10.18  Problems

Chapter  start  Previous  page   Next page

10.18  Problems

* = Difficult ** = Very difficult *** = Extremely difficult

10.1 (Hello World, 10 min.) Set up a new, empty, directory (use mkdir VHDL , for example) to run your VHDL simulator (the exact details will depend on your computer and simulator). Copy the code below to a file called hw_1.vhd in your VHDL directory (leave out comments to save typing). Hint: Use the vi editor ( i inserts text, x deletes text, dd deletes a line, ESC :w writes the file, ESC :q quits) or use cat > hw_1.vhd and type in the code (use CTRL-D to end typing) on a UNIX machine. Remember to save in 'Text Only' mode (Frame or MS Word) on an IBM PC or Apple Macintosh.

Analyze, elaborate, and simulate your model (include the output in your answer). Comment on how easy or hard it was to follow the instructions to use the software and suggest improvements.

entity HW_1 is end; architecture Behave of HW_1 is
constant M : STRING := "hello, world"; signal Ch : CHARACTER := ' ';
begin process begin
	for i in M'RANGE loop Ch <= M(i); wait for 1 ns; end loop; wait; 
end process; end;

10.2 (Running a VHDL simulation, 20 min.) Copy the example from Section 10.1 into a file called Counter1.vhd in your VHDL directory (leave out the comments to save typing). Complete the compile (analyze), elaborate (build), and execute (initialize and simulate) or other equivalent steps for your simulator. After each step list the contents of your directory VHDL and any subdirectories and files that are created (use ls -alR on a UNIX system).

10.3 (Simulator commands, 10 min.) Make a "cheat sheet" for your simulator, listing the commands that can be used to control simulation.

10.4 (BNF addresses, 10 min.) Create a BNF description of a name including: optional title (Prof., Dr., Mrs., Mr., Miss, or Ms.), optional first name and middle initials (allow up to two), and last name (including unusual hyphenated and foreign names, such as Miss A-S. de La Salle, and Prof. John T. P. McTavish-f Fiennes). The lowest level constructs are letter ::= a-Z , '.' (period) and '-' (hyphen). Add BNF productions for a postal address in the form: company name, mail stop, street address, address lines (1 to 4), and country.

10.5 (BNF e-mail, 10 min.) Create a BNF description of a valid internet e-mail address in terms of letters, '@' , '.' , 'gov' , 'com ', 'org ', and 'edu' . Create a state diagram that "parses" an e-mail address for validity.

10.6 (BNF equivalence) Are the following BNF productions exactly equivalent? If they are not, produce a counterexample that shows a difference.

		term ::= factor { multiplying_operator factor }
		term ::= factor | term multiplying_operator factor

10.7 (Environment, 20 min.) Write a simple VHDL model to check and demonstrate that you can get to the IEEE library and have the environment variables, library statements, and such correctly set up for your simulator.

10.8 (Work, 20 min.) Write simple VHDL models to demonstrate that you can retrieve and use previously analyzed design units from the library work and that you can also remove design units from work . Explain how your models prove that access to work is functioning correctly.

10.9 (Packages, 60 min.) Write a simple package (use filename PackH.vhd ) and package body (filename PackB.vhd ). Demonstrate that you can store your package (call it MyPackage ) in the library work . Then store, move, or rename (the details will depend on your software) your package to a library called MyLibrary in a directory called MyDir , and use its contents with a library clause ( library MyLibrary ) and a use clause ( use MyLibrary.MyPackage.all ) in a testbench called PackTest (filename PackT.vhd ) in another directory MyWork . You may or may not be amazed at how complicated this can be and how poorly most software companies document this process.

10.10 (***IEEE Std 1164, 60 min.) Prior to VHDL-93 the xnor function was not available, and therefore older versions of the std_logic_1164 library did not provide the xnor function for STD_LOGIC types either (it was actually included but commented out). Write a simple model that checks to see if you have the newer version of std_logic_1164 . Can you do this without crashing the simulator?

You are an engineer on a very large project and find that your design fails to compile because your design must use the xnor function and the library setup on your company's system still points to the old IEEE std_logic_1164 library, even though the new library was installed. You are apparently the first person to realize the problem. Your company has a policy that any time a library is changed all design units that use that library must be rebuilt from source. This might require days or weeks of work. Explain in detail, using code, the alternative solutions. What will you recommend to your manager?

10.11 (**VHDL-93 test, 20 min.) Write a simple test to check if your simulator is a VHDL-87 or VHDL-93 environment--without crashing the simulator.

10.12 (Declarations, 10 min.) Analyze the following changes to the code in Section 10.8 and include the simulator output in your answers:

Uncomment the declarations for Bad100 and Bad4 in Declaration _1 .

Add the following to Constant_2:

signal wacky : wackytype (31 downto 0); -- wacky

Remove the library and use clause in Constant_2.

10.13 ( STRING type, 10 min.) Replace the write statement that prints the string " count=" in Text(Behave) in Section 10.6.3 with the following, compile it, and explain the result:

write(L, " count=" ); -- No type qualification.

10.14 (Sequential statements, 10 min.) Uncomment the following line in Wait_1(Behave) in Section 10.10, analyze the code, and explain the result:

wait on x(1 to v); -- v is a variable.

10.15 (VHDL logical operators, 10 min.)

Explain the problem with the following VHDL statement:

Z <= A nand B nand C;

Explain why this problem does not occur with this statement:

Z <= A and B and C;

What can you say about the logical operators: and , or , nand , nor , xnor , xor ?

Is the following code legal?

Z <= A and B or C;

10.16 (*Initialization, 45 min.) Consider the following code:

entity DFF_Plain is port (Clk, D : in BIT; Q : out BIT); end;
architecture Bad of DFF_Plain is begin process (Clk) begin
	if Clk = '0' and Clk'EVENT then Q <= D after 1 ns; end if;
end process; end;

Analyze and simulate this model using a testbench.

Rewrite architecture Bad using an equivalent process including a wait statement. Simulate this equivalent model and confirm the behaviors are identical.

What is the behavior of the output Q during initial execution of the process?

Why does this happen?

Why does this not happen with the following code:

architecture Good of DFF_Plain is
begin process begin wait until Clk = '0'; Q <= D after 1 ns;
end process; end;

10.17 (Initial and default values, 20 min.) Use code examples to explain the difference between: default expression, default value, implicit default value, initial value, initial value expression, and default initial value.

10.18 (Enumeration types, 20 min.) Explain the analysis results for the following:

type MVL4 is ('X', '0', '1', 'Z'); signal test : MVL4;
process begin
	test <= 1; test <= Z; test <= z; test <= '1'; test <= 'Z';
end process;

Alter the type declaration to the following, analyze your code again, and comment:

type Mixed4 is (X , '0', '1', Z);

10.19 (Type declarations, 10 min.) Correct these declarations:

type BadArray is array (0 to 7) of BIT_VECTOR;
type Byte is array (NATURAL range 7 downto 0) of BIT;
subtype BadNibble is Byte(3 downto 0);
type BadByte is array (range 7 downto 0) of BIT;

10.20 (Procedure parameters, 10 min.) Analyze the following package; explain and correct the error. Finally, build a testbench to check your solution.

package And_Pkg_Bad is procedure V_And(a, b : BIT; c: out BIT); end;
package body And_Pkg_Bad is
procedure V_And(a,b : BIT;c : out BIT) is begin c <= a and b;end;
end And_Pkg_Bad;

10.21 (Type checking, 20 min.) Test the following code and explain the results:

type T is INTEGER range 0 to 32; variable a: T;
a := (16 + 17) - 12; a := 16 - 12 + 17; a := 16 + (17 - 12);

10.22 (Debugging VHDL code, 30 min.) Find and correct the errors in the following code. Create a testbench for your code to check that it works correctly.

entity UpDownCount_Bad is 
port(clock, reset, up: STD_LOGIC; D: STD_LOGIC_VECTOR (7 to 0));
end UpDownCount_Bad;
architecture Behave of UpDownCount_Bad is
begin process (clock, reset, up); begin 
if (reset = '0') then D <= '0000000'; 
elseif (rising_edge(clock)) then
if (up = 1) D <= D+1; else D <= D-1; end if;
end if; end process; end Behave;

10.23 (Subprograms, 20 min.) Write and test subprograms for these declarations:

function Is_X_Zero (signal X : in BIT) return BIT;
procedure Is_A_Eq_B (signal A, B : BIT; signal Y : out BIT);

10.24 (Simulator error messages, 10 min.) Analyze and attempt to simulate Arithmetic_2(Behave) from Section 10.12 and compare the error message you receive with that from the MTI simulator (not all simulators are as informative). There are no standards for error messages.

10.25 (Exhaustive property of case statement, 30 min.) Write and simulate a testbench for the state machine of Table 10.8 and include your results. Is every state transition tested by your program and is every transition covered by an assignment statement in the code? (Hint: Think very carefully.) Repeat this exercise for the state machine in Section 10.10.6.

10.26 (Default values for inputs, 20 min.) Replace the interface declaration for entity Half_Adder in Section 10.5 with the following (to remove the default values):

port (X, Y: in BIT ; Sum, Cout: out BIT);

Attempt to compile, elaborate, and simulate configuration Simplest (the other entities needed, AndGate and XorGate , must already be in work or in the same file). You should get an error at some stage (different systems find this error at different points--just because an entity compiles, that does not mean it is error-free).

The LRM says "... A port of mode in may be unconnected ...only if its declaration includes a default expression..." [VHDL 93LRM1.1.1.2].

We face a dilemma here. If we do not drive inputs with test signals and leave an input port unconnected, we can compile the model (since it is syntactically correct) but the model is not semantically correct. On the other hand, if we give the inputs default values, we might accidentally forget to make a connection and not notice.

10.27 (Adder generation, 10 min.) Draw the schematic for Adder_1(Structure) of Section 10.13.7, labeling each instance with the VHDL instance name.

10.28 (Generate statement, 20 min.) Draw a schematic corresponding to the following code (label the cells with their instance names):

B1: block begin L1 : C port map (T, B, A(0), B(0)) ;
L2: for i in 1 to 3 generate L3 : for j in 1 to 3 generate
L4: if i+j > 4 generate L5: C port map (A(i-1), B(j-1), A(i), B(j)) ;
end generate; end generate; end generate;
L6: for i in 1 to 3 generate L7: for j in 1 to 3 generate
L8: if i+j < 4 generate L9: C port map (A(i+1), B(j+1), A(i), B(j)) ;
end generate; end generate; end generate;
end block B1;

Rewrite the code without generate statements. How would you prove that your code really is exactly equivalent to the original?

10.29 (Case statement, 20 min.) Create a package ( my_equal ) that overloads the equality operator so that 'X' = '0' and 'X' = '1' are both TRUE . Test your package. Simulate the following design unit and explain the result.

entity Case_1 is end; architecture Behave of Case_1 is
signal r : BIT; use work.my_equal.all;
begin process variable twobit:STD_LOGIC_VECTOR(1 to 2); begin 
	twobit := "X0";
	case twobit is 
		when "10" => r <= '1';
		when "00" => r <= '1'; 
		when others => r <= '0';
	end case; wait;
end process; end;

10.30 (State machine) Create a testbench for the state machine of Section 10.2.5.

10.31 (Mealy state machine, 60 min.) Rewrite the state machine of Section 10.2.5 as a Mealy state machine (the outputs depend on the inputs and on the current state).

10.32 (Gate-level D flip-flop, 30 min.) Draw the schematic for the following D flip-flop model. Create a testbench (check for correct operation with combinations of Clear , Preset , Clock , and Data ). Have you covered all possible modes of operation? Justify your answer of yes or no.

architecture RTL of DFF_To_Test is
signal A, B, C, D, QI, QBarI : BIT; begin 
A <= not (Preset and D and B) after 1 ns;
B <= not (A and Clear and Clock) after 1 ns;
C <= not (B and Clock and D) after 1 ns;
D <= not (C and Clear and Data) after 1 ns;
QI <= not (Preset and B and QBarI) after 1 ns;
QBarI <= not (QI and Clear and C) after 1 ns;
Q <= QI; QBar <= QBarI;

10.33 (Flip-flop model, 20 min.) Add an asynchronous active-low preset to the D flip-flop model of Table 10.3. Generate a testbench that includes interaction of the preset and clear inputs. What issue do you face and how did you solve it?

10.34 (Register, 45 min.) Design a testbench for the register of Table 10.4. Adapt the 8-bit register design to a 4-bit version with the following interface declaration:

entity Reg4 is port (D : in  STD_LOGIC_VECTOR(7 downto 0);
Clk,Pre,Clr : in STD_LOGIC;Q,QB : out STD_LOGIC_VECTOR(7 downto 0));
end Reg8;

Create a testbench for your 4-bit register with the following component declaration:

 component DFF
port(Preset,Clear,Clock,Data:STD_LOGIC;Q,QBar:out STD_LOGIC_VECTOR);
end component;

10.35 (*Conversion functions, 30 min.) Write a conversion function from
NATURAL to STD_LOGIC_VECTOR using the following declaration:

function Convert (N, L: NATURAL) return STD_LOGIC_VECTOR;
-- N is NATURAL, L is length of STD_LOGIC_VECTOR

Write a similar conversion function from STD_LOGIC_VECTOR to NATURAL:

function Convert (B: STD_LOGIC_VECTOR) return NATURAL;

Create a testbench to test your functions by including them in a package.

10.36 (Clock procedure, 20 min.) Design a clock procedure for a two-phase clock (C1, C2) with variable high times ( HT1 , HT2 ) and low times ( LT1 , LT2 ) and the following interface. Include your procedure in a package and write a model to test it.

procedure Clock (C1, C2 : out STD_LOGIC; HT1, HT2, LT1, LT2 : TIME);

10.37 (Random number, 20 min.) Design a testbench for the following procedure:

procedure uniform (seed : inout INTEGER range 0 to 15) is
	variable x : INTEGER;
	begin x := (seed*11) + 7; seed := x mod 16;
end uniform;

10.38 (Full-adder, 30 min.) Design and test a behavioral model of a full adder with the following interface:

entity FA is port (X, Y, Cin : STD_LOGIC; Cout, Sum : out STD_LOGIC);

Repeat the exercise for inputs and outputs of type UNSIGNED .

10.39 (8-bit adder testbench, 60 min.) Write out the code corresponding to the generate statements of Adder_1 ( Structure ) in Section 10.13.7. Write a testbench to check your adder. What problems do you encounter? How thorough do you believe your tests are?

10.40 (Shift-register testbench, 60 min.) Design a testbench for the shift register of Table 10.4. Convert this model to use STD_LOGIC types with the following interface:

entity ShiftN is 

10.41 (Multiplier, 60 min.) Design and test a multiplier with the following interface:

entity Mult8 is 
port (A, B : STD_LOGIC_VECTOR(3 downto 0); 
Start, CLK, Reset : in STD_LOGIC;
Result : out STD_LOGIC_VECTOR(7 downto 0); Done : out BIT);

Create testbench code to check your model.

Catalog each compile step with the syntax errors as you debug your code.

Include a listing of the first code you write together with the final version.

An interesting class project is to collect statistics from other students working on this problem and create a table showing the types and frequency of syntax errors made with each compile step, and the number of compile steps required. Does this information suggest ways that you could improve the compiler, or suggest a new type of tool to use when writing VHDL?

10.42 (Port maps, 5 min.) What is wrong with this VHDL statement?

U1 : nand2 port map (a <= set, b <= qb, c <= q);

10.43 (DRIVING_VALUE, 15 min.) Use the VHDL-93 attribute Clock'DRIVING_VALUE to rewrite the following clock generator model without using a temporary variable.

entity ClockGen_2 is port (Clock : out BIT); end;
architecture Behave of ClockGen_2 is
begin process variable Temp : BIT := '1'; begin
	Temp := not Temp ; Clock <= Temp after 10 ns; wait for 10 ns;
	if (now > 100 ns) then wait; end if; end process;

10.44 (Records, 15 min.) Write an architecture (based on the following skeleton) that uses the record structure shown:

entity Test_Record_1 is end; architecture Behave of Test_Record_1 is
begin process type Coordinate is record X, Y : INTEGER; end record;
-- a record declaration for an attribute declaration:
attribute Location:Coordinate; -- an attribute declaration
begin wait; end process; end Behave;

10.45 (**Communication between processes, 30 min.) Explain and correct the problem with the following skeleton code:

variable v1 : INTEGER := 1; process begin v1 := v1+3; wait; end process;
process variable v2 : INTEGER := 2; begin v2 := v1  ; wait; end process;

10.46 (*Resolution, 30 min.) Explain and correct the problems with the following:

entity R_Bad_1 is port (i : in BIT; o out BIT); end;
architecture Behave of R_Bad_1 is
begin o <= not i after 1 ns; o <= i after 2 ns; end;

10.47 (*Inputs, 30 min.) Analyze the following and explain the result:

entity And2 is port (A1, A2: in BIT; ZN: out BIT); end;
architecture Simple of And2 is begin ZN <= A1 and A2; end;
entity Input_Bad_1 is end; architecture Netlist of Input_Bad_1 is
component And2 port (A1, A2 : in BIT; ZN : out BIT); end component;
signal X, Z : BIT begin G1 : And2 port map (X, X, Z); end;

10.48 (Association, 15 min.) Analyze the following and explain the problem:

entity And2 is port (A1, A2 : in BIT; ZN : out BIT); end;
architecture Simple of And2 is begin ZN <= A1 and A2; end;
entity Assoc_Bad_1 is port (signal X, Y : in BIT; Z : out BIT); end;
architecture Netlist of Assoc_Bad_1 is
component And2 port (A1, A2 : in BIT; ZN : out BIT); end component;
G1: And2 port map (X, Y, Z);
G2: And2 port map (A2 => Y, ZN => Z, A1 => X);
G3: And2 port map (X, ZN => Z, A2 => Y);

10.49 (Modes, 30 min.) Analyze and explain the errors in the following:

entity And2 is port (A1, A2 : in BIT; ZN : out BIT); end;
architecture Simple of And2 is begin ZN <= A1 and A2; end;
entity Mode_Bad_1 is port (X : in BIT; Y : out BIT; Z : inout BIT); end;
architecture Netlist of Mode_Bad_1 is
component And2 port (A1, A2 : in BIT; ZN : out BIT); end component;
begin G1 : And2 port map (X, Y, Z); end;
entity Mode_Bad_2 is port (X : in BIT; Y : out BIT; Z : inout BIT); end;
architecture Netlist of Mode_Bad_1 is
component And2 port (A1, A2 : in BIT; ZN : inout BIT); end component;
begin G1 : And2 port map (X, Y, Z); end;

10.50 (*Mode association, 60 min.) Analyze and explain the errors in the following code. The number of errors, types of error, and the information in the error messages given by different simulators vary tremendously in this area.

entity Allmode is port 
(I : in BIT; O : out BIT; IO : inout BIT; B : buffer BIT); 
architecture Simple of Allmode is begin O<=I; IO<=I; B<=I; end;
entity Mode_1 is port 
(I : in BIT; O : out BIT; IO : inout BIT; B : buffer BIT); 
architecture Netlist of Mode_1 is
component Allmode port 
(I : in BIT; O : out BIT; IO : inout BIT; B : buffer BIT); end component;
G1:Allmode port map (I , O , IO, B ); 
G2:Allmode port map (O , IO, B , I ); 
G3:Allmode port map (IO, B , I , O ); 
G4:Allmode port map (B , I , O , IO); 

10.51 (**Declarations, 60 min.) Write a tutorial (approximately two pages of text, five pages with code) with examples explaining the difference between: a component declaration, a component configuration, a configuration declaration, a configuration specification, and a block configuration.

10.52 (**Guards and guarded signals, 60 min.) Write some simple models to illustrate the use of guards, guarded signals, and the disconnect statement. Include an experiment that shows and explains the use of the implicit signal GUARD in assignment statements.

10.53 (** Std_logic_1164 , 120 min.) Write a short (two pages of text) tutorial, with (tested) code examples, explaining the std_logic_1164 types, their default values, the difference between the 'ulogic' and 'logic' types, and their vector forms. Include an example that shows and explains the problem of connecting a std_logic_vector to a std_ulogic_vector.

10.54 (Data swap, 20 min.) Consider the following code:

library ieee; use ieee.std_logic_1164.all;
package config is
type type1 is record
f1 : std_logic_vector(31 downto 0); f2 : std_logic_vector(3 downto 0);
end record;
type type2 is record
f1 : std_logic_vector(31 downto 0); f2 : std_logic_vector(3 downto 0);
end record;
end config;
library ieee; use ieee.STD_LOGIC_1164.all; use work.config.all;
entity Swap_1 is
port (Data1 : type1; Data2 : type2; sel : STD_LOGIC;
Data1Swap : out type1; Data2Swap : out type2); end Swap_1;
architecture Behave of Swap_1 is begin
Swap: process (Data1, Data2, sel) begin case sel is
when '0' => Data1Swap <= Data1; Data2Swap <= Data2;
when others => Data1Swap <= Data2; Data2Swap <= Data1;
end case; end process Swap; end Behave;

Compile this code. What is the problem? Suggest a fix. Now write a testbench and test your code. Have you considered all possibilities?

10.55 (***RTL, 30 min.) "RTL stands for register-transfer level. ...when referencing VHDL, the term means that the description includes only concurrent signal assignment statements and possibly block statements. In particular, VHDL data flow descriptions explicitly do not contain either process statements (which describe behavior) or component instantiation statements (which describe structure)" (Dr. VHDL from VHDL International).

With your knowledge of process statements and components, comment on Dr. VHDL's explanation.

In less than 100 words offer your own definition of the difference between RTL, data flow, behavioral, and structural models.

10.56 (*Operators mod and rem , 20 min.) Confirm and explain the following:

 	i1 := (-12) 					rem 		7; 					-- i1 = -5
	i2 := 12 					rem 		(-7); 					-- i2 =  5
	i3 := (12) 					rem 		(-7); 					-- i3 = -5
	i4 := 12 					mod 		7; 					-- i4 =  5
	i5 := (-12) 					mod 		7; 					-- i5 =  2
	i6 := 12 					mod 		(-7); 					-- i6 = -2
	i7 := (12) 					mod 		(-7); 					-- i7 = -5

Evaluate -5 rem 2 and explain the result.

10.57 (***Event and stable, 60 min.) Investigate the differences between clk'EVENT and clk'STABLE . Write a minitutorial (in the form of a "cheat sheet") with examples showing the differences and potential dangers of using clk'STABLE .

10.58 (PREP benchmark #2, 60 min.) The following code models a benchmark circuit used by PREP to measure the capacity of FPGAs. Rewrite the concurrent signal assignment statements (labeled mux and comparator) as equivalent processes. Draw a datapath schematic corresponding to PREP2(Behave_1). Write a testbench for the model. Finally (for extra credit) rewrite the model and testbench to use STD_LOGIC instead of BIT types.

library ieee; use ieee.STD_LOGIC_1164.all;
use ieee.NUMERIC_BIT.all; use ieee.NUMERIC_STD.all;
entity PREP2 is 
port(CLK,Reset,Sel,Ldli,Ldhi : BIT; D1,D2 : STD_LOGIC_VECTOR(7 downto 0);
	DQ:out STD_LOGIC_VECTOR(7 downto 0));
architecture Behave_1 of PREP2 is 
signal EQ : BIT; signal y,lo,hi,Q_i : STD_LOGIC_VECTOR(7 downto 0);
outputDriver: Q <= Q_i;
mux: with Sel select y <= hi when '0', D1 when '1';
comparator: EQ <= '1' when Q_i = lo else '0';
	register: process(Reset, CLK) begin
		if Reset = '1' then hi <= "00000000"; lo <= "00000000";
		elsif CLK = '1' and CLK'EVENT then
			if Ldhi='1' then hi<=D2;end if;if Ldlo='1' then lo<=D2;end if;
		end if;
	end process register;
	counter: process(Reset, CLK) begin
		if Reset = '1' then Q_i <= "00000000";
		elsif CLK = '1' and CLK'EVENT then
			if EQ = '1' then Q_i <= y;
			elsif EQ = '0' then Q_i  <= Q_i + "00000001";
			end if;
		end if;
	end process counter;

10.59 (PREP #3, state machine) Draw the state diagram for the following PREP benchmark (see Problem 10.58). Is this a Mealy or Moore machine? Write a testbench and test this code.

library ieee; use ieee.STD_LOGIC_1164.all; 
entity prep3_1 is port(Clk, Reset: STD_LOGIC;
	I : STD_LOGIC_VECTOR(7 downto 0); O : out STD_LOGIC_VECTOR(7 downto 0));
end prep3_1;
architecture Behave of prep3_1 is
	type STATE_TYPE is (sX,s0,sa,sb,sc,sd,se,sf,sg);
	signal state : STATE_TYPE; signal Oi : STD_LOGIC_VECTOR(7 downto 0);
	O <= Oi;
	process (Reset, Clk) begin 
		if (Reset = '1') then state <= s0; Oi <= (others => '0');
		elsif rising_edge(Clk) then 
			case state is 
				when s0 =>
					if (I = X"3c") then state <= sa; Oi <= X"82"; 
				 	else state <= s0; Oi <= (others => '0');
					end if;
				when sa =>
					if (I = X"2A") then state <= sc; Oi <= X"40"; 
					elsif (I = X"1F") then state <= sb; Oi <= X"20"; 
					else state <= sa; Oi <= X"04"; 
					end if;
				when sb =>
					if (I = X"AA") then state <= se; Oi <= X"11";
				 	else state <= sf; Oi <= X"30"; 
					end if;
				when sc => state <= sd; Oi <= X"08"; 
				when sd => state <= sg; Oi <= X"80"; 
				when se => state <= s0; Oi <= X"40"; 
				when sf => state <= sg; Oi <= X"02"; 
				when sg => state <= s0; Oi <= X"01"; 
				when others => state <= sX; Oi <= (others => 'X'); 
			end case;
		end if;
	end process;

10.60 (Edge detection, 30 min) Explain the construction of the IEEE 1164 function to detect the rising edge of a signal, rising_edge(s) . List all the changes in signal s that correspond to a rising edge.

 function rising_edge (signal s : STD_ULOGIC) return BOOLEAN is
	begin return 
	(s'EVENT and (To_X01(s) = '1') and (To_X01(s'LAST_VALUE) = '0')); end;

10.61 (*Real, 10 min.) Determine the smallest real in your VHDL environment.

10.62 (*Stop, 30 min.) How many ways are there to stop a VHDL simulator?

10.63 (*Arithmetic package, 60 min.) Write a function for an arithmetic package to subtract two's complement numbers. Create a test bench to check your function. Your declarations in the package header should look like this:

type TC is array (INTEGER range <>) of STD_LOGIC;
function "-"(L : TC; R : TC) return TC;

10.64 (***Reading documentation, hours) There are a few gray areas in the interpretation of the VHDL-87 LRM some of which were clarified in the VHDL-93 revision. One VHDL system has a "compatibility mode" that allows alternative interpretations. For each of the following "issues" taken from the actual tool documentation try to interpret what was meant, determine the interpretation taken by your own software, and then rewrite the explanation clearly using examples.

* "Unassociated variable and signal parameters. Compatibility mode allows variable and signal parameters to subprograms to be unassociated if they have a default value. Otherwise, an error is generated."

Example answer: Consider the following code:

package Util_2 is
procedure C(signal Clk : out BIT; signal P : TIME := 10 ns); 
end Util_2;
package body Util_2 is
procedure C(signal Clk : out BIT; signal P : TIME := 10 ns) is
begin loop Clk <= '1' after P/2, '0' after P;
wait for P; end loop; end; end Util_2;
entity Test_Compatibility_1 is end; use work.Util_2.all;
architecture Behave of Test_Compatibility_1 is 
signal v,w,x,y,z : BIT; signal s : TIME := 5 ns;
begin process variable v : TIME := 5 ns; begin
C(v, s); 							-- parameter s is OK since P is declared as signal 
-- C(w, v); 							-- would be OK if P is declared as variable instead
-- C(x, 5 ns); 							-- would be OK if P is declared as constant instead
-- C(y); 							-- unassociated, an error if P is signal or variable
-- C(z,open); 							-- open, an error if P is signal or variable
end process; end;

The Compass Scout simulator (which does not have a compatibility mode) generates an error during analysis if a signal or variable subprogram parameter is open or unassociated (a constant subprogram parameter may be unassociated or open).

* "Allow others in an aggregate within a record aggregate. The LRM [] defines nine situations where others may appear in an aggregate. In compatibility mode, a tenth case is added. In this case, others is allowed in an aggregate that appears as an element association in a record element."

* " BIT'('1') parsed as BIT ' ('1') . The tick ( ' ) character is being used twice in this example. In the first case as an attribute indicator, in the second case, to form a character literal. Without the compatibility option, the analyzer adopts a strict interpretation of the LRM, and without white space around the first tick, the fragment is parsed as BIT '('1') , that is, the left parenthesis ( '(' ) is the character literal."

** "Generate statement declarative region. Generate statements form their own declarative region. In compatibility mode, configuration specifications will apply to items being instantiated within a generate statement."

** "Allow type conversion functions on open parameters. If a parameter is specified as open, it indicates a parameter without an explicit association. In such cases, the presence of a type conversion function is meaningless. Compatibility mode allows the type conversion functions."

*** "Entity class flexibility. Section [3.1.2] of the LRM defines the process of creating a new integer type. The type name given is actually assigned to a subtype name, related to an anonymous base type. This implies that the entity class used during an attribute specification [LRM 5.1] should indicate subtype, rather than type. Because the supplied declaration was type rather than subtype, compatibility mode allows type."

*** "Allowing declarations beyond an all/others specification. Section [5.1] of the LRM states that the first occurrence of the reserved word all or others in an attribute specification terminates the declaration of the related entity class. The LRM declares that the entity/architecture and package/package body library units form single declaration regions [LRM 10.1] that are the concatenation of the two individual library declarative regions. For example, if a signal attribute specification with all or others was specified in the entity, it would be impossible to declare a signal in the architecture. In compatibility mode, this LRM limitation is removed."

*** "User-defined attributes on overloaded functions. In compatibility mode, user-defined attributes are allowed to be associated with overloaded functions. Note: Even in compatibility mode, there is no way to retrieve the different attributes."

10.65 (*1076 interpretations, 30 min.) In a DAC paper, the author writes: `It was experienced that (company R) might have interpreted IEEE 1076 differently than (company S) did, e.g. concatenations (&) are not allowed in "case selector" expressions for (company S).' Can you use concatenation in your VHDL tool for either the expression or choices for a case statement?

10.66 (**Interface declarations, 15 min.) Analyze the following and comment:

entity Interface_1 is 
	generic (I : INTEGER; J : INTEGER := I; K, L : INTEGER);
	procedure X(P, Q : INTEGER; R : INTEGER range P to Q);
	procedure Y(S : INTEGER range K to L); 
end Interface_1;

10.67 (**Wait statement, 10 min.) Construct the sensitivity set and thus the sensitivity list for the following wait statement (that is, rewrite the wait statement in the form wait on sensitivity_list until condition ).

entity Complex_Wait is end;
architecture Behave of Complex_Wait is
	type A is array (1 to 5) of BOOLEAN; 
	function F (P : BOOLEAN) return BOOLEAN;
	signal S : A; signal i, j : INTEGER range 1 to 5;
	begin process begin
		wait until F(S(3)) and (S(i) or S(j));
	end process;

10.68 (**Shared variables, 20 min.) Investigate the following code and comment:

architecture Behave of Shared_1 is 
subtype S is INTEGER range 0 to 1; shared variable C : S := 0; begin
process begin C := C + 1; wait; end process;
process begin C := C - 1; wait; end process;

10.69 (Undocumented code and ranges, 20 min.) Explain the purpose of the following function (part of a package from a well-known synthesis company) with a parameter of type SIGNED. Write a testbench to check your explanation. Investigate what happens when you call this function with a string-literal argument, for example with the statement X <= IM("11100"). What is the problem and why does it happen? Rewrite the code, including documentation, to avoid this problem.

type SIGNED is array (NATURAL range <> ) of BIT;
function IM (L : SIGNED) return INTEGER is variable M : INTEGER;
begin M := L'RIGHT-1;
	for i in L'LEFT-1 downto L'RIGHT loop
		if (L(i) = (not L(L'LEFT))) then M := i; exit; end if;
	end loop; return M;

10.70 (Timing parameters, 20 min.) Write a model and a testbench for a two-input AND gate with separate rising (tpLH) and falling (tpHL) delays using the following interface:

entity And_Process is
generic (tpLH, tpHL : TIME); port (a, b : BIT; z : out BIT) end;

10.71 (Passive code in entities, 30 min.) Write a procedure (CheckTiming, part of a package Timing_Pkg ) to check that two timing parameters (tPLH and tPHL) are both greater than zero. Include this procedure in a two-input AND gate model ( And_Process ). Write a testbench to show your procedure and gate model both work. Rewrite the entity for And_Process to include the timing check as part of the entity declaration. You are allowed to include passive code (no assignments to signals and so on) directly in each entity. This avoids having to include the timing checks in each architecture.

10.72 (Buried code, 30 min.) Some companies bury instructions to the software within their packages. Here is an example of part of the arithmetic package from an imaginary company called SissyN:

function UN_plus(A, B : UN) return UN is
variable CRY : STD_ULOGIC; variable X,SUM : UN (A'LEFT downto 0);
-- pragma map_to_operator ADD_UNS_OP
-- pragma type_function LEFT_UN_ARG
-- pragma return_port_name Z
-- sissyn synthesis_off
if (A(A'LEFT) = 'X' or B(B'LEFT) = 'X') then SUM := (others => 'X'); 
end if;
-- sissyn synthesis_on
CRY := '0'; X := B;
for i in 0 to A'LEFT loop
SUM(i) := A(i) xor X(i) xor carry;
CRY := (A(i) and X(i)) or (A(i) and CRY) or (CRY and X(i));
end loop; return SUM;

Explain what this function does. Can you now hazard a guess at what each of the comments means? What are the repercussions of using comments in this fashion?

10.73 (*Deferred constants, 15 min.) "If the assignment symbol ':=' followed by an expression is not present in a constant declaration, then the declaration declares a deferred constant. Such a constant declaration may only appear in a package declaration. The corresponding full constant declaration, which defines the value of the constant, must appear in the body of the package" [VHDL 93LRM4.3.1.1].

package Constant is constant s1, s2 : BIT_VECTOR; end Constant;
package body Constant is
constant s0 : BIT_VECTOR := "00"; constant s1 : BIT_VECTOR := "01";
end Constant;

It is tempting to use deferred constants to hide information. However, there are problems with this approach. Analyze the following code, explain the results, and correct the problems:

entity Deferred_1 is end; architecture Behave of Deferred_1 is
use work.all; signal y,i1,i2 : INTEGER; signal sel : INTEGER range 0 to 1;
begin with sel select y <= i1 when s0, i2 when s1; end;

10.74 (***Viterbi code, days) Convert the Verilog model of the Viterbi decoder in Chapter 11 to VHDL. This problem is tedious without the help of some sort of Verilog to VHDL conversion process. There are two main approaches to this problem. The first uses a synthesis tool to read the behavioral Verilog and write structural VHDL (the Compass ASIC Synthesizer can do this, for example). The second approach uses conversion programs (Alternative System Concepts Inc. at http://www.ascinc.com is one source). Some of these companies allow you to e-mail code to them and they will automatically return a translated version.

10.75 (*Wait statement, 30 min.) Rewrite the code below using a single wait statement and write a testbench to prove that both approaches are exactly equivalent:

entity Wait_Exit is port (Clk : in BIT); end;
architecture Behave of Wait_Exit is
	begin process begin
		loop wait on Clk; exit when Clk = '1'; end loop;
	end process;

10.76 (Expressions, 10 min.) Explain and correct the problems with the following:

variable b : BOOLEAN; b := "00" < "11";
variable bv8 : BIT_VECTOR (7 downto 0) := "1000_0000";

10.77 (Combinational logic using case statement, 10 min.) A Verilog user suggests the following method to model combinational logic. What are the problems with this approach? Can you get it to work?

entity AndCase is port (a, b : BIT; y : out BIT); end;
architecture Behave of AndCase is begin process (a , b) begin
	case a & b is 
	 when '1'&'1' => y <= '1'; when others => y <= '0';
	end case;
end process; end;

10.78 (*Generics and back-annotation, 60 min.)

Construct design entities And_3(Behave), a two-input AND gate, and Xor_3(Behave) , a two-input XOR gate. Include generic constants to model the propagation delay from each input to the output separately. Use the following entity declaration for And_3:

entity And_3 is port (I1, I2 : BIT; O : out BIT);
	generic (I1toO, I2toO : DELAY_LENGTH := 0.4 ns); end;

Create and test a package, P_1, that contains And_3 and Xor_3 as components.

Create and test a design entity Half_Adder_3 (Structure_3) that uses P_1, with the following interface:

entity Half_Adder_3 is port (X, Y : BIT; Sum, Carry : out BIT); end;

Modify and test the architecture Structure_3 for Half_Adder_3 so that you can use the following configuration:

configuration Structure_3 of Half_Adder_3 is
for Structure_3
for L1 : XOR generic map (0.66 ns,0.69 ns); end for; 
for L2 : AND generic map (0.5 ns, 0.6 ns) port map (I2 => HI); end for;
end for; end;

10.79 (SNUG'95, *60 min.) In 1995 John Cooley organized a contest between VHDL and Verilog for ASIC designers. The goal was to design the fastest 9-bit counter in under one hour using Synopsys synthesis tools and an LSI Logic vendor technology library. The VHDL interface is as follows:

library ieee; use ieee.std_logic_1164.all;
-- use ieee.std_logic_arith.all; -- substitute your package here
entity counter is port (
data_in    : in std_logic_vector(8 downto 0);
up         : in std_logic;
down       : in std_logic;
clock      : in std_logic;
count_out  : inout std_logic_vector(8 downto 0);
carry_out  : out std_logic;
borrow_out : out std_logic;
parity_out : out std_logic ); end counter;
architecture example of counter is begin
-- insert your design here 
end example;

The counter is positive-edge triggered, counts up with up = '1' and down with down = '1' . The contestants had the advantage of a predefined testbench with a set of test vectors, you do not. Design a model for the counter and a testbench. How confident are you that you have thoroughly tested your model? (In the real contest none of the VHDL contestants managed to even complete a working design in under one hour. In addition, the VHDL experts that had designed the testbench omitted a test case for one of the design specifications.)

10.80 (*A test procedure, 45 min.) Write a procedure all (for a package test ) that serially generates all possible input values for a signal spaced in time by a delay, dly . Use the following interface:

library ieee; use ieee.std_logic_1164.all; package test is
procedure all (signal SLV : out STD_LOGIC_VECTOR; dly : in TIME);
end package test ;

10.81 (Direct instantiation, 20 min.) Write an architecture for a full-adder, entity Full_Adder_2, that directly instantiates units And_2(Behave) and Xor_2(Behave). This is only possible in a VHDL-93 environment.

entity And_2 is port (i1, i2 : BIT; y : out BIT); end;
entity Xor_2 is port (i1, i2 : BIT; y : out BIT); end;
entity Full_Adder_2 is port (a, b, c : BIT ; sum, cout : out BIT); end;

10.82 (**Shift operators for 1164, 60 min.) Write a package body to implement the VHDL-93 shift operators, sll and srl, for the type STD_LOGIC_VECTOR. Use the following package header:

package 1164_shift is
function "sll"(x : STD_LOGIC_VECTOR; n : INTEGER)
function "srl"(x : STD_LOGIC_VECTOR; n : INTEGER)
end package 1164_shift;

10.83 (**VHDL wait statement, 60 min.) What is the problem with the following VHDL code? Hint: You may need to consult the VHDL LRM.

procedure p is begin wait on b; end;
process (a) is begin procedure p; end process;

10.84 (**Null range, 45 min.) A range such as 1 to -1 or 0 downto 1 is a null range ( 0 to 0 is a legal range). Write a one-page summary on null ranges, including code examples. Is a null range treated as an ascending or descending range?

10.85 (**Loops, 45 min.) Investigate the following issues with loops, including code examples and the results of analysis and simulation:

Try to alter the loop parameter within a loop. What happens?

What is the type of the loop parameter?

Can the condition inside a loop depend on a loop parameter?

What happens in a for loop if the range is null?

Can you pass a loop parameter out of a procedure as a procedure parameter?

10.86 (Signals and variables, 30 min.) Write a summary on signals and variables, including code examples.

10.87 (Type conversion, 60 min.) There are some very subtle rules involving type conversion, [VHDL 93LRM7.3.5]. Does the following work? Explain the type conversion rules.

BV <= BIT_VECTOR("1111");

Chapter  start   Previous  page   Next  page

S2C: FPGA Base prototyping- Download white paper

Internet Business Systems © 2016 Internet Business Systems, Inc.
595 Millich Dr., Suite 216, Campbell, CA 95008
+1 (408)-337-6870 — Contact Us, or visit our other sites:
AECCafe - Architectural Design and Engineering TechJobsCafe - Technical Jobs and Resumes GISCafe - Geographical Information Services  MCADCafe - Mechanical Design and Engineering ShareCG - Share Computer Graphic (CG) Animation, 3D Art and 3D Models
  Privacy Policy