Lab Exercise Two arithmetic logic    unit    (ALU).

COMP4300
Lab    Exercise    Two
Objective
This    lab    develops    the    first    building    block for    this    semester’s    CPU,    the    arithmeticlogic    unit    (ALU).    It    will    be    combined    with    the    other    data    path    components    from    later    
labs,    resulting    in    a    complete    CPU by    Lab    4.
Instructions
Develop    VHDL    for    the    following    component.    You    should    define    an    architecture    for    
the    ALU    entity    given    below.    You    should    test    your    architecture    by    developing    
simulation    files    for    the    entity.    Your    architecture    should    implement    the    functionality    
described    in    this    document.    To    make    the    simulation    results    more    readable,    we    will    
use    a    32-bit    datapath.    
You    should    use    the    types    from    the    package    “dlx_types”    (available    on    the    files    section    
of    Canvas)     and    the    conversion and    math routines    from    the    package    “bv_arithmetic”    
in    files    bva.vhd,    and    bva-b.vhd    on    Canvas.    The    propagation delay    through    the    ALU
should    be    15 nanoseconds.     Code    to    make    use    of    them    should    be    included    in    your    
vhdl    file    for    the    ALU:    
use work.dlx_types.all;
use work.bv_arithmetic.all;
Arithmetic-Logic Unit This unit takes in two 32-bit values, and a 4-bit operation code
that specifies which ALU operation (eg. add, subtract, multiply, etc) is to be performed
on the two operands. For non commutative operations like subtract or divide, operand1
always goes on the left of the operator and operand2 on the right.
The operation codes are
0000 = unsigned add
0001 = unsigned subtract
0010 = two’s complement add
0011 = two’s complement subtract
0100 = two’s complement multiply
0101 = two’s complement divide
0111 = bitwise AND
1001 = bitwise OR
1011 = bitwise NOT of operand1 (ignore operand2)
1100 = pass operand1 through to the output
1101 = pass operand2 through to the output
1110 = output all zero’s
1111 = output all one’s
The unit returns the 32-bit result of the operation and a 4-bit error code. The meaning of
the error code should be
0000 = no error
0001 = overflow (either negative or positive)
0010 = divide by zero
The ALU should be sensitive to changes on all the input ports. Note that the ALU is
purely combinational logic, no memory, and that there is no clock signal.
The entity declaration is
entity alu is
generic(prop_delay : Time := 5 ns);
port(operand1, operand2: in dlx_word; operation: in
alu_operation_code;
result: out dlx_word; error: out error_code);
end entity alu;
Deliverables
Please    turn    in    the    following    things    for    this    lab:
1.    The    text    of    your    VHDL    source    code.
2.    Your    simulation    test    file.    Do    not    exhaustively    test    these    designs    since    they    take    
lots    of    input    bits,    but    do    test    a    reasonable    number    of    things.     For    example,    test    both    
normal    function    and    an    error,    if    applicable    (e.g.    overflow    for    unsigned    add,    subtract,    
multiply,    nothing    for    and/or/not)        for    each    of    the    alu    operations.    
3.    Transcripts    of    tests    or    screenshots        running    your    simulations.    It    should    be    clear    
what    is    being    tested.    If    not,    add    text    to    explain    it. You    cannot    test    exhaustively    but    
you    should    test    each    function    code    for    correct    operation    and    demonstrate    how    each    
error    code    is    generated.    
Please    turn    in    all    files    on    Canvas.    If    I    have    questions,    I    may    ask    you    to    schedule    a    
time    to    demo    your    code,    if    I    can't    figure    out    how    something    works    by    reading the    
code.