VHSIC Hardware

Description Language.
• VHSIC is an abbreviation for Very High
Speed Integrated Circuit, a project
sponsered by the US Government and Air
Force begun in 1980 to advance techniques
for designing VLSI silicon chips.
VHDL is an IEEE standard.
 Using VHDL for design synthesis

The design process is a 6 step cycle:

• Define the design requirement
• Describe the design in VHDL code
• Simulate the source code
• Synthesize, optimize, and fit the design
• Simulate the design
• Implement the design
 Design requirements

Required setup, clock requirements,

maximum operating frequency, critical
paths.
Describe the design in VHDL code
Prior to this a design methodology should be used
to describe the system. The most common of these
are:

• Top-down : In top-down, the main functional

blocks are first defined, where each block has its
defined inputs and outputs and is performing a
specific function. Then a description of lower
levels follows.
•Bottom-up : In this case first the
individual blocks are defined and designed
and then they are brought together to form
the overall design.

•Flat : In this case the design is normally

simple and does not need a hierarchical
approach.
 Simulate the source code

• This is an important feature of Hardware

Description Languages as the simulation is
done before the synthesis and design stage.
 Synthesize, optimize and fit the design

• Synthesis tools in VHDL, allow designers to

design logic circuits by creating design
descriptions without necessarily having to
perform Boolean algebra or create
technology-specific, optimized functions.
Optimization process, is the process of
minimizing the circuitry by means of
reducing the sum of product terms.
However, the process is multi-objective
and dependent on the expression being
realized and the device used in implementation.
Fitting, is the process of taking the logic
produced by the synthesis and optimization
processes and placing it into a logic device.

Place and route, is fitting the logic into the

logic device and placing the logic elements
in optimal locations and routing the signals
from one unit to another in an optimal way.
 Design Entity

A design entity is design element in VHDL

and consists of:
• an entity declaration and
• an architecture body.
A
B

C
D
 Entity

An entity body describes the input and output

pins (ports) of the design entity.

entity AOI is
port ( A , B , C , D : in STD_LOGIC ;
F : out STD_LOGIC ) ;
end AOI ;
architecture V1 of AOI is
begin
F <= not ( ( A and B ) or ( c and D ) ) ;
end V1 ;
 Entity Declaration

Each port in an entity should include :

• name of the signal (identifier)
• direction (mode)
• data type
Modes : There are four default modes in
VHDL
in : input into the entity ( unidirectional)
out : output from the entity( unidirectional)
inout : input and output to and from the entity
( bidirectional signals )
buffer : behaves in a similar way as inout,
but the source of the buffered signal is always
determined by the driving value of the port.
 Data Types
There are scalar and array data types in
VHDL.

array types example

string “abc”
bit_vector “1001”
std_logic_vector “101Z”
 Scalar example
type
character ‘a’
bit ‘1’ ‘0’
std_logic ‘0’ forcing 0
‘1’ forcing 1
‘x’ forcing unknown
‘z’ high impedance
‘-‘ don’t care
‘L’ weak 0
‘H’ weak 1
‘U’ uninitialized
boolean true false
real 2.35 -1.0E+38
integer 832 -1
time fs, ps, ns, us, ms, sec, min, hr
 Signals
An internal connection is described in VHDL as a
signal defined inside the architecture.
A AB
B
O
F
C
D CD

architecture V2 of AOI is
signals AB, CD, O : STD_LOGIC ;
begin
AB <= A and B after 2 NS;
CD <= C and D after 2 NS;
O <= AB or CD after 2 NS;
F <= not O after 1 NS;
end V2 ;
 Concurrent Statements
• VHDL statements are grouped into sequential and
concurrent statements. Concurrent statements are
used in data flow and structural descriptions.
Sequential statements are used in bahavioral
descriptions

Concurrent statements are mainly:

Concurrent signal assignment statements with
BOOLEAN equations
• Selective signal-assignment (with-select-when)
Conditional statements (if-then-else)
 Concurrent Statements
A concurrent signal assignment is triggered by an
event on a signal. An event is a change in value.

An event on the input port A would trigger the

assignment to AB

port (A, B, C, D: in STD_LOGIC;

...
AB <= A and B after 2 NS;
 port (A, B, C, D: in STD_LOGIC;
...
AB <= A and B after 2 NS;

A
B 2 NS

AB
AB <= A and B after 2 NS;
CD <= C and D after 2 NS;
O <= AB or CD after 2 NS;
F <= not O after 1 NS;

AB

F
 Variables
Previously we have looked at signals as electrical
connections or “pieces of wires”.
Variables can be pieces of wire too, but they can also
be more abstract.
Variables can represent wires, registers, or be used to
store intermediate values in abstract calculations.

variable V: STD_LOGIC;

Variables must be defined inside a process, before

begin (unlike signals which are defined in an
architecture).
 Components
A component is analogous to a chip socket. It
creates an extra interface which allows
more flexibility when building hierarchical
system out of its component parts.
library IEEE;
use IEEE.STD_LOGIC_1164.all;

entity MUX2 is
port (SEL, A, B: in STD_LOGIC;
F: out STD_LOGIC);
end MUX2;

SEL
A

F
SELB
B
architecture STRUCTURE of MUX2 is
component INV
port (A: in STD_LOGIC;
F: out STD_LOGIC);
end component;
component AOI
port (A, B, C, D: in STD_LOGIC;
F: out STD_LOGIC);
end component;
signal SELB: STD_LOGIC;
begin
G1: INV port map(SEL, SELB);
G2: AOI port map(SEL, A, SELB, B, F);
end STRUCTURE;
 Variables

Previously we have looked at signals as electrical

connections or “pieces of wires”.
Variables can be pieces of wire too, but they can also be
more abstract.
Variables can represent wires, registers, or be used to store
intermediate values in abstract calculations.

variable V: STD_LOGIC;

Variables must be defined inside a process, before begin

(unlike signals which are defined in an architecture).
 A variable has a name and a type, just like a signal.

-- Assume A = '0', B = '0', C = '1'

process (A, B, C)
variable V: Std_logic; [ V = ’U’ ]
begin
V := A nand B; [ V = ’1’ ]
V := V nor C; [ V = ’0’ ]
F <= not V;
end process;

Variables behave quite differently to signals; variable assignments never have

a delay. Signal assignments always have a delay (even if it is a delta delay ).
architecture ... When this process is synthesized, each
variable assignment creates a new logic level.
signal F: Std_logic; A new wire is synthesized to hole the value of
the variable each time it is assigned.
begin
process (A, B, C)
variable V: Std_logic;
begin A V
V := A nand B; B V
V := V nor C;
F
C
F <= not V;
end process;

process (F)
begin
... A variable can be used ONLY inside the
end process; process. So, if we want a value passed
between processes, we MUST use a signal.
 Structural Description

library IEEE;
use IEEE.STD_LOGIC_1164.all;

entity COMP4 is
port ( A, B : in STD_LOGIC_VECTOR ( 3 downto 0);
F : out STD_LOGIC) ;
end COMP4;

use WORK.GATESPKG.all ;
architecture STRUCT of COMP4 is
signal x : STD_LOGIC_VEXTOR ( 0 to 3 );
begin
u0: xnor2 port map (a(0) , b(0) , x(0)) ;
u1: xnor2 port map (a(1) , b(1) , x(1)) ;
u2: xnor2 port map (a(2) , b(2) , x(2)) ;
u3: xnor2 port map (a(3) , b(3) , x(3)) ;
u4: and4 port map (x(0) , x(1) , x(2) , x(3) , F) ;

end STRUCT;
 Process Statement Execution

A process is an operation that involves a number of

parameters identified in a sensitivity list:

proc_x : process (a , b , c )
begin
x <= a and b and c;
end process;

A process is either being executed or suspended. It is

executed when one of the signals in its sensitivity list had an
event, a change of value.

The process continues to execute until the last statement is

reached or it encounters a wait statement. It then suspends
itself.
A wait statement: is a sequential statement which causes a
process to be suspended. It can wait for a period of time
(wait for), an event on a signal in the sensitivity list (wait
on), a boolean condition to become true (wait until), or
forever (wait ;).

proc_x : process
begin
x <= a and b and c;
wait on a , b , c ;
end process;

The process is used to describe the behaviour of a part of a

system without getting into the details of the implementation,
so can be used to describe hardware at high level of
abstraction.
Concurrent statement : It is a statement which execute in
parallel with other concurrent statements in the design
hierarchy. They are written between begin and end in an
architecture. The most common concurrent statements are
a) the process
b) the component instantiation
c) concurrent signal assignment
 Sequential Statements
If Statements
An if statement is a sequential statement which conditionally executes other
sequential statements, based on the value of a Boolean condition.
An if statement can contain any number of other statements, including other
nested if statements.
if C1 = '0' then
V := A;
end if;

if C2 = '0' then
V := B;
W := C;
if C3 = '0' then
X := D;
Y := E;
end if;
else
V := C;
W := D;
end if;
An if statement is synthesised by generating a
multiplexer for every signal or variable assigned
within the statement.
The condition given at the top of the if statement
forms the select input to the multiplexer.

if C1 = '0' then A
V := A; V
else
V := B; B
end if
In the example below the if statement does not have an else part,
but the assignment V within the if still synthesizes to a
multiplexer.

if C2 = '1' and C3 = '1' then

V := not V;
end if;
C1
C2

V
V

If the condition is false, the previous value of V is fed through the

multiplexer unaltered.
process (C1, C2, C3, A, B)
variable V: Std_logic;
begin
if C1 = '0' then
V := A;
else
V := B;
end if;
if C2 = '1' and C3 = '1'
then
C1 C2
V := not V;
C3
end if;
F <= V; A
V
end process; V
B
Nested if Statements

Each if must be balanced with an end if.

process (C0, C1, C2, A, B, C, D)

begin
if C0 = '1' then
F <= A;
elseif C1 = '1' then C2
F <= B; D
elseif C2 = '1' then
F <= C; C
else
F <= D; F
B
end if;
A
end if;
end if;
end process
 The elseif allows multiple tests to be cascaded together,
such that only the branch following the first true
condition is executed.

process (C0, C1, C2, A, B, C, D)

begin
if C0 = '1' then
F <= A;
elsif C1 = '1' then
F <= B;
elsif C2 = '1' then
F <= C;
else
F <= D;
end if;
end process;

Here there is no end if after each elseif.

Example:

process
begin
if Clock = ‘0’ then
F <= ‘0’;
else
F <= A;
end if;
end process;

What kind of hardware would by synthesized?

 Case Statements
The expression at the top of the case is evaluated and compared with the
expressions following the whens. The statments within the matching when
branch are executed, then control jumps to the statement following end case.

case SEL is
when "00" => SEL
F <= A; 2
when "01" =>
F <= B;
A
when "10" =>
F <= C; B
when "11" => F
F <= D; C
when others => D
F <= 'X';
end case;

The others branch will catch any cases not already mentioned explicitly by the
when.
It is possible to cover several cases in the same when branch
by separating the values with vertical bars. In this example
the branch is executed if ADDRESS is:
16, 20, 24 or 28.

case ADDRESS is

when 16 | 20 | 24 | 28 =>
A <= '1';
B <= '1';

when others =>

end case;
Each branch can also include any number of sequential
statements including other nested case statements as well!!

case ADDRESS is

when 16 | 20 | 24 | 28 =>
A <= '1';
B <= '1';

when others =>

end case;
The null statement means do nothing

A <= '0';
B <= '0';

case ADDRESS is

when 0 to 7 =>
A <= '1';
when 8 to 15 =>
B <= '1';
when 16 | 20 | 24 | 28 =>
A <= '1';
B <= '1';
when others =>
null;

end case;
It is possible also to cover a whole range using the case statement.

case ADDRESS is

when 0 to 7 =>
A <= '1';

when 8 to 15 =>
B <= '1';

when 16 | 20 | 24 | 28 =>
A <= '1';
B <= '1';

when others =>

end case;
 If ADDRESS is:
1) =9
2) =19
What will be the values of A and B?
A <= '0';
B <= '0';

case ADDRESS is

when 0 to 7 =>
A <= '1';

when 8 to 15 =>
B <= '1';

when 16 | 20 | 24 | 28 =>
A <= '1';
B <= '1';

when others =>

null;

end case;
Case versus if:

1) if is used when priority is desired.

2) in cases of more than three branches, case is
preferred.
 FOR Loops
The FOR loop is a sequential statement which is
used to execute a set of sequential statements
repeatedly.

for I in 0 to 3 loop
F(I) <= A(I) and B(3-I);
V := V xor A(I);
end loop;
The range can be ascending or descending, but the loop
parameter cannot change in increments greater than 1.

for I in 0 to 3 loop
F(I) <= A(I) and B(3-I);
V := V xor A(I);
end loop;

for I in 3 downto 0 loop

F(I) <= A(I) and B(3-I);
V := V xor A(I);
end loop;
The loop parameter is technically a constant, which
means that its value cannot be changed using an
assignment.

You cannot jump out of a loop by forcing the value of the

loop parameter

for I in 0 to 3 loop
F(I) <= A(I) and B(3-I);
V := V xor A(I);
if V = 'X' then
I := 4;
end if;
end loop;
FOR Loop are synthesized by making
multiple copies of the logic synthesized
inside the loop, one copy for each possible
value of the loop parameter.
 A(0)
process (A, B) F(0)
B(3)
variable V: Std_logic;
begin
A(1)
V := '0'; B(2)
F(1)

for I in 0 to 3 loop
F(I) <= A(I) and B(3-I); A(2)
V := V xor A(I); B(1)
F(2)

end loop;
G <= V; A(3)
F(3)
end process; B(0)

A(0)
A(1) G
A(2)
A(3)
 LOOPS
There are 3 different kinds of loop statements in
VHDL:
1) The while loop, tests a boolean condition at the
top of the loop, and only leaves the loop when
the condition is false.

while CONDITION loop

. . .
end loop;
2) The unbounded loop statement simply loops forever.
Loop
...
exit;
...
exit when CONDITION;
...
end loop;

3) for loop
Only the for loop is synthesizable, and that only if the
bounds are constant. The other loops can be useful in
test benches and behavioural models.
L1: for I in 0 to 7 loop

L2: for J in 0 to 7 loop

C := C + 1;

exit L2 when A(J) = B(I);

exit L1 when B(C) = 'U';

end loop L2;

end loop L1;

ClockGenerator_1: process
begin
for I in 1 to 1000 loop
Clock <= '0';
wait for 5 NS;
Clock <= '1';
wait for 10 NS;
end loop;
wait;
end process ClockGenerator_1;
ClockGenerator_2: process
begin
while NOW < 15 US loop
Clock <= '0';
wait for 5 NS;
Clock <= '1';
wait for 10 NS;
end loop;
wait;
end process ClockGenerator_2;
ClockGenerator_3: process
begin
loop
Clock <= '0';
wait for 5 NS;
Clock <= '1';
wait for 10 NS;
exit when NOW >= 15 US;
end loop;
wait;
end process ClockGenerator_3;
 Sequential Logic:Latches
library IEEE;
use IEEE.STD_LOGIC_1164.all;
entity LATCH is
port (ENB, D: in STD_LOGIC;
Q: out STD_LOGIC);
end;
architecture BEHAVIOUR of LATCH
is
begin
process (ENB, D)
begin
if ENB = '1' then
Q <= D;
end if;
end process;
end;
process (ENB, D)
begin
if ENB = '1' then
Q <= D;
end if;
end process;

The use of transparent latches is discouraged in some FPGA

technologies because the implementation requires asynchronous
feedback
 Edge Triggered Flip-Flops

This is achieved by using the signal attribute’EVENT as follows:

process (Clock)
begin

if Clock'EVENT and Clock = '1' then

Q0 <= D0;
Q1 <= D1;

end if;

end process;

Each signal has a signal attribute S’EVENT, which has a boolean value True
or False.
S’Evant changes during any delta in which there is an event on the
signal (i.e. signal changes value).

process
begin
...
S <= '1';
wait for 10 NS;
S <= '1';
wait for 10 NS;
S <= '0';
wait for 10 NS;
S <= '0';
...
 Rising_edge and Falling_edge Functions
Testing for a clock edge is such a common requirement that the package
std_logic_1164 includes standard functions Rising_edge and
Falling_edge .

The functions are given a signal of type STD_LOGIC and return a value
which is either True or False.

library IEEE;
use IEEE.STD_LOGIC_1164.all;
...
process (Clock)
begin

if RISING_EDGE(Clock) then
Q <= D;
end if;

end process;
process (Clock)
begin
if Clock'EVENT and Clock = '1' then
Q <= D;
end if;
end process;

process (Clock)
begin
if RISING_EDGE(Clock) then
Q <= D;
end if;
end process;

Not all synthesis tools support the RISING_EDGE and

FALLING_EDGE functions.
 Asynchronous Reset
Describing an asynchronous reset requires that the reset signal is added to the
sensitivity list of the process.

process (Clock, Reset)

begin

if Reset = '0' then

Count <= "00000000";

elsif RISING_EDGE(Clock) then

if Load = '1' then

Count <= Data;
else
Count <= Count + '1';
end if;
end if;
end process;
 Synchronous operations and Asynchronous operations can be
described within a single process.

...
signal Count :
STD_LOGIC_VECTOR(7 downto 0);
begin
process (Clock, Reset)
begin Load
Data Reset
if Reset = '0' then
Count <= "00000000"; Count
elsif RISING_EDGE(Clock) then
if Load = '1' then '1'
Count <= Data; >
else + Clock

Count <= Count + '1';

end if;
end if;
end process;
process (Clock)
begin
if RISING_EDGE(Clock) then

Reset
if Reset = '1' then
Q <= '0';
Q
Data
elsif Enable = '1' then
Enable
Q <= Data;
end if; Clock
end if; >
end process;
Wait until Statement

It suspends the process until a condition becomes True. The condition

can be any Boolean expression.

Another_Flipflop: process
begin Reset
wait until Clock = '1';
if Reset = '1' then Q
Q <= '0'; Data
else
Q <= D;
end if;
end process; Clock >
Another_Flipflop: process

begin
wait until RISING_EDGE(Clock);

if Reset = '1' then Reset

Q <= '0';
else Q
Q <= D; Data
end if;
end process;
Clock >
 Finite State Machines

• A finite state machine is a sequential logic circuit which moves

between a finite set of states, dependent upon the values of the
inputs and the previous state. The state transitions are
synchronized on a clock.

There are many ways to describe a finite state machine in VHDL.

The most convenient is with a process statement.
The state of the machine can be stored in a variable or signal, and the
possible states conveniently represented with an enumeration type.
architecture Explicit of FSM is
begin
process
type StateType is (Idle, Start, Stop, Clear);
variable State: StateType;
begin
wait until RISING_EDGE(Clock);
if Reset = '1' then
State := Idle; F <= '0'; G <= '1';
else
case State is
when Idle => State := Start; G <= '0';
when Start => State := Stop;
when Stop => State := Clear; F <= '1';
when Clear => State := Idle; F <= '0';
G <= '1';
end case;
end if;
end process;
end;
 State Names

Using enumeration type allows you to give symbolic names to the states,
but say nothing about the hardware implementation.

You should choose meaningful names, as this makes the VHDL code
easy to understand. The names will also be visible during simulation,
which makes debugging easier.

In practice, it is important that finite state machines are initialized by

means of an explicit reset signal.

Otherwise, there is no reliable way to get the VHDL and gate level
representations of the FSM into the same known state, and thus no way
to verify their equivalence.
architecture Explicit of FSM is
begin
process
type StateType is (Idle, Start, Stop, Clear);
variable State: StateType;
begin
wait until RISING_EDGE(Clock);
if Reset = '1' then
State := Idle; F <= '0'; G <= '1';
else
case State is
when Idle => State := Start; G <= '0';
when Start => State := Stop;
when Stop => State := Clear; F <= '1';
when Clear => State := Idle; F <= '0';
G <= '1';
end case;
end if;
end process;
end;
The following description of an FSM consists of a process synchronized
on a clock edge, and assigning the variable state (the state vector) and
signal F and G (the outputs).

Thus, four flip flops will be synthesized, two for the state vector, and one
each of the two outputs.
architecture Explicit of FSM is
begin
process
type StateType is (Idle, Start, Stop, Clear);
variable State: StateType;
begin
wait until RISING_EDGE(Clock);
if Reset = '1' then
State := Idle; F <= '0'; G <= '1';
else
case State is
when Idle => State := Start; G <= '0';
when Start => State := Stop;
when Stop => State := Clear; F <= '1';
when Clear => State := Idle; F <= '0';
G <= '1';
end case;
end if;
end process;
end;
architecture SeparateDecoding of FSM is
type StateType is (Idle, Start, Stop, Clear);
signal State: StateType;
begin
Change_state: process
begin
wait until RISING_EDGE(Clock);
if State = Clear or Reset = '1' then
State <= Idle;
elsif State = Idle then State <= Start;
elsif State = Start then State <= Stop;
else State <= Clear;
end if;
end process;
Output: process (State)
begin
F <= '0'; G <= '0';
if State = Clear then F <= '1';
elsif State = Idle then G <= '1';
end if;
end process;
end;
architecture RegistersPlusLogic of FSM is
type StateType is (Idle, Start, Stop, Clear);
signal State: StateType;
begin
Registers: process
begin
wait until RISING_EDGE(Clock);
if Reset = '0' then
State <= Idle;
else
State <= NextState;
end if;
end process;
C_logic: process (State)
begin
if State = Clear then NextState <= Idle;
F <= '1';
elsif State = Idle then NextState <= Start;
G <= '1';
elsif State = Start then NextState <= Stop;
else NextState <= Clear;
end if;
end process;
end;
 Concatenation
• The concatenation operator “&” is used to
join together two arrays end to end to make
one longer array

Signal A,B: STD_LOGIC_VECTOR(7 downto 0);

Signal F: STD_LOGIC_VECTOR(15 downto 0);

F<= A & B;
 • Concatenation can also be used to
concatenate arrays with single bits, or even
bits with bits.

Signal A,B,C: STD_LOGIC;

Signal F: STD_LOGIC_VECTOR(3 downto 0);

F(3 downto 1) <= (A & B) & C;

F(3) = A
F(2) = B
F(1) = C
F(0) = unchanged
• Shift and rotate operations can be
performed in one line by combining
concatenation with slice names.

Signal Reg: STD_LOGIC_VECTOR(7 downto 0);

Reg <= Reg(6 downto 0) & ‘0’;

Shift left one digit

Reg <= Reg(6 downto 0) & Reg(7);

Rotate left one digit

 Shift and Rotate
• In VHDL ’93 these operations can be
performed also using the shift and rotate
operators:
• Shift left / right logical : sll srl
• Shift left / right arithmatic : sla sra
• Rotate left / right : rol ror
Signal Reg: STD_LOGIC_VECTOR(7 downto 0);

Reg <= Reg(6 downto 0) & ‘0’;

Shift left one digit
Reg <= Reg sll 1;

Reg <= Reg(6 downto 0) & Reg(7);

Rotate left one digit
Reg <= Reg rol 1;
 Example
Variable A: STD_LOGIC_VECTOR (3 downto 0);
...
A:= “0110”;

• What is the value of this expression?

A(0) & ‘1’ & A(2 downto 1)

“0111”
 Operator Overloading
library IEEE;
use IEEE.STD_LOGIC_1164.all; • This is illigal
because the
entity ADDER is
operator “+”
port(A, B: in
STD_LOGIC_VECTOR(7 downto 0); is not defined
SUM: out to work on
STD_LOGIC_VECTOR(7 downto 0)); type
end; Std_logic_vec
tor.
architecture A1 of ADDER is
begin
SUM <= A + B;
end;
• The Std_logic_vector type is declared in the
package Std_logic_1164.

library IEEE;
use IEEE.STD_LOGIC_1164.all;

• The package also declares a set of operators

which work on the type.
• Defining operators to work on new types is
called operator overloading.
• There are also some operators that are implicitly
defined for all array types.
• These are the relational operators

>
>=
<
<=
=
/=
• “+” is not defined in std_logic_1164, but it
can be defined for std_logic_vector and put
in a package.

• All VHDL synthesis tools provide a

package that overloads certain arithmatic
operators on std_logic_vector or on related
types.
 library IEEE;
use IEEE.STD_LOGIC_1164.all;
• Package Numeric_Std
use IEEE.NUMERIC_STD.all;
includes new array types
entity ADDER is
singed and unsigned to
port(A, B: in UNSIGNED(7
represent numbers and
downto 0);
overloaded operators.
SUM: out UNSIGNED(7
downto 0));
end;
architecture A1 of ADDER is
begin
SUM <= A + B;
end;
signal A, B, C: STD_LOGIC;

F<= (A nand B ) xor (not C);

A
B F

signal A, B: STD_LOGIC_VECTOR;

F <= A + B ;
 Synthesis of Arithmetic
• The synthesis of the arithmetic operators depend
on the synthesis tool.
• Some tools will map the operators to discrete
gates. Others will make use of macro-cells
optimized for the target technology.
• A plus operator (+ ) can be implemented in
hardware using a ripple carry or a carry look-
ahead scheme. The implementation will be smaller
or faster, depending on which one you choose.
 F <= A + B ;

Discrete
Gates ADDER8

Non-optimal Optimized
connection of structure
cells of cells
 F <= A + B ;
Technology
ADDER8 independent
VHDL
code

C1: ADDER8 port map (A,B,B);

Vendor
ADDER8 Specific
VHDL
code
• If your synthesis tool can not map an arithmetic
operator to an optimized macro-cell, you will need
to instantiate the macro-cell directly in your
VHDL code.

• This would make your code vendor specific.

• However, sometimes you are forced to
compromise technology independence in favour of
efficiency.
 F <= A + B ;
+ Constraints

Ripple Carry
Carry Look-ahead
C1: SMALL-ADDER … (Ripple Carry Adder)
C1: FAST_ADDER8 … (Carry Look-Ahead Adder)

Ripple Carry
Carry Look-ahead
 Resource Sharing
• Resource sharing allows a single hardware to be
shared by more than one VHDL operator
• Some tools share resources automatically.
process (A, B, C, D, K)
begin C D
A B
if K then
Z <= A + B; -- 8 Bit
+ +
else
Z <= C + D; K

end if; Z

end process;
 process (A, B, C, D, K)
begin
if K then
Z <= A + B; -- 8 Bit
else
Z <= C + D;
end if;
end process;
A C B D
A B C D
K

+ +

K +
Z
Z
 process (A, B, C, D, K)
variable V1, V2 : ...
• If your tool does not do begin
resource sharing, you must if K then
rewrite your code to V1 := A;
achieve same results. V2 := B;
else
V1 := C;
V2 := D;
end if;
Z <= V1 + V2;
end process;
 State Machines (Again!)

Input Output outputs

Logic Logic
Inputs
• A State Machine should not have registers
at the output stage.
• In order to eliminate the output registers and
synthesize a pure state machine, we must
re-write the VHDL description.
• In order to synthesize the correct hardware,
we must split the description into at least
two processes.
 Exercise
• What is the minimum number of processes
necessary to describe this architecture for
synthesis?

A B C D

E
• 2 processes (AC) (BDE)
• 4 processes (AC) (B) (D) (E)
• 3 processes (A) (CE) (BD)
• 4 processes (A) (BC) (D) (E)
• 2 processes (ABCE) (D)
• 3 processes (A) (BCE) (D)
entity FSM1 is
port (Clock : in std_logic;
SlowRAN: in std_logic;
Read, Write: out std_logic); ST_Read ST_Write
end entity;
architecture RTL of FSM1 is
begin
SEQ_abd_COMB: process
type StateType is (ST_Read, ST_Write, ST_Delay);
variable State:StateType := ST_Read; ST_Delay
begin
wait until rising_edge*Clock);
case State is
when ST_Read => Read <= ‘1’;
Write <= ‘0’;
State := ST_Write;
when ST_Write => Read <= ‘0’;
Write <= ‘1’;
if (SlowRam = ‘1’) then
State := ST_Delay;
else
State := ST_Read
end if;
when ST_Delay => Read <= ‘0’;
Write <= ‘0’;
State := ST_Read;
end case;
end process SEQ_AND_COMB;
end architecture RTL;
entity FSM2 is
port (Clock : in std_logic;
SlowRAN: in std_logic;
Read, Write: out std_logic);
end entity;
architecture RTL of FSM2 is
type StateType is (ST_Read, ST_Write, ST_Delay);
signal CurrentState, NextState: StateType;
begin ST_Read ST_Write
SEQ: process (Clock)
begin
if rising_edge(Clock) then
if (Reset = ‘1’) then
CurrentState <= ST_Read;
else
CurrentState <= NextState; ST_Delay
end if;
end process SEQ;
COMB: process (CurrentState)
begin
case CurrentState is
when ST_Read => Read <= ‘1’;
Write <= ‘0’;
NextState := ST_Write;
when ST_Write => Read <= ‘0’;
Write <= ‘1’;
if (SlowRam = ‘1’) then
State := ST_Delay;
else
NextState := ST_Read
end if;
when ST_Delay => Read <= ‘0’;
Write <= ‘0’;
NextState := ST_Read;
end case;
end process COMB;
end architecture RTL;
 SEQ: process (Clock)
begin
if rising_edge(Clock) then
if (Reset = ‘1’) then
CurrentState <= ST_Read;
else
CurrentState <= NextState;
end if;
end process SEQ;
COMB: process (CurrentState)
begin
case CurrentState is
when ST_Read => Read <= ‘1’;
Write <= ‘0’;
NextState := ST_Write;
when ST_Write => Read <= ‘0’;
Write <= ‘1’;
if (SlowRam = ‘1’) then
State := ST_Delay;
else
NextState := ST_Read
end if;
when ST_Delay => Read <= ‘0’;
Write <= ‘0’;
NextState := ST_Read;
end case;
end process COMB;
SlowRam
Read
D Q

Q
Reset

D Q
Write

Clock